Kinetic and Stoichiometric Characterisation of Streptavidin‐Binding Aptamers

DOI: 10.1002/cbic.201100774

Kinetic and Stoichiometric Characterisation ofStreptavidin-Binding AptamersVincent J. B. Ruigrok,*[a] Esther van Duijn,[b] Arjan Barendregt,[b] Kevin Dyer,[c] John A. Tainer,[c]

Regina Stoltenburg,[d] Beate Strehlitz,[d] Mark Levisson,[a] Hauke Smidt,*[a] andJohn van der Oost*[a]

Introduction

Aptamers are oligonucleotides (RNA or single-stranded DNA)that serve as high-affinity binding ligands, selected to interactspecifically with a target. Potential aptamer targets range fromsmall organic molecules (e.g. , ethanolamine[1] and acetylcho-line[2]) to proteins and large protein complexes,[3] and even tocells.[4] Research into aptamers has steadily increased aftertheir initial development in 1990.[3a, 5] Although at presentthere are only a few commercial applications of aptamers, theycertainly have beneficial properties for application as therapeu-tics (e.g. , FDA-approved Pegaptanib (Macugen)[6]). In addition,aptamers have potential in the fields of targeted drug-deliverysystems, biological recognition tools in biosensors and imagingtools.[7]

Aptamers that specifically bind predefined targets are gener-ally selected in vitro by SELEX (systematic enrichment of li-gands by exponential enrichment). The SELEX procedure con-sists of multiple selection rounds, starting with a large pool ofsynthetic oligonucleotides containing a variable region (1014–1015 variants) flanked by constant regions. Rounds of selectionare initiated by exposure of RNA or ssDNA oligonucleotides toan immobilised target molecule. Subsequently, nonbinding oli-gonucleotide molecules are washed away, after which boundmolecules are recovered, amplified by PCR or RT-PCR andmade single-stranded again. Several selection procedures havebeen successfully applied over the years, including affinitychromatography,[8] capillary electrophoresis[9] and enrichmentthrough the use of target-coated magnetic beads.[10] After suc-cessful selection, the minimal binding sequence should be ob-tained, in order to increase specificity and to reduce synthesiscosts. Approaches to identification and characterisation of the

minimal binding sequence can be straightforward, but onlywhen a conserved nucleotide motif and folding pattern is en-riched.

A relevant feature of an aptamer is its affinity for its target.The affinity constant (KD [m]) is often measured by approachesthat do not provide insight into the actual kinetics of binding.However, the kinetic parameters can reveal valuable informa-tion: the association rate constant (ka [m�1 s�1]) is a measure ofhow rapidly a complex is formed, whereas the dissociation rateconstant (kd [s�1]) is a measure of how rapidly a complex fallsapart. The affinity constant can easily be calculated from theratio of the dissociation and association rate constants (KD = kd/

Aptamers are oligonucleotide ligands that are selected forhigh-affinity binding to molecular targets. Only limited knowl-edge relating to relations between structural and kinetic prop-erties that define aptamer–target interactions is available. Tothis end, streptavidin-binding aptamers were isolated andcharacterised by distinct analytical techniques. Binding kineticsof five broadly similar aptamers were determined by surfaceplasmon resonance (SPR); affinities ranged from 35–375 nm

with large differences in association and dissociation rates.Native mass spectrometry showed that streptavidin can ac-commodate up to two aptamer units. In a 3D model of one

aptamer, conserved regions are exposed, strongly suggestingthat they directly interact with the biotin-binding pockets ofstreptavidin. Mutational studies confirmed both conserved re-gions to be crucial for binding. An important result is the ob-servation that the most abundant aptamer in our selections isnot the tightest binder, emphasising the importance of havinginsight into the kinetics of complex formation. To find thetightest binder it might be better to perform fewer selectionrounds and to focus on post-selection characterisation,through the use of complementary approaches as described inthis study.

[a] V. J. B. Ruigrok, Dr. M. Levisson, Prof. Dr. H. Smidt, Prof. Dr. J. van der OostLaboratory of Microbiology, Wageningen UniversityDreijenplein 10, 6703 HB Wageningen (The Netherlands)E-mail : [email protected]

[email protected]@wur.nl

[b] Dr. E. van Duijn, A. BarendregtBiomolecular Mass Spectrometry and Proteomics GroupBijvoet Center for Biomolecular ResearchUtrecht Institute for Pharmaceutical Sciences, Utrecht UniversityPadualaan 8, 3584 CH Utrecht (The Netherlands)

[c] K. Dyer, Dr. J. A. TainerLife Science Division, Lawrence Berkeley National Laboratory1 Cyclotron Road, Berkeley, CA 94720 (USA)

[d] Dr. R. Stoltenburg, Dr. B. StrehlitzDepartment of Environmental BiotechnologyCentre for Environmental Research Leipzig-Halle GmbH (UFZ)Permoserstrasse 15, 04318 Leipzig (Germany)

Supporting information for this article is available on the WWW underhttp ://dx.doi.org/10.1002/cbic.201100774.

ChemBioChem 0000, 00, 1 – 9 � 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim &1&

These are not the final page numbers! ��

ka) ; this illustrates how different kinetics can result in similar af-finities. Knowledge of underlying kinetic parameters is there-fore relevant for better understanding of the molecular basisof complex formation and stability. One of the methods usedto determine kinetic parameters (ka and kd) is surface plasmonresonance (SPR), an advanced method for following binding in-teractions in real time.[11]

Streptavidin, a 60 kDa homo-tetrameric protein originallyisolated from Streptomyces avidinii, is used extensively in mo-lecular biology, because the interaction between streptavidinand its natural ligand biotin is one of the strongest noncova-lent molecular interactions known (KD�10�14

m). In the ab-sence of biotin, a loop close to the binding pocket is open;however, upon biotin binding the loop moves to cover thepocket. This conformational change explains the exceptionallyhigh affinity—the result of a very low kd value—for biotin.[12] Intwo recent studies, streptavidin has been used as target foraptamer selection. Remarkably, the predicted secondary struc-tures of the enriched binding motifs were quite similar, al-though only a few nucleotides appeared to be conserved.[10, 13]

From equilibrium analyses and competition experiments, affini-ty constants ranging from 40 to 85 nm were reported, al-though no data on the kinetics are available. In addition, biotinwas shown to compete with streptavidin-binding aptamers,suggesting that the aptamers bind in or near the biotin-bind-ing pocket.[13] Nevertheless, the stoichiometry and the locationof binding are not known.

Here we set out to test whether or not multiple independ-ent selection experiments would result in similar aptamers, interms of sequence and structure as well as kinetic properties.In addition, we wanted to obtain a thorough understanding ofthe kinetics underlying aptamer selection, ideally resulting inaptamers with improved binding affinities. Streptavidin waschosen as target because streptavidin-binding aptamers havealready been described, because detailed knowledge on itsstructure is available, and because its stability makes it relative-ly easy to perform SPR experiments. Seven rounds of selectionresulted in streptavidin-binding oligonucleotides with a con-served secondary structure. Five different aptamer familiescould be distinguished, and SPR analysis showed that, despitetheir similarity, they differ greatly in their affinities (35–375 nm)and their kinetic behaviour. In addition, the stoichiometries ofthe complexes were determined by native mass spectrometry,and small angle X-ray scattering (SAXS) data were used to gen-erate a structural model of the aptamers.

Results

Streptavidin-binding aptamers

Through the use of the experimental approaches describedabove we aimed to obtain ssDNA aptamers that would bind tostreptavidin. After seven rounds of selection, the enriched poolof aptamers was cloned. In total, 91 clones were sequencedand used for subsequent analysis. Assembly of individual se-quences yielded 12 clusters that each contained two to 13fully identical sequences, whereas 19 sequences occurred only

once, resulting in a total of 31 unique sequences (Figure S1 inthe Supporting Information).

Structural predictions of all unique sequences (clusters andsinglets) were made by use of Mfold (a selection is depicted inFigure S2). In 26 out of the 31 sequences a similar structure ofa stem-bulge-stem-loop was pre-dicted (Scheme 1); in the remain-ing five sequences (two clustersrepresenting 16 clones, and threesinglets) no such structure waspredicted. After manual sequenceinspection, however, regions thatcould form stem–bulge–stem–loopstructures were found in these fivesequences as well. Submission ofthese sequence fragments toMfold confirmed this prediction. Asexpected, the locations of thestructure motifs are not fixedwithin the DNA sequences; onsome occasions the primers, flank-ing the random regions, seem tobe involved in the structures (Fig-ure S1).

In order to gain insight into theconservation of nucleotide sequen-ces and structures, the sequencespredicted to form the stem-bulge-stem-loop structures were manual-ly aligned according to their posi-tions within the predicted struc-tures (Scheme 2). Despite the over-all structural similarity some small variations could be ob-served: loop sizes vary between five and seven nucleotides,the bulges contain either three or five nucleotides, andstems 2 have three base pairs when the loops contain sevennucleotides and four base pairs in cases of smaller loop sizes.In particular, the nucleotides in the bulges (CGC) and loops(CGCA) are well conserved, whereas the actual compositions ofthe stems do not seem to be important as long as properWatson–Crick base pairing is maintained.

From the alignment and the small differences between se-quences it was possible to distinguish five aptamer familiesthat occur at different frequencies in the library of selectedclones. A shortened aptamer representing each family—Stre-pApt1 to StrepApt5—was designed for subsequent experi-ments while maintaining the specific features of each family(Scheme 3).

Aptamer binding kinetics

Shortened aptamers—StrepApt1 to StrepApt5—as well as St-2-1 described by Bing et al. ,[13] were used for the determinationof aptamer binding kinetics by SPR. A titration-kinetics ap-proach was used: in one run increasing aptamer concentra-tions were sequentially injected over a streptavidin-coated sur-face. Double referenced data were used for fitting according to

Scheme 1. Secondary struc-ture of the predicted stem-bulge-stem-loop sequences.Nucleotides in capitals areconserved in the majority ofsequences, nucleotides inlower case italics are con-served within a family (orfamilies), and nucleotides be-tween parentheses are absentin some families. n representsone of four nucleotides, ands represents C or G.

&2& www.chembiochem.org � 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemBioChem 0000, 00, 1 – 9

�� These are not the final page numbers!

V. J. B. Ruigrok, H. Smidt, J. van der Oost et al.

a simple 1:1 binding interactionmodel designed for fitting titra-tion kinetics data.[14] Experimen-tal data and corresponding fitsare shown in Figure 1 A, whereasrate constants (ka, kd) and affinityconstants (KD) derived fromfitted data are presented inTable 1.

Although the differences in se-quences and predicted struc-tures are small, the binding ki-netics of the five streptavidinaptamer types varied dramatical-ly. In direct relation to the differ-ences in their kinetics, the affini-ty constants (KD) range from 35to 375 nm (Table 1). The affinityconstant found for St-2-1 (79�2 nm) agrees well with data re-ported earlier (40�18 nm).[13] Itshould be noted that associationand dissociation rates for thisaptamer were not reported inthat study because of limitations

of the analytical approach used.From the association rates (ka) of the five shortened aptam-

ers designed in this study it was possible to distinguish twogroups: those that form complexes rapidly (StrepApt2 andStrepApt3; Table 1) and those that form complexes slowly(StrepApt1, StrepApt4 and StrepApt5). From the dissociationrates (kd) it was possible to distinguish two other groups:those that dissociate rapidly (StrepApt1, StrepAtp2 and Stre-pApt3) and those that dissociate slowly (StrepApt4 and Stre-pApt5).

StrepApt4 and StrepApt5 (low values of ka and kd) representonly minor fractions of sequences in the clone library, which isdominated by StrepApt2, representing 53 % of all sequences.Despite being the most abundant sequence, StrepApt2 doesnot have the highest affinity (77 nm). Rather, the highest-affini-ty aptamer (StrepApt5, 35 nm) represented only 12 % of the se-quences in the clone library. It is interesting to note here thatthe affinity and kinetic behaviour of St-2-1 are similar to thoseof StrepApt2 (Table 1).

Scheme 3. Shortened aptamers representing aptamer families. Nucleotidesin bold are those conserved in the majority of sequences belonging toa given family. Nucleotides in italics are not based on specific features in thealignment, but instead are added to assure stable conformations.

Scheme 2. Alignment of stem-bulge-stem-loop sequences. Nucleotides con-served in the majority of sequences andthose conserved in families are high-lighted. Ctg is a contig based on thenumber of sequences indicated in pa-rentheses. Percentages indicate the oc-currence of a certain family within theclone library.

ChemBioChem 0000, 00, 1 – 9 � 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chembiochem.org &3&

These are not the final page numbers! ��

Streptavidin-Binding Aptamers

Influence of mutations andbiotin

Two conserved regions, sharinga common CGC motif, werefound in the majority of theshortened aptamers (Scheme 3).In order to determine the impor-tance of the conserved guaninesfor streptavidin binding, theywere replaced by thymines inStrepApt5, our best binder. Re-placement of one guanine re-sulted in a strongly reduced af-finity in the micromolar range (�2 mm ; Figure 1 B). Although weperformed only one injectionseries for these mutants the datawere still fitted for the purposeof obtaining a rough estimate ofthe KD value. Replacement ofboth guanines by thymines com-pletely abolished binding, indi-cating that these residues are es-sential for binding in all aptamerfamilies.

SPR was furthermore used forcompetition experiments. Biotinis the natural ligand bound bystreptavidin, with an exception-ally high affinity (KD�10�14

m).[12]

In order to assess the impact ofbiotin on aptamer binding, thestreptavidin-coated chip was firstloaded with StrepApt5, the apta-mer with the highest affinity, fol-lowed by normal dissociation(Figure 1 C). After regenerationthe chip was reloaded with apta-mer, directly followed by injec-tion of biotin (by use of the co-inject command). The faster dis-

sociation and increased baseline, even after regeneration, re-vealed that biotin outcompetes StrepApt5. In addition, subse-quent aptamer injections did not result in any binding.

Stoichiometries of the complexes

The binding stoichiometries of the various streptavidin-apta-mer complexes were determined by native mass spectrometry.Native mass spectrometry is a powerful method for study ofnoncovalent complexes, the exact mass analysis often beingsufficient to determine the stoichiometry of a complex unam-biguously. Streptavidin was found to bind a maximum of twoaptamer units simultaneously, regardless of the aptamer used(Figure 2). In addition to identifying binding stoichiometries,native mass spectrometry can also be applied to monitor the

Figure 1. SPR data. A) Titration kinetics of shortened aptamers, B) titration kinetics of mutated variants of Stre-pApt5, and C) biotin competition experiment. Double referenced data for shortened aptamer titration kinetics areshown in black; fits according to a simple 1:1 binding model are depicted in grey.

Table 1. Relative occurrence of shortened aptamers in the clone library,together with kinetic parameters of aptamer–streptavidin binding as de-termined by SPR analysis. Highest association rates and slowest dissocia-tion rates are highlighted in bold.

Occurrence in KD [nm] ka (� 104) kd (� 10�3)clone library [%] [m�1 s�1] [s�1]

StrepApt1 10 375�50 1.37�0.05 5.1�0.7StrepApt2 53 77�3 3.98�0.09 3.07�0.09StrepApt3 19 105�4 4.8�0.8 5.0�0.8StrepApt4 6 181�7 0.73�0.07 1.32�0.08StrepApt5 12 35�1 1.04�0.03 0.36�0.01St-2-1 – 79�2 3.2�0.3 2.5�0.2

&4& www.chembiochem.org � 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemBioChem 0000, 00, 1 – 9

�� These are not the final page numbers!

V. J. B. Ruigrok, H. Smidt, J. van der Oost et al.

binding strengths of aptamers to streptavidin. The relativeabundances of free streptavidin, of streptavidin bound to oneaptamer and of streptavidin bound to two aptamers confirmedthat aptamer 5 has the highest affinity for streptavidin (Fig-ure 2 B and Table S1).

Structural model of StrepApt2 and the complex

Modelling of the 3D structure of StrepApt2 was performedwith use of the MC-Fold and MC-Sym pipeline to provide in-sight into the structural basis of aptamer binding to the strep-tavidin tetramer. In total, 57 3D structures were predicted forStrepApt2, from its sequence and the 2D structure predictedby Mfold. In order to identify the best prediction, SAXS datafor the aptamer were compared with theoretical scattering ofthe 57 predicted models. The structure (Figure 3) with thelowest Chi2 value (5.32, Figure S3) was considered the bestmatch with the experimental scattering. This structure indi-cates that the conserved nucleotides in the loop (CGCA) andbulge (CGC) are solvent-exposed and are therefore the mostlikely candidates to interact with streptavidin.

Discussion

DNA aptamers that bind to streptavidin, with affinities in thelow to middle nanomolar range (35–375 nm), were enriched inseven sequential rounds of selection. Sequence analysis of 91

clones revealed 31 unique sequences at different relative abun-dances. Structural predictions, based on Mfold, and sequenceanalysis show that all sequences contain stem-bulge-stem-loopstructures. Interestingly, similar structures for streptavidin-bind-ing aptamers had been reported earlier, although the aptamersequences vary slightly, and nucleotides in the bulge (CGC)and loop (CGCA) regions seem to be conserved throughoutvarious independent selection experiments.[10, 13] Variation inloop composition and size (five to seven nucleotides), bulgesize (either three or five nucleotides) and stem 2 length (threeor four base pairs) further allowed five aptamer families to bedistinguished. The compositions of the stems do not seem tobe important as long as proper Watson–Crick base pairing ismaintained. The length of stem 1, however, is important forstability of the aptamer, as has been shown previously by Bingand co-authors.[13] Shortened aptamers—StrepApt1 to Stre-pApt5—for each of the five types, based on the alignment,were designed, with the specific features of each family main-tained. Single mutations of the conserved guanines in theloops and bulges resulted in reduced binding of the aptamers,whereas in the double mutant binding was abolished, indicat-ing that these conserved guanines play an essential role instreptavidin binding.

We further determined the kinetic properties of the short-ened aptamers StrepApt1 to StrepApt5 by use of SPR titrationkinetics. The obtained results showed widely different affinityconstants (35–375 nm), but no obvious relation between affini-ty and abundance in the clone library was observed. Thenumber of clones is limited and might not represent the com-plete range of aptamers present after seven selection rounds,but results of independent studies showed enrichment of simi-lar aptamers. Kinetic parameters of representative aptamers in-dicate that the rapid binders (StrepAtp2 and StrepApt3, loopsize 6 nt) dominate the clone library, together representing72 % of all sequences. In contrast, the slowly dissociating ap-tamers (StrepApt4 and StrepApt5, loop size 7 nt) are a minority,together representing 18 % of the clone library. The kinetic pa-rameters determined in this study show that slow dissociationseems to be related to a loop size of seven nucleotides, where-as rapid association seems to be associated with a loop size of

Figure 2. Native mass spectrometry data. A) Spectra of the lowest-affinitybinder and of the highest-affinity binder. B) Relative abundances of freestreptavidin, streptavidin bound to one aptamer and streptavidin bound totwo aptamers. Diamonds show data corresponding to free streptavidin,squares data corresponding to streptavidin + one aptamer, and trianglesdata corresponding to streptavidin + two aptamers.

Figure 3. Predicted 3 D model of StrepApt2. Prediction 28 of the MC-Foldand MC-Sym pipeline that best matches experimental scatter. Stems are inblue, bulge and loop in gold. The two guanines that are essential for bind-ing are depicted in red.

ChemBioChem 0000, 00, 1 – 9 � 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chembiochem.org &5&

These are not the final page numbers! ��

Streptavidin-Binding Aptamers

six nucleotides. It is interesting to note here that the aptamerwith the highest affinity was not the most abundant in ourclone library. Because of its structural similarity we also includ-ed St-2-1, selected by Bing and co-authors,[13] in our SPR analy-sis. The kinetic parameters of St-2-1, as well as its predictedfolding motif, were most similar to those of StrepApt2. Two in-dependent selection experiments hence resulted in similarbinders, with rapid association rates; it is therefore temptingto speculate that the selection procedure is somewhat biasedtowards selection of rapid binders, rather than high-affinitybinders. This is also observed in the extensive work by Winet al.[15] on codeine-binding RNA aptamers. These authorsfound several sequences to be present more than once in theirclone library, whereas the five best binders, those with thelowest KD values, occurred only once.

Consistently with the study of Bing et al. , we also observedcompetition between aptamer (StrepApt5) and biotin. Never-theless, our data clearly showed aptamer binding to be com-pletely suppressed when streptavidin is saturated with biotin,whereas Bing et al. still observed some binding in the presenceof biotin.[13] Both studies, however, unequivocally suggest thatthe aptamer binds in, or close to, the biotin-binding pocket.

To gain insight in the stoichiometry of the streptavidin–apta-mer complex we performed native mass spectrometry, whichindicated that streptavidin can bind a maximum of two apta-mer moieties simultaneously, regardless of the aptamer usedand its affinity. This leaves the question, however, of whetherone aptamer binds two biotin pockets at once, or if access toa single biotin pocket is blocked by a bound aptamer. In orderto gain insight in the location of binding, and to provide ananswer to the question of whether or not one aptamer bindstwo biotin pockets, structural predictions (57) of StrepApt2were generated by use of the MC-Fold and MC-Sym pipeline.The conserved nucleotides in the loop and bulge are solvent-exposed and are therefore the most likely candidates to inter-act with the loops that cover the two biotin pockets oncebiotin is bound (Figure S4). On the basis of this model it canbe hypothesised that one aptamer occupies two biotin pock-ets. This is supported by the presence of two conserved re-gions and by the fact that two simultaneous mutations, one ineach of these conserved regions, are required to yield a non-binding aptamer. Our approach of obtaining a 3D model foran aptamer, supported by SAXS data, might also be helpful asa new strategy to afford insight into the binding sites of otheraptamers.

Conclusions

All streptavidin-binding aptamers selected in this work havesimilar structures and conserved nucleotides, but still varyslightly at sequence level. Although the differences in se-quence are small, the differences in binding kinetics are sub-stantial. One streptavidin tetramer can accommodate up totwo aptamers, and we propose that one aptamer occupiestwo biotin-binding pockets. It is interesting to note that theaptamer with the highest affinity is not the most abundant. In-stead, selection appears to be driven by the binding kinetics.

This could be the result of the presence of a limited number ofbinding sites that are quickly saturated by rapid binders whenselection progresses through subsequent rounds. This findingindicates that it might be better either to increase the numberof target binding sites (by using more beads, or less DNA) orto limit the number of selection rounds. Moreover, these find-ings indicate that one should focus on post-selection charac-terisation of potential aptamers, by using complementary ap-proaches introduced here, for example, in order to find theaptamer with the highest affinity.

Experimental Section

Oligonucleotides : Unmodified primers AP10 (ATACC AGCTT ATTCAATT)[10] and AP30 (CTAAC TGATT ACGAT TGT) (R. Stoltenburg, per-sonal communication), the random pool (ATACC AGCTT ATTCAATT-N64-A CAATC GTAAT CAGTT AG, PAGE purified), primer AP60,equivalent to AP10 but containing 5’-fluorescein, and shortenedaptamers (for sequences see Table 2) were obtained from Biolegio(Nijmegen, the Netherlands). Primer TER-AP30, equal to AP30 butcontaining 5’-poly-dA20-hexa(ethylene glycol), was ordered fromIBA (Gçttingen, Germany).

SELEX procedure (FluMag-SELEX): The selection procedure was asdescribed previously.[10] Briefly, before each selection round ssDNAwas heated to 95 8C for 8 min, cooled on ice for 10 min and left atroom temperature for at least 10 min. In the first round, 26 pmolof the random pool was added to streptavidin-coated beads (Dy-nabeads M-280, Invitrogen), and in subsequent rounds 250 mL ofthe ssDNA preparation (see below) was added. Approximately 1 �108 beads were washed with selection buffer [NaCl (100 mm),MgCl2 (2 mm), KCl (5 mm), CaCl2 (1 mm), Tris·HCl (pH 7.6, 20 mm)] ,and after addition of the ssDNA to the beads they were incubatedat 25 8C with mild shaking for 30 min. Unbound oligonucleotideswere removed by washing three times with selection buffer. Subse-quently, bound oligonucleotides were eluted by addition of selec-tion buffer and incubation at 95 8C for 5 min with mild shaking.

Eluted oligonucleotides were PCR-amplified in 15 parallel PCR reac-tions, each mixture containing MgCl2 (1.9 mm), dNTPs (0.2 mm

each), primer (AP60 and TER-AP30, 1 mm of each), 1 � buffer B2, andHOT FIREPol (Solis BioDyne, 4 units) in a total volume of 100 mL.The PCR program was as follows: 15 min at 95 8C, 30 cycles of 30 s

Table 2. Sequences of shortened and mutated aptamers used in thisstudy. Uppercase letters indicate nucleotides that are conserved in themajority of sequences within a given family. G!T mutations in Stre-pApt5.1, StrepApt5.2 and StrepApt5.3 are given in bold letters and under-lined. All sequences are shown in the 5’-to-3’ direction. The sequence ofSt-2-1 was taken from.[13]

StrepApt1 gggaACGCgttaTTGGGtaacTtcccStrepApt2 gggaatCGCcaccCGACGCAgggtTtcccStrepApt3 gggaACGCataCGCCGCAGtatTtcccStrepApt4 gggaACGCccgTATTGCTcggTtcccStrepApt5 gggaACGCaccGATCGCAggtTtcccStrepApt5.1 gggaACTCaccGATCGCAggtTtcccStrepApt5.2 gggaACGCaccGATCTCAggtTtcccStrepApt5.3 gggaACTCaccGATCTCAggtTtcccSt-2-1 ATTGACCGCTGTGTGACGCAACACTCAAT

&6& www.chembiochem.org � 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemBioChem 0000, 00, 1 – 9

�� These are not the final page numbers!

V. J. B. Ruigrok, H. Smidt, J. van der Oost et al.

at 95 8C, 30 s at 52 8C, 1 min at 72 8C, followed by 7 min at 72 8Cafter the last cycle. Electrophoresis on a 12 % polyacrylamide (PAA)gel was used to confirm successful amplification of a DNA frag-ment of the correct size.

From the second selection round onwards, fluorescence in thestarting sample, the non-bound DNA fraction, the wash fractionsand the eluted DNA was measured in 96-well microtiter plates(Greiner bio-one) with a SpectraMax M2 instrument (Molecular De-vices).

ssDNA preparation : Unpurified PCR products were pooled andconcentrated by ethanol precipitation and resuspended in TEbuffer [Tris (10 mm), EDTA (1 mm), pH 7.4, 100 mL]. Strands wereseparated by electrophoresis of the concentrated sample on a de-naturing PAA gel, containing PAA (8 %) and urea (7 m) in TBE buffer[Tris (10 mm), borate (90 mm), EDTA (2 mm)] . Fluorescein-labelledDNA was visualised with a Safe imager (Invitrogen) and excisedfrom the gel. The DNA was eluted from the gel slice by mashingand subsequent incubation in 600 mL EDTA (2 mm) and sodiumacetate (pH 7.8, 300 mm) at 80 8C with mild shaking for 2 h. Afterethanol precipitation the DNA was resuspended in selection buffer(300 mL).

Cloning, sequencing and structure analysis : The DNA eluted inSELEX round 7 was amplified, under the conditions describedabove, by using unmodified primers AP10 and AP30 and ligatedinto pGEM-T (Promega). Escherichia coli XL1-Blue cells (Stratagene)were transformed with the vector constructs, 96 colonies weresubsequently randomly picked, and inserts were sequenced (GATC,Konstanz, Germany). The DNA folding form of Mfold[16] was usedfor secondary structure prediction with use of the default settingsexcept for temperature (25 8C) and ionic conditions [Na+ (100 mm),Mg2 + (2 mm)] .

Native mass spectrometry : High-resolution mass spectra of strep-tavidin complexes with StrepApt1 to StrepApt5 and St-2-1 were re-corded with an LCT instrument (Waters, Manchester, UK) in positiveion mode. Samples were loaded into gold-plated borosilicate capil-laries made in-house (with a Sutter P-97 puller (Sutter InstrumentsCo., Novato, CA, USA) and an Edwards Scancoat six sputter-coater(Edwards Laboratories, Milpitas, CA, USA)) for analysis with anLCT 1 mass spectrometer (Waters Corp. , Milford, MA, USA), adjust-ed for optimal performance in high-mass detection.[17] A capillaryvoltage of 1300 V was used, with a sampling cone voltage of 90 V.The source backing pressure was elevated in order to promote col-lisional cooling to approximately 6 mbar. All samples were sprayedfrom ammonium acetate (pH 7.6, 125 mm). Calibration of the in-strument was carried out with a cesium iodide solution(25 mg mL�1). Processing of the acquired spectra was performedwith MassLynx 4.1 software (Waters Corp., Milford, MA, USA). Mini-mal smoothing was used, after which the spectra were centred.The mass of the species was calculated with use of each chargestate in a series. The corresponding intensities of each charge statewere assigned by MassLynx and summed. This approach allowsthe relative quantification of all species in a sample.

Surface plasmon resonance : In a Biacore 3000 system (BIACORE,Uppsala, Sweden), streptavidin (recombinant produced in E. coli,SIGMA, approximately 2000 RU), dissolved in sodium acetate(pH 4.0, 10 mm, GE Healthcare), was immobilised on a CM5 chip at25 8C. The “aim-for-immobilised-level” wizard was used, togetherwith reagents provided in the amine coupling kit (GE Healthcare).A reference channel was prepared by activating and subsequentblocking of the surface by the same procedure.

The shortened aptamers StrepApt1 to StrepApt5.3 and St-2-1 weredissolved in selection buffer at concentrations of 100 mm and sub-sequently diluted in selection buffer to concentrations of 2.50,0.83, 0.28, 0.09 and 0.03 mm. Titration kinetics were performed bysequential injections of each concentration (60 mL, starting withthe lowest concentration) at 30 mL min�1, followed by 60 s dissocia-tion. The fifth injection was followed by 20 min dissociation andsubsequent regeneration with glycine (pH 3.0, 10 mm, 10 mL) at10 mL min�1. Titration kinetics reduce assay times as well as thenumber of chip regenerations required. Prior to aptamer injectionsa series of blank injections, with only selection buffer, was run toallow for double referencing of the data. Selection buffer was usedas running buffer. Each aptamer was injected three times anddouble referenced data were used for fitting according to a 1:1binding model as described by Karlsson et al.[14] The concentrationseries of StrepApt5.1 to StrepApt5.3, which are mutated variants ofStrepApt5, were injected only once.

Small-angle X-ray scattering and 3D modelling : StrepApt2 wasdissolved in selection buffer to concentrations of 4.2, 2.9 and1.5 mg mL�1. SAXS data were collected at the SIBYLS beamline at10 8C, by procedures described previously.[18] Briefly, the threesample concentrations and a buffer blank were measured in order(buffer, low, middle, and high concentration, and again buffer) atexposures of 0.5, 1.0 and 6.0 s. Scattering curves were merged withuse of PRIMUS[19] in the ATSAS program suite.[20] The MC-Fold andMC-Sym pipeline was used to obtain predictions of the 3D struc-ture of StrepApt2.[21] Theoretical scattering of these predictionswas calculated and compared with experimentally observed scat-tering of StrepApt2 with use of the FoXS webserver.[22]

Acknowledgements

This work was supported by a grant from the Netherlands Or-ganisation for Scientific Research and the Netherlands Institutefor Space Research [ALW-GO-PL/08-08] . We thank Willem Haas-noot (RIKILT Institute of Food Safety, Wageningen) for help withinitial SPR experiments.

Keywords: aptamers · native mass spectrometry · SAXS ·SELEX · surface plasmon resonance

[1] D. Mann, C. Reinemann, R. Stoltenburg, B. Strehlitz, Biochem. Biophys.Res. Commun. 2005, 338, 1928 – 1934.

[2] J. G. Bruno, M. P. Carrillo, T. Phillips, B. King, In Vitro Cell. Dev. Biol. Anim.2008, 44, 63 – 72.

[3] a) C. Tuerk, L. Gold, Science 1990, 249, 505 – 510; b) L. C. Bock, L. C. Grif-fin, J. A. Latham, E. H. Vermaas, J. J. Toole, Nature 1992, 355, 564 – 566.

[4] a) X. X. Cao, S. H. Li, L. C. Chen, H. M. Ding, H. Xu, Y. P. Huang, J. Li, N. L.Liu, W. H. Cao, Y. J. Zhu, B. F. Shen, N. S. Shao, Nucleic Acids Res. 2009,37, 4621 – 4628; b) Z. W. Tang, D. Shangguan, K. M. Wang, H. Shi, K.Sefah, P. Mallikratchy, H. W. Chen, Y. Li, W. H. Tan, Anal. Chem. 2007, 79,4900 – 4907.

[5] A. D. Ellington, J. W. Szostak, Nature 1990, 346, 818 – 822.[6] E. S. Gragoudas, A. P. Adamis, E. T. Cunningham, Jr. , M. Feinsod, D. R.

Guyer, N. Engl. J. Med. 2004, 351, 2805 – 2816.[7] V. J. B. Ruigrok, M. Levisson, M. H. M. Eppink, H. Smidt, J. van der Oost,

Biochem. J. 2011, 436, 1 – 13.[8] C. Wilson, J. W. Szostak, Nature 1995, 374, 777 – 782.[9] S. D. Mendonsa, M. T. Bowser, Anal. Chem. 2004, 76, 5387 – 5392.

[10] R. Stoltenburg, C. Reinemann, B. Strehlitz, Anal. Bioanal. Chem. 2005,383, 83 – 91.

[11] Y. S. N. Day, C. L. Baird, R. L. Rich, D. G. Myszka, Protein Sci. 2002, 11,1017 – 1025.

ChemBioChem 0000, 00, 1 – 9 � 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chembiochem.org &7&

These are not the final page numbers! ��

Streptavidin-Binding Aptamers

[12] O. H. Laitinen, V. P. Hytçnen, H. R. Nordlund, M. S. Kulomaa, Cell. Mol.Life Sci. 2006, 63, 2992 – 3017.

[13] T. Bing, X. Yang, H. Mei, Z. Cao, D. Shangguan, Bioorg. Med. Chem. 2010,18, 1798 – 1805.

[14] R. Karlsson, P. S. Katsamba, H. Nordin, E. Pol, D. G. Myszka, Anal. Bio-chem. 2006, 349, 136 – 147.

[15] M. N. Win, J. S. Klein, C. D. Smolke, Nucleic Acids Res. 2006, 34, 5670 –5682.

[16] M. Zuker, Nucleic Acids Res. 2003, 31, 3406 – 3415.[17] N. Tahallah, M. Pinkse, C. S. Maier, A. J. R. Heck, Rapid Commun. Mass

Spectrom. 2001, 15, 596 – 601.[18] G. L. Hura, A. L. Menon, M. Hammel, R. P. Rambo, F. L. Poole II, S. E. Tsu-

takawa, F. E. Jenney, Jr. , S. Classen, K. A. Frankel, R. C. Hopkins, S. J.

Yang, J. W. Scott, B. D. Dillard, M. W. W. Adams, J. A. Tainer, Nat. Methods2009, 6, 606 – 612.

[19] P. V. Konarev, V. V. Volkov, A. V. Sokolova, M. H. J. Koch, D. I. Svergun, J.Appl. Crystallogr. 2003, 36, 1277 – 1282.

[20] P. V. Konarev, M. V. Petoukhov, V. V. Volkov, D. I. Svergun, J. Appl. Crystal-logr. 2006, 39, 277 – 286.

[21] M. Parisien, F. Major, Nature 2008, 452, 51 – 55.[22] D. Schneidman-Duhovny, M. Hammel, A. Sali, Nucleic Acids Res. 2010,

38, W540 – W544.

Received: December 13, 2011Published online on && &&, 0000

&8& www.chembiochem.org � 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemBioChem 0000, 00, 1 – 9

�� These are not the final page numbers!

V. J. B. Ruigrok, H. Smidt, J. van der Oost et al.

FULL PAPERS

V. J. B. Ruigrok,* E. van Duijn,A. Barendregt, K. Dyer, J. A. Tainer,R. Stoltenburg, B. Strehlitz, M. Levisson,H. Smidt,* J. van der Oost*

&& –&&

Kinetic and StoichiometricCharacterisation of Streptavidin-Binding Aptamers

Tightest binders not the most abun-dant : Detailed characterisation of fivebroadly similar streptavidin-binding ap-tamers revealed widely different kineticbehaviour, and even improved affinity,in relation to previously described ap-tamers. Mutation studies confirm thatnucleotides that had been predicted tobe important by sequence analysis andstructural modelling are indeed essen-tial for binding.

ChemBioChem 0000, 00, 1 – 9 � 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chembiochem.org &9&

These are not the final page numbers! ��