Biomolecular Engineering

16 (1999) 67–72

Extremely high thermal stability of streptavidin and avidinupon biotin binding

Martın Gonzalez, Carlos E. Argarana, Gerardo D. Fidelio *Departamento de Quımica Biologica, CIQUIBIC, Facultad de Ciencias Quımicas, Uni6ersidad Nacional de Cordoba, Ciudad Uni6ersitaria,

5000 Cordoba, Argentina

Abstract

The effect of biotin binding on the thermal stability of streptavidin (STV) and avidin (AVD) was evaluated using differentialscanning calorimetry. Biotin binding increases the midpoint of temperature Tm of thermally induced denaturation of STV andAVD in phosphate buffer from 75 and 83°C to 112 and 117°C at full biotin saturation, respectively. This thermostability is thehighest reported for proteins coming from either mesophilic or thermophilic organisms. In both proteins, biotin also increases thecalorimetric enthalpy and the cooperativity of the unfolding. Thermal stability of STV was also evaluated in the presence of highconcentrations of urea or guanidinium hydrochloride (GuHCl). In 6 M GuHCl, STV remains as a tetramer and the Tm of theSTV-biotin complex is centered at 108°C, a few degrees below the value obtained in phosphate buffer. On the contrary, STVunder fully saturating condition remains mainly in its dimeric form in 8 M urea and the thermogram shows two endotherms. Themain endotherm at a lower temperature has been ascribed to the dimeric liganded state with a Tm of 87°C, and the highertemperature endotherm to the tetrameric liganded form with a Tm of 106°C. As the thermostability of unliganded protein in thepresence of urea is unchanged upon binding we related the extremely high thermal stability of this protein to both an increase instructural ordering and compactness with the preservation of the tetramer integrity. © 1999 Published by Elsevier Science B.V. Allrights reserved.

Keywords: Streptavidin; Avidin; Protein stability; Ligand–protein interaction; Thermal unfolding; Differential scanning calorimetry

www.elsevier.com/locate/geneanabioeng

1. Introduction

In this work we described some aspects about thethermal stability of both streptavidin (STV) and avidin(AVD) and the changes induced by its ligand biotin, theextensively used protein-ligand systems in biochemistry.The biotin–STV interaction has demonstrated to be ahighly versatile tool for biotechnology and biochemistryin more than hundreds of applications. An extensiveand perusal review of these applications has been doneby Wilchek and Bayer [1].

We carried out detailed thermodynamic analysis ofthermal denaturation of STV under subsaturating andsaturating ligand conditions so as to describe the ap-pearance of bimodality; this being the presence of morethan one endothermic peak in a thermogram, as afunction of the amount of ligand present. Analysis of

the effect of the ligand binding on the two-state thermaldenaturation and the magnitude of the ligand associa-tion constant led us to conclude in a rationale modelfor the biotin–STV interaction in which strong cooper-ativity among the STV subunits takes place with con-siderable structural changes upon biotin binding. Theincrease in the thermostability is in keeping with anincrement of protein packing induced by the ligandinteraction (structural cooperativity) without substan-tial changes in the binding affinity [2].

2. Features of the protein streptavidin

STV is a homotetrameric protein in which the molec-ular weight of its subunit is 14.5 kDa. The nativeprotein is present in two different species with a slightdifference in length [3]. Both species have a high affinityfor its ligand biotin with an association constant ofabout 1015 M−1 [4]. The formation of the biotin-STV

* Corresponding author. Tel.: +54-351-4334171; fax: +54-351-4334074.

E-mail address: gfidelio@dqbfcq.uncor.edu (G.D. Fidelio)

1389-0344/99/$ - see front matter © 1999 Published by Elsevier Science B.V. All rights reserved.PII: S 1 0 5 0 -3862 (99 )00041 -8

M. Gonzalez et al. / Biomolecular Engineering 16 (1999) 67–7268

complex does not involve any covalent bonds, but itsolely involves hydrogen bonds, electrostatic and hy-drophobic interactions [5–8]. The binding site is builtby sharing regions of two subunits. In the hydrophobicpocket, the binding site involves several aromatic aminoacids, particularly Trp 79, 92 and 108 from a subunitand Trp 120 from the adjacent subunit [4]. Trp 120from the adjacent subunit seems to be critical for thehigh affinity: a mutant (STV-38) in which Trp 120 isreplaced by Phe showed a considerably lower affinitywith the association constant reduced from 1015 to 108

M−1 [4]. This fact indicates that the loop from theadjacent subunit that contains Trp 120 contributes tothe dimer–dimer stabilization and that an adequateintersubunit contact is necessary for the high affinity.

Spectroscopic studies of the biotin–STV complexhave been done using several techniques: far UV andCD were done by Green and Melaned [9], fluorescencestudies by Kurzban et al. [10] and FTIR assayed byGonzalez et al. [2]. Measurement of the biotin–STVinteraction by force rupture studies was done by Leck-band et al. [11] and Grubmuller et al. [8]. Free energyperturbation was investigated by Miyamoto and Koll-man [12]. The cooperativity in the biotin binding wasinvestigated by measuring the electrophoretic behaviorof the biotin–STV complex [4,14].

The biotin binding induces an increase in the proteinordering. In fact, the x-ray data reported by Weber etal. [6] showed that two loops in a truncated form ofSTV that cannot be well defined in the apo or unli-ganded STV, are clearly identified in the STV–biotincomplex. Moreover, new FT-IR data suggested that

residues not included in the truncated form are alsodisordered in apo STV [2]. X-ray studies have clearlydefined the general structure of STV [6]. The proteinhas around 70% of its residues forming a b-barrel. Thisbarrel involves the b-strands and the connecting turns,besides other turns and two more flexible loops. ByFT-IR, the increment in protein ordering upon biotinbinding was detected as an increase in the intensity at1647 cm−1 (more structured) at the expense of un-ordered assigned bands [2].

Information obtained from the differential scanningcalorimetry (DSC) experiments has given us an ade-quate approach to understand the behavior of STVduring thermal unfolding in the presence of biotin. Thecalorimetric data and the use of the van’t Hoff analysishave permitted to evaluate the nature of STV unfoldingand the changes in the thermal stability (and in thecooperativity of denaturation) induced by biotin.

3. High thermal stability of streptavidin induced by itsligand biotin

DSC thermograms obtained for STV under subsatu-rating conditions of the ligand clearly show a biphasicbehavior (or bimodality) during thermal unfolding (Fig.1). This behavior is seen only at a ligand:STV (subunit)molar ratio ranging from 0.25 to 0.50 where two wellidentified endotherm components are recorded (In Fig.1 is shown only at a 0.50 ligand:protein molar ratio). Ata ligand:protein molar ratio of 0 or 2, the thermogramis monophasic with a single peak (Fig. 1).

Such bimodality arises from two different relatedbiophysical processes that may occur during unfolding[2]. One of the processes is related to the thermody-namic coupling of protein denaturation with ligandbinding equilibrium, a rather known phenomenon de-scribed also for many protein-ligand systems [15]. How-ever, this model assumes that the ligand is releasedprior to denaturation and the protein unfolds from itsunliganded state. The bimodality and the increase inthermal stability at subsaturating conditions can also beexplained if the ligand bound tightly at one of thesubunit remains associated to its site as the unboundprotein subunits unfold. As the binding induces inter-digitation of part of a subunit into the other, it isreasonable to expect that the ligand–protein interactionis propagated through the neighbor subunit influencingthe overall thermal behavior of, at least, the monomer-monomer entity.

STV without biotin shows a single thermogram witha symmetric profile and a Tm centered at 75.5°C. Thisthermogram behaves in a monophasic manner. How-ever, this single endotherm can be deconvoluted in twocomponents that reflect the heterogeneity found in ourSTV samples (not shown). For a system in which the

Fig. 1. Thermograms of Streptavidin. Upper left: unliganded protein;center: biotin–streptavidin at 0.50 subsaturating ligand:monomerprotein molar ratio; bottom right: biotin–streptavidin at full saturat-ing ligand condition (ligand:protein molar ratio of 2). All runs weredone at 0.066 mM protein concentration using a scan rate of 55.6°C/h in 100 mM phosphate buffer pH 7.4. The thermograms at subsatu-rating ligand condition behave in a biphasic manner such as the 0.50ligand:protein molar ratio endotherm.

M. Gonzalez et al. / Biomolecular Engineering 16 (1999) 67–72 69

Table 1Midpoint temperature of denaturation of STV and AVD under different conditions

DTm (°C)Protein/condition Calorimetric enthalpy (DHcal, Kcal mol−1)Tm (°C)

75.5STV (unliganded in phosphate buffer) 150STV (subsaturating condition 0.5 ligand:protein molar ratio)

5.6Lower temperature endothermic peak 64.481.132.5108.0 90Higher temperature endothermic peak

STV (fully liganded in phosphate buffer) 112.2 36.7 210

AVD (unliganded in phosphate buffer) 11583.833.2AVD (fully liganded in phosphate buffer) 151117

STV (unliganded in urea 8 M) 11573.9STV (fully liganded in urea 8 M)

13.587.4 820Lower temperature endothermic peak106Higher temperature endothermic peak 32.1 104

63.4 60STV (unliganded in GuHCl 6 M)108STV (fully liganded in GuHCl 6 M) 44.6 141

tetramer is bound with 1 or 2 biotins (at subsaturatingligand:protein molar ratios of 0.25 and 0.50, respec-tively) a biphasic profile was observed in the ther-mograms. If the thermally induced protein denaturationis affected by changing the ligand:protein ratio and ifwe assume that the increase in the Tm (thermal stability)is due only to the thermodynamic coupling of theunfolding with the ligand binding equilibria, the ex-pected Tm of STV can be theoretically estimated at anybiotin:protein ratio [2]. The expected Tm increases pro-portionally as the free biotin increases reaching a valueof about 96°C at saturating ligand condition [2]. How-ever, the experimental value of Tm at saturating biotincondition is 16°C above the value expected by consider-ing only the affinity of Kass:1015 M−1. At saturatingligand condition, a considerable increase in the Tm ofSTV by about 37°C is observed compared to the un-bound protein, and the STV–biotin complex has asingle peak centered at 112.2°C (Fig. 1, Table 1), arather unusually high temperature of denaturation for aprotein–ligand system. The denaturation of ligand-sta-bilized STV is characterized also by both a 3-foldincrease in excess heat capacity and by an increment inthe calorimetric enthalpy, from 150 kcal mol−1 forunliganded STV to 210 kcal mol−1 for the full satu-rated biotin–STV complex (Fig. 1).

4. Avidin is calorimetrically homologous to streptavidin

The high stability acquired by the STV–biotin sys-tem was also observed for the homologous AVD–bi-otin system. We found for unliganded AVD a Tm of84°C increasing 33°C to 117°C under saturating condi-tions (Fig. 2). Similar to the STV–biotin system, aconsiderable increase in excess heat capacity is foundfor the AVD–biotin complex (Fig. 2).

AVD is a glycoprotein with a 10% of carbohydratesin its composition. However, the influence of the glyco-sylation on the thermal stability of unliganded AVDwas negligible [16]. This author reported a Tm of about74°C for both native and deglycosylated AVD but thethermograms were done in buffer containing 1 M ofGuHCl. In absence of the denaturant for unligandedAVD we found a Tm of 83.8°C in agreement with theTm of 85°C reported for this protein by Donovan andRoss [17]. In the presence of GuHCl for unligandedSTV we found a reduction of the Tm by almost 10°C(see below, Table 1). This difference is similar to thatfound by Wang et al. [16] for AVD in presence ofGuHCl. Donovan and Ross [17] reported a Tm of132°C for the AVD–biotin complex, nearly 15°Chigher than the value reported here (Fig. 2 and Table1). However, this higher value may be inaccurate, since

Fig. 2. Avidin and Avidin–biotin thermograms in phosphate buffer.Left: protein without ligand; right: full saturated protein with biotin.The protein concentration was 0.032 mM in both cases. Scan rate andbuffer as indicated in Fig. 1.

M. Gonzalez et al. / Biomolecular Engineering 16 (1999) 67–7270

Fig. 3. Comparison of the thermograms of STV and STV–biotin atfull saturating ligand condition in 8 M urea. Two peaks are present infully liganded protein endotherm, one of them is ascribed to theliganded tetrameric form of STV (at higher temperature). The proteinconcentration was 0.063 mM for all runs. Scan rate as indicated inFig. 1.

also studied by DSC. The unliganded STV melts at73.9°C which is practically the same Tm found in phos-phate buffer (Table 1) but with a lower calorimetricenthalpy. In the full saturating ligand condition, STVhas a thermogram with two endothermic components.In the presence of urea the main endothermic peak iscentered at a Tm of 87°C whereas the second highertemperature endotherm has a Tm at 106°C (Fig. 3).According to the exclusion chromatography data men-tioned above, the dimeric form of STV is the predomi-nant species independent of the presence of biotin.Therefore the main endotherm found in 8 M urea insaturating condition of ligand may be ascribable todimers and the higher temperature endotherm to theremaining tetrameric form. It is not simple to explainthe higher calorimetric enthalpy found at high ureaconcentrations compared to phosphate buffer. This re-sult may be ascribed to an additional interaction ofurea with the protein surface as postulated by Kurzbanet al. [10]. If in 8 M urea the lower main endothermcould be attributed to unfolding of the liganded dimerspecies, it carries near 90% of the total enthalpywhereas the rest is assigned to the tetrameric form. Thisratio between endotherms is not in complete agreementwith the relation found in the exclusion chromatogra-phy experiment described above (60% for dimers and40% for tetrameric form). However, it should be men-tioned that the column chromatography is normallyrun at 22°C and the higher temperature imposed to theSTV–biotin complex in 8 M urea during DSC experi-ment may probably alter the dimer–tetramer equi-librium favoring the dimeric form.

It has been reported by Kurzban et al. [10] that 6 MGuHCl induces slow unfolding of STV with a half-timeof 50 days without altering the tetrameric form. If STVis dissolved with this denaturant and the DSC ther-mogram is run immediately, the STV modifies its ther-mal stability in a rather peculiar manner (Table 1).Unliganded STV substantially decreases its Tm from75.5°C in phosphate buffer to 63.4°C in 6 M GuHCl(around 12°C), whereas under full saturating ligandcondition the Tm is changed slightly from 112.2 to108°C (Table 1). Similar results are observed for theunfolding enthalpies in which, in presence of GuHCland taking into account the values obtained in phos-phate buffer, they have been reduced to 60 and 33% forunliganded and saturated biotin condition respectively(Table 1). It is noteworthy to mention that even whenthe stability of unliganded STV at 6 M GuHCl is highlyaffected, both the tetrameric state and the bindingaffinity are preserved. The thermal stability induced bythe biotin binding in GuHCl condition is also consider-ably high, since the change in Tms (DTm) reaches morethan 44°C (Table 1).

this author used a heating rate 2.5 times faster thanours and it has been reported that as the scan rate isincreased the Tm is shifted to higher temperatures[18,19]. This fact can explain the apparent discrepancybetween both values.

5. The effect of denaturant on the thermal stability ofstreptavidin–biotin complex

It is well known that high concentrations of urea orGuHCl act as denaturants for many proteins [20].However, the resistance to perturbants and denaturantsis a general property of STV. Urea at 6 M does notunfold STV but rather binds to the protein inducing ablue shift of its intrinsic fluorescence [10]. The effect ofurea on STV tetramers integrity has been a matter ofcontroversy. Sano and Cantor [4] suggested that 6 Murea induces a dissociation of a STV tetramer into twosubunit dimers. However, Kurzban et al. [10] hasclaimed that STV in the presence of 6 M urea remainsmainly in the tetrameric form. In contrast and usingfast protein liquid chromatography (FPLC), we foundthat 60% of STV in 8 M urea/sodium phosphate 20mM buffer pH 7.6 either in the absence or presence ofbiotin at saturating condition eluted as dimer form, theremaining 40% eluted as tetramer. For the conditionsassayed, the Superdex-200 column was equilibratedwith the mentioned running buffer, and the molecularweight standards were bovine serum albumin (66 kDa),ovoalbumin (45 kDa) and carbonic anhydrase (29kDa). The thermal stability of STV in 8 M urea was

M. Gonzalez et al. / Biomolecular Engineering 16 (1999) 67–72 71

6. Tetramer integrity is required for a great thermalstability

Independent of either STV or AVD, and the denatu-rant used, urea or GuHCl, the difference in thermalstability of the unliganded state compared to the fulltetrameric liganded state (DTm) is over 30°C (Table 1).If the main lower endotherm found at high urea con-centrations is assigned to STV (dimer)–biotin complex,it should be concluded that the tetrameric condition isalso contributing to the overall thermal stability but inits liganded form. Even when at high urea concentra-tions where a considerable portion of STV remains asdimers, the unliganded protein does not change its Tm,indicating that the dimer–dimer interaction is ratherweak in unliganded condition and this interaction doesnot contribute to overall thermal stability. The entranceof a biotin molecule in the binding site shared by twosubunits has an additional effect besides the increase inmonomer–monomer interaction. The propagated con-formational communication among the dimer subunitspostulated before [6,13] is clearly evidenced by thethermogram observed under subsaturating conditions[see Figure 1 in [2]]. As the exchange of biotin bound toSTV is hard to achieve by free ligand [4], it is reason-able to think that under subsaturating conditions ther-mal unfolding of the unbound subunit occurs at a lowertemperature than the bound subunit due to their differ-ent thermal stabilities (Fig. 1). The lower melting en-dotherm found for the unbound subunits of a tetramerthat has one or two bound biotins has a higher stabilitywith an increased Tm compared to the midpoint denatu-ration temperature of biotin-free STV [Fig. 1; [2]] indi-cating that the neighboring dimer that has bound biotininfluences the unliganded dimer. Therefore, the appar-ent instability of liganded STV induced by 8 M ureashown in Fig. 3 is mainly due to the dissociation of thetetramer into dimers rather than to an influence on themonomer STV conformation. The facts that the ther-mal stability of STV in GuHCl, a denaturant that doesnot influence the oligomeric state of this protein, haspractically the same Tm as the remaining tetramerunder 8 M urea conditions and that both values areclose to that found in phosphate buffer give furthersupport to the idea that the integrity of bound tetrameris needed for the maximal STV thermal stability. Asmentioned before, the biotin binding induces an in-crease in the structural ordering of STV, and the sub-unit barrels become more tightly wrapped. Because thebarrels interact at the dimer–dimer interface, the con-formational change is cooperatively propagated and theliganded STV brings about quaternary changes com-pared to apo STV [6]. The quaternary structuralchanges observed upon biotin binding are not only seenin the modification of the thermal stability but they arealso evidenced from the analysis of the changes in the

cooperative unit of thermal unfolding [2]. The coopera-tive unit defined as DHcal/DHVH, the ratio of calorimet-ric entalphy (area under the transition peak showed inFig. 1) to van’t Hoff entalphy (calculated from theshape of the transition peak), thermodynamically indi-cates whether the protein unfolding represents a simpleor complex cooperative system [21]. This value meansin physical terms the number of independent unfoldingunits per STV subunit (or monomer). The value of thecooperative units ranges from about 2 in apo STV,indicating a rather independent unfolding of one sub-unit with respect to the other, to around 0.5 in fullliganded condition, indicating a marked increment inthe cooperativity of the unfolding [2].

7. Concluding remarks

STV and its homologous AVD in the presence of itsligand biotin have one of the greatest thermal stabilitiesknown for any protein described in biology frommesophilic organisms. To the authors’ knowledge noother proteins have been described with similar featureseven for proteins coming from hyperthermophilic or-ganisms. Both proteins in their liganded state remaincompletely folded at the ordinary water boiling temper-ature. The midpoint temperature of denaturation forSTV and AVD under saturating biotin conditions is112 and 117°C, respectively, in pressurized phosphatebuffer. The integrity of the quaternary structure seemsto be critical for maximal thermal stability. It can beclaimed that a closely packed b-barrel with a highaffinity bound ligand inside may be the main reason forsuch a large increase in thermal stability. However, ithas been reported that the enzyme aldehyde ferredoxinoxidoreductase (AOR) from the Pyrococcus furiosus,the first structurally characterized enzyme from ahiperthermophilic organism growing optimally at100°C, is a homodimer and has a ‘normal’ structuralpattern with a 45% of a-helix and a 14% of b-sheet [22].The only properties shared by STV and AVD withAOR may be the relatively small solvent-exposed sur-face area with a relatively large number of buried atomsincluding an organic ligand co-factor. A close-packedarray, independent of the secondary structure, withmaximal van der Waals and ion-pairs interactions of aprotein may be responsible for the high thermalstability.

Acknowledgements

This paper was supported by CONICOR, SeCyT-UNC, FONCyT and CONICET. CEA and GDF aremembers of the Research Career from CONICET-Argentina.

M. Gonzalez et al. / Biomolecular Engineering 16 (1999) 67–7272

References

[1] Wilchek M, Bayer EA. Anal Biochem 1988;171:1–32.[2] Gonzalez M, Bagatolli L, Echabe I, Arrondo JLR, Argarana

CE, Cantor CR, Fidelio GD. J Biol Chem 1997;272:11288–94.[3] Argarana CE, Kuntz ID, Birkin S, Axel R, Cantor CR. Nucl

Acids Res 1986;14:1871–82.[4] Sano T, Cantor CR. Proc Natl Acad Sci USA 1995;92:3180–4.[5] Green NM. Adv Protein Chem 1975;29:85–133.[6] Weber PC, Ohlendorf DH, Wendoloski JJ, Salemme FR. Science

1989;243:85–8.[7] Hendrickson WA, Palher A, Smith JL, Satow Y, Merritt EA,

Phizackerley RP. Proc Natl Acad Sci 1989;86:2190–4.[8] Grubmuller H, Heymann B, Tavan P. Science 1996;271:997–

9.[9] Green NM, Melamed MD. Biochem J 1966;100:614–21.

[10] Kurzban GP, Bayer EA, Wilchek M, Horowitz PM. J Biol Chem1991;266:1470–7.

[11] Leckband DE, Schmitt FJ, Israelachvili JN, Knoll W. Biochem-istry 1994;33:4611–24.

[12] Miyamoto S, Kollman PA. Proc Natl Acad Sci 1993;90:8402–6.[13] Sano T, Cantor CR. J Biol Chem 1990;265:3369–73.[14] Jones ML, Kurzban GP. Biochemistry 1995;34:11750–6.[15] Shrake A, Ross PD. J Biol Chem 1990;265:5055–9.[16] Wang C, Eufemi M, Turano C, Giartosi A. Biochemistry

1996;35:6299–7307.[17] Donovan JW, Ross KD. Biochemistry 1973;12:512–7.[18] Sanchez Ruiz JM. Biophys J 1992;61:921–35.[19] Lepock JR, Ritchie KP, Kolios MC, Rodahl M, Heinz KA,

Kruuv J. Biochemistry 1992;31:12706–12.[20] Pace NC, Shirley BA, Thomson JA. Measuring the conforma-

tional stability of a protein. In: Protein Structure, a PracticalApproach. IRL press, Oxford University press, 1990, pp. 311–330 [Chapter 13].

[21] Privalov PL. Adv Prot Chem 1982;35:1–104.[22] Chan MK, Mukund S, Kletzin A, Adams MWW, Rees DC.

Science 1995;267:1463–9.

.