Mutations for decreasing the immunogenicity and maintaining the function of core streptavidin


  • Kyohei Yumura,

    1. Institute of Medical Science, The University of Tokyo, Minato-ku, Tokyo 108-8639, Japan
    2. Department of Medical Genome Sciences, Graduate School of Frontier Sciences, The University of Tokyo, Kashiwa 277-8562, Japan
    Search for more papers by this author
  • Mihoko Ui,

    1. Department of Medical Genome Sciences, Graduate School of Frontier Sciences, The University of Tokyo, Kashiwa 277-8562, Japan
    Search for more papers by this author
  • Hirofumi Doi,

    1. Research Center for Advanced Science and Technology, The University of Tokyo, Komaba, Tokyo 153-8904, Japan
    Search for more papers by this author
  • Takao Hamakubo,

    1. Research Center for Advanced Science and Technology, The University of Tokyo, Komaba, Tokyo 153-8904, Japan
    Search for more papers by this author
  • Tatsuhiko Kodama,

    1. Research Center for Advanced Science and Technology, The University of Tokyo, Komaba, Tokyo 153-8904, Japan
    Search for more papers by this author
  • Kouhei Tsumoto,

    Corresponding author
    1. Institute of Medical Science, The University of Tokyo, Minato-ku, Tokyo 108-8639, Japan
    2. Department of Medical Genome Sciences, Graduate School of Frontier Sciences, The University of Tokyo, Kashiwa 277-8562, Japan
    • Institute of Medical Science, Transmittal form, The University of Tokyo, Minato-ku, Tokyo 108-8639, Japan
    Search for more papers by this author
  • Akira Sugiyama

    Corresponding author
    1. Research Center for Advanced Science and Technology, The University of Tokyo, Komaba, Tokyo 153-8904, Japan
    • Institute of Medical Science, Transmittal form, The University of Tokyo, Minato-ku, Tokyo 108-8639, Japan
    Search for more papers by this author


The defining property of core streptavidin (cSA) is not only its high binding affinity for biotin but also its pronounced thermal and chemical stability. Although potential applications of these properties including therapeutic methods have prompted much biological research, the high immunogenicity of this bacterial protein is a key obstacle to its clinical use. To this end, we have successfully constructed hypoimmunogenic cSA muteins in a previous report. However, the effects of these mutations on the physicochemical properties of muteins were still unclear. These mutations retained the similar electrostatic charges to those of wild-type (WT) cSA, and functional moieties with similar hydrogen bond pattern. Herein, we performed isothermal titration calorimetry, differential scanning calorimetry, and sodium dodecyl sulfate–polyacrylamide gel electrophoresis to gain insight into the physicochemical properties and functions of these modified versions of cSA. The results indicated that the hypoimmunogenic muteins retained the biotin-binding function and the tetramer structure of WT cSA. In addition, we discuss the potential mechanisms underlying the success of these mutations in achieving both immune evasion and retention of function; these mechanisms might be incorporated into a new strategy for constructing hypoimmunogenic proteins.


Core streptavidin (cSA), a protein from Streptomyces avidinii, is well known to have an extremely strong binding affinity to biotin (KD ∼ 10−14M).1, 2 cSA is also highly resistant to heat,3, 4 denaturants,5 and proteolysis.6, 7 Because of its unique properties, cSA has been put to practical use in various areas, including biochemistry,8 and shows therapeutic potential. One possible therapeutic use is in a pretargeting system for drug delivery.9 However, the immunogenicity of this bacteria-derived protein is an obstacle to its clinical application.10–12 One of the most successful approaches for reducing immunogenicity13, 14 is that used for humanizing antibodies.15 Although grafting sequences from the human homolog into a target protein can reduce the immunogenicity of the resulting hybrid molecule, this strategy cannot be applied to cSA because its human counterpart has not yet been found.

Site-directed mutagenesis is another strategy for reducing the immunogenicity of nonhuman proteins.16, 17 Previous reports have suggested that the immunogenicity of nonhuman proteins was decreased through the elimination of charged or aromatic residues on B-cell epitopes by the replacement of these residues mainly with alanine.16, 17 This mutagenesis strategy has been applied successfully to cSA.18 Because charged and aromatic residues on the protein surface are considered to be important for immune recognition of proteins including cSA, these residues not directly involved in biotin binding or tetramerization were mutated.18 This method diminished the immunoreactivity of the mutated cSA to murine and human antistreptavidins, because some of these mutations, especially those at Y83, involved key epitopes for immune recognition.16 However, the mutations also increased the dissociation rate from biotin,16 suggesting that the biotin-binding function was reduced. The relationship between the structure and function of cSA seems to be strict,19 and some mutations reportedly have led to dimerization, destabilization, and decreased binding activity of cSA.19 Therefore, strategies for mutagenesis of cSA should be considered carefully in regard to both the target residues and those to be substituted.

We also previously performed site-directed mutagenesis of various charged and aromatic residues of cSA that were proposed to be involved in its immune recognition (Fig. 1)20 and designed the hypoimmunogenic muteins. The immunoreactivities of these muteins against crab-eating monkey antiserum were evaluated by surface plasmon resonance (Table I and Supporting Information Fig. S1).21, 22 The immunogenicity analyses about T-cell epitope were performed in silico model (Supporting Information Table S1).22–25 Some muteins had exceedingly low immunoreactivities and the number of T-cell epitope was decreased. However, the effects of these mutations on the physicochemical properties and functions of the resulting cSA muteins had not been studied enough previously. Furthermore, mutations induced here are different from alanine. These mutations, except for E116N and E116Q, are considered to retain similar electrostatic charges to those of the wild-type (WT) residues and have functional moieties that form similar hydrogen bonds (Table I).22 Here, we performed thermodynamic analyses to reveal the functions of these cSA muteins and the effects of mutations from a biophysical viewpoint. We evaluated the interaction with biotin by using isothermal titration calorimetry (ITC) and assessed thermodynamic stability by using differential scanning calorimetry (DSC) and sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE). Our results showed that all of the eight muteins we tested bound biotin in the same manner as does WT cSA, and all muteins maintained the thermodynamic stabilities of the WT form. The present results suggest that the strategy of mutating surface target sites by preserving electrostatic charge and functional moieties prevents immune recognition while maintaining the structure and function of cSA. This strategy may offer an innovative method for constructing hypoimmunogenic muteins of other proteins for therapeutic applications.

Figure 1.

The crystal structure and mutated residues of cSA. (A) Tetrameric structure of cSA WT (from Ref. 20). (B) Mutated residues are indicated as stick forms in a monomeric subunit. All mutations were induced at surface residues not directly involved in either biotin binding or tetramerization.

Table I. Mutated Residues and the Immunoreactivities of the Resulting Muteinsa
MuteinMutated residue(s)Relative immunoreactivity to wild-type cSA (%)
  • a

    From Ref. 22.

030Y83S, R84K, E116N11
040Y83S, R84K, E116Q7
314Y22S, Y83S, R84K, E101D, R103K, E116N1
414Y22S, Y83S, R84K, E101D, R103K, E116Q1

cSA, core streptavidin; DSC, differential scanning calorimetry; FFF-MALS, field-flow fractionation–multiangle light scattering; ITC, isothermal titration calorimetry; SDS–PAGE, sodium dodecyl sulfate–polyacrylamide gel electrophoresis; WT, wild-type.


Asymmetric field-flow fractionation–multiangle light scattering analyses of mutant cSAs

Recombinant cSA WT and the muteins listed in Table I were expressed in Escherichia coli Rosetta2 (DE3) cells as insoluble forms and were refolded to obtain cSA tetramers by using stepwise dialysis as described previously, with slight modification.26 Size-exclusion chromatography showed single peaks for all of these proteins except mutein 083, which had a shoulder peak (Supporting Information Fig. S2); this shoulder peak was collected as mutein 083′. Field-flow fractionation–multiangle light scattering (FFF-MALS) analyses to evaluate the molecular sizes and weights of the various cSAs showed that the main peaks identified on size-exclusion chromatography had molecular weights consistent with that of tetrameric cSA WT and that, compared with cSA WT, muteins 314 and 414 might have larger molecular radii in the absence of biotin (Supporting Information Fig. S3).

ITC analyses of cSAs

Because biotin-binding activity is a primary, characteristic property of cSA, we used ITC to measure the thermodynamics of biotin-binding among the hypoimmune muteins [Fig. 2(A)]. The WT form and all of the cSA muteins had very strong binding constants (Ka > 1010 M−1) that exceeded the measurement limit of ITC. We note that the values of stoichiometry N were found in the range 1.0–1.2 in all the constructs examined, except that of mutein 083′ (N = 0.8) [Fig. 2(B)]. The binding enthalpies (ΔHbinds) were similar among cSA WT and all muteins (Table II). Therefore, the ITC analyses implied that all muteins other than 083′ had the same biotin-binding characteristics as those of cSA WT.

Figure 2.

Results of ITC. Ka exceeded the measurement limit of ITC, and ΔS could not be calculated. (A) cSA WT (left); mutein 314 (middle); and mutein 414 (right). (B) 083′.

Table II. Results of ITC
MuteinStoichiometry (N)ΔHbind (kcal mol−1)

DSC analyses of cSAs

Thermal stability is an important factor for therapeutic proteins, such as the antibody fragment single-chain Fv.27 To investigate the thermal stabilities of various cSAs, DSC analyses were carried out. The denaturation temperature (TM) was estimated as the peak top of heat capacity curves. Accordingly, the TMs of cSA WT and the hypoimmunogenic muteins were higher when biotin was bound than in its absence [Fig. 3(A)]. The TMs of muteins 314 and 414 were lower than that of WT in both the absence and presence of biotin. In comparison, the TMs of muteins 083, 030, and 314, which all contain the mutation E116N, were greatly decreased compared with that of cSA WT, whereas those of muteins 001 and 091 were increased (Table III). Only mutein 083′ yielded multiple peaks on DSC analysis [Fig. 3(B)], indicating that its components contained several tetrameric structures. Although the TMs of the hypoimmunogenic muteins were lower than that of cSA WT (Table III), destabilization was negligible, and all muteins demonstrated sufficient thermal stability under the biological condition.

Figure 3.

Results of DSC analysis in the absence and presence of biotin. The denaturation temperature was increased in the presence of biotin. (A) cSA WT, mutein 314, and mutein 414. (B) Mutein 083′, which showed multiple peaks.

Table III. Results of DSC Analysis
 TM (°C)
MuteinBiotin (–)Biotin (+)

SDS–PAGE analyses of cSAs

To analyze the thermal and chemical stabilities of the hypoimmunogenic muteins in more detail, we performed SDS–PAGE analyses of the various cSAs with stepwise heating from 45 to 95 °C.28, 29 Streptavidin retains its tetrameric structure even during SDS–PAGE after heating of samples at low temperature.28, 29 The tetrameric band in SDS–PAGE dissociates to a monomeric band only after heating samples at high temperatures thus indicating the thermal stability of streptavidin.28, 29

Stepwise increases in the sample-heating temperature were associated with progressive migration from high- to low-molecular-weight bands (Fig. 4). The temperatures at which this migration began coincided well with the denaturation temperature determined from the DSC data [Fig. 4(A) and Supporting Information Fig. S6]. Unlike cSA WT and all other muteins, mutein 083′ showed multiple bands at low temperatures [Fig. 4(B)], indicating that this sample may contain several different oligomeric assemblies. The ease with which the high-molecular-weight bands and their migration temperatures could be appreciated indicated that the cSA muteins had sufficient thermal and chemical stability for this analysis, thus indicating the considerable stabilities of cSAs. However, all muteins except cSA WT, mutein 001, and mutein 091 yielded faint low-molecular-weight bands at low temperatures [Fig. 4(A) and Supporting Information Fig. S6), suggesting that the induced mutations slightly decreased the muteins' chemical stability to the denaturant SDS.

Figure 4.

Analysis of stability by using SDS–PAGE. Each sample was prepared as a PBS solution and heated for 3 min at the indicated temperature. (A) cSA WT (left), mutein 314 (middle), and mutein 414 (right). (B) Mutein 083′ showed multiple bands at low temperatures.


Few previous studies have investigated the biophysical properties of cSA muteins in detail. Here, we assessed the thermodynamic properties of several cSA muteins that have low immunoreactivities.

ITC implied that biotin-binding activity was similar among cSA WT and all of the muteins we evaluated (Table II). The migrations of TM due to biotin binding were consistent with those of previous reports (Table III).3, 4 SDS–PAGE analysis showed considerable stabilities (Fig. 4). Because cSA requires a strict tetrameric structure for high biotin-binding affinity,19 we concluded that the induced mutations do not disrupt the conformational relationships required for biotin binding. While the R84I mutation has been reported to cause decreased biotin-binding activity compared with that of cSA WT,18 the R84K variation (mutein 001) did not radically affect Ka or ΔHbind in the current study. This difference may reflect differences in electrostatic charge and functional moieties between isoleucine and lysine. Lysine is similar in electrostatic charge to arginine and likewise carries N[BOND]H bonds in its side chain, which enables it to form similar hydrogen bonds. In contrast, isoleucine is apparently unable to form hydrogen bonds because it carries only C[BOND]H bonds in its side chain.

The mutation E116N (mutein 083) greatly destabilized the resulting cSA protein in stability analyses (Table III), and mutein 083′ showed depression of binding stoichiometry and tetramerization [Figs. 2(B), 3(B), and 4(B)]. In contrast, the mutation E116Q (mutein 091) stabilized the protein (Table III and Supporting Information Fig. S6). This pronounced difference between mutations at E116 and other mutations also seems to reflect the differences in electrostatic charge and functional moiety (carbamoyl group, [BOND]CONH2) from those of the WT residue, glutamic acid (carboxyl group, [BOND]COOH). Our analysis of crystal structural data indicated that the loop and helix near E116 contacts two interfaces of the cSA tetramer (Fig. 5), and mutation of W120 on this helix to lysine was reported to cause dimerization of cSA.30 Although the mutations E116N and E116Q apparently maintain the conformational structure of cSA, changes in electrostatic charge and functional moieties in this region may alter the thermodynamic stability and tetramerization of the resulting mutein. Therefore, our data regarding R84 and E116 strongly suggest that mutations that preserve the electrostatic charge of the original residues and their functional moieties for hydrogen bonds maintain the functions and structures of the proteins for therapeutic application.

Figure 5.

The loop and helix (magenta) near cSA E116 (from Ref. 20) on the interfaces of the tetramer (blue and cyan).

Although the cSA muteins seemed to maintain the tetrameric structure of cSA, the TMs of cSA muteins decreased in DSC analyses [Fig. 3(C)], and monomer bands appeared even at low sample-heating temperatures in SDS–PAGE analyses (Fig. 4 and Supporting Information Fig. S6) except for muteins 001 (R84K) and 091 (E116Q). This indicated that each of these mutations slightly decreased the thermal and chemical stabilities of the resulting muteins. However, the decreased tetramerization ability of mutein 083 (E116N only) at refolding was rescued by inducing additional mutations (mutein 030; Y83S, R84K, and E116N) (data not shown). The key for successful refolding is achieving a balance between structural collapse and flexibility of unfolded structures during treatment with denaturant.31 Our current results suggest that the various mutations we induced in cSA did not totally disrupt its tetrameric structure but instead likely “loosened” it. Although this looseness slightly decreases the thermal and chemical stabilities of the resulting mutein, it may increase the flexibility of unfolded structures during exposure to denaturants and improve the refolding efficiency. The FFF–MALS analyses of muteins 314 and 414 indicated larger molecular radii in the absence of biotin (Supporting Information Fig. S3), in support of our hypothesis of increased structural looseness in our cSA muteins.

We constructed cSA muteins that had decreased immunoreactivity but that preserved the tetrameric structures and biotin-binding functions of the WT protein. Previous reports have suggested that human and murine immune systems recognize cSA through a conformational epitope.18 The mutations induced here were expected to maintain the higher-order structures of cSA; however, these mutations probably also changed the residual structures of side chains exposed on the protein surface. These changes might disrupt the conformation of the cSA epitope and thus achieve immune evasion. In this regard, a low-immunogenicity recombinant of staphylokinase (K35A, E65Q, K74R, D82A, S84A, T90A, E99D, T101S, E108A, K109A, K130T, K135R) had several similar mutations (underlined)16 to those of our hypoimmunogenic cSA muteins. In contrast, the single mutations E116N and E116Q (muteins 083 and 091) increased the immune response to these cSAs (Table I), perhaps due to unexpected effects of structural or thermodynamic changes.

The increased immunoreactivities of muteins 083 and 091 were ameliorated by inducing additional mutations (those in muteins 030, 040, 314, and 414) (Table I). This outcome indicates that looseness of the tetrameric structure might also contribute to increasing immune evasion. This looseness we allude to may generate mobility in the epitope and increase the energy loss associated with antibody binding to prevent immune recognition. Furthermore, the looseness may also contribute to the immunoreactivity correlated to T-cell epitope. It is reported that the high resistance to the lysosomal proteolysis controlled the amount of T-cell epitope and favored its presentation to antigen-presenting cells for longer period, resulting the enhanced immunogenicity of the antigen.32–34 The decreased stability of cSAs we discussed here may weaken the immunoreactivity of T-cells. However, in this view point, the immunogenicity in vivo animal models should be considered, because cSAs become substantially stable in the presence of biotin.

The current study showed that all of our muteins maintained the tetramic structure and functions of cSA WT. We propose that mutein 314 and 414 are the candidates for therapeutic applications. Furthermore, in at least cSAs, our findings also might suggest a new strategy for constructing hypoimmunogenic muteins, in which mutations are induced at charged or aromatic residues on the surface of the target protein; the mutations induced should preserve the electrostatic charge and functional moieties of the WT residues. Consequently, these mutations maintain high-order structures and functions while modifying the residual structures on the surface and generating structural looseness in the resulting protein that leads to immune evasion. Viruses such as picornaviruses are known to escape antibody recognition through substitutions of surface-exposed residues;35 this scenario may be somewhat analogous to the strategy we propose here. Unlike previous strategies such as the construction of chimeric antibodies,15 our current design does not require knowledge of the human counterpart to a target protein. For therapeutics including pretargeting method, we expect that this design will be applied to the other biotin-binding protein such as rhizavidin, the dimeric avidin,36 or other various proteins, and that the immunogenicity of those designed proteins will be evaluated using animal models.

To date, site-directed mutageneses have been performed only to eliminate B-cell epitopes mainly by using alanine,16, 17 and there are no reliable methodologic guidelines for retaining function and structure in the mutated protein. Our current strategy warrants confirmation through its application to other nonhuman proteins with therapeutic potential.

Materials and Methods


pET28b vector was obtained from Novagen (Darmstadt, Germany). Tetramethylethylenediamine, bromophenol blue, and 30% (w/v)-acrylamide/bis (37.5:1) mixed Solution was purchased from Nacalai Tesque (Kyoto, Japan). Other chemicals including biotin were purchased from Wako (Osaka, Japan). Primers were synthesized by Operon (Tokyo, Japan).

Cloning, expression, and purification of recombinant proteins

cSA (residues 13–139 of streptavidin) WT and its muteins were cloned into pET28b vector by using forward primer 5′-GCGGAAGCTGGTATCACTG-3′ and reverse primer 5′-GCTAGCAGCAGAAGGCTTAAC-3′. A His6-tag was fused to the C-terminus. E. coli Rosetta2 (DE3) cells were transformed with cSA plasmids (WT and muteins) and grown at 28 °C in 2 × YT Broth containing 50 μg mL−1 kanamycin. Expression was induced with 1.0 mM Isopropyl β-D-1-thiogalactopyranoside at O.D.600 = 0.8 and then the cells were cultured at 37 °C. Cells were harvested 18 h later by centrifugation (7000 × g, 10 min, 4 °C) and suspended in a buffer comprising 500 mM NaCl and 20 mM Tris-HCl (pH 8.0 at 4 °C). The suspension was sonicated (model UD-201, TOMY, Tokyo, Japan) followed by centrifugation (6000 × g, 30 min, 4 °C). The pellet containing inclusion bodies was washed with wash buffer (2% Triton X-100 and 50 mM Tris-HCl pH 8.0 at 4 °C), acetone, and water purified by reverse osmosis. The pellet then was suspended in buffer A (6 M guanidinium HCl, 500 mM NaCl, 20 mM Tris-HCl; pH 7.9 at 4 °C) containing 5 mM imidazole, followed by shaking overnight at 4 °C. The supernatant was collected by centrifugation (40,000 × g, 30 min, 4 °C) and loaded onto a Ni-NTA resin (Qiagen, Valencia, CA) equilibrated with buffer A containing 5 mM imidazole. Then, the resin was washed with buffer A containing 5 mM imidazole and then buffer A containing 30 mM imidazole. cSA (WT and muteins) were eluted by using buffer A containing 50 mM imidazole and buffer A containing 300 mM imidazole. Eluates were dialyzed against buffer B (200 mM NaCl, 50 mM Tris-HCl, 1 mM EDTA; pH 8.0 at 4 °C) containing 6 M guanidinium HCl for 12 h followed by dialyses against buffer B containing 3 M guanidinium HCl and 2 M guanidinium HCl for 6 h each and buffer B with 0.4 M L-arginine containing 1 M guanidinium HCl and 0.5 M guanidinium HCl for 12 h each. Then the dialysate was dialyzed three times against buffer B for more than 4 h each, followed by centrifugation (6000 × g, 30 min, 4 °C) and filtration over Millex-GP (0.22 µm, PES, 33 mm, sterile; Millipore, Bedford, MA). The supernatant was concentrated by using Amicon Ultra-15 (Millipore, Bedford, MA) and loaded onto a 26/60 Superdex column (GE Healthcare, Piscataway, NJ) equilibrated with buffer B. Eluate fractions were analyzed by using SDS–PAGE, pooled and stored at 4 °C.

FFF–MALS analyses of cSAs

FFF-MALS analyses for the estimation of protein size and weight were done in a Wyatt Eclipse separation system (Wyatt Technology, Santa Barbara, CA). This instrument was equipped with refractive index and MALS detectors. Protein samples were concentrated to 1.8 mg mL−1 and supplemented with excess biotin, when needed. The focused flow rate was 3 mL min−1. Separation was performed by using a cellulose membrane with a nominal molecular weight cutoff of 5 kDa. At the elution step, the channel flow was kept at a constant rate of 1 mL min−1 and the cross-flow decreased linearly from 3 to 0 mL min−1 in 15 min. Buffer B was used for these analyses. The eluted peaks were detected with a refractive index detector, and the molecular weights of these peaks were calculated by using MALS measurement. Data were analyzed with the program ASTRA ver. 5.3 (Wyatt Technology, Santa Barbara, CA) to determine molecular weights.

ITC analyses of cSAs

The thermodynamics of biotin binding were measured by using a iTC200 instrument (MicroCal, Northampton, MA). Purified cSAs were dialyzed against PBS overnight. Biotin was dissolved in the same buffer as that of the cSAs to obtain a 500 μM stock solution. This stock solution was adjusted to a concentration tenfold higher than that of each mutein tested. In a calorimeter cell, 9–20 μM cSAs was titrated with biotin solution at 25 °C. Thermography data were analyzed by using the software package ORIGIN (Light Stone, Tokyo, Japan), and titration curves were fitted to a one-site binding isotherm.

DSC analyses of cSAs

Heat capacity curves were measured by using a VP-DSC ultrasensitive scanning calorimeter (MicroCal, Northampton, MA). The heating rate was 1 K min−1. The sample cell was filled with approximately 0.4 mL of PBS containing 20–50 μM cSA or its muteins. When necessary, samples were supplemented with excess biotin. The data were analyzed by using the software package ORIGIN. The buffer baseline was subtracted from the raw data, which were normalized by protein concentration to obtain thermodynamic parameters and then fitted to two-state thermal transition models.

SDS–PAGE analyses of cSAs

SDS–PAGE analyses were performed as described previously with slight modification.28, 29 Samples of cSAs were prepared as 6 μL aliquots of 9–12 μM PBS solutions and then heated for 3 min at the indicated temperature by using a GeneAmp 9700 PCR system (Applied Biosystems, Tokyo, Japan). Immediately thereafter, 12 μL of sample buffer (80 mM SDS, 62.5 mM Tris-HCl, 8% glycerol; pH 6.8 at room temperature) and 2 μL of 1 mg mL−1 bromophenol blue solution containing 10% glycerol were added to the heated samples. The samples were placed on ice and then loaded onto a 15% polyacrylamide gel containing 0.1% SDS. Electrophoresis was performed at 150 V for 80 min. Gels were stained with 0.5% Coomassie brilliant blue R-250, 50% methanol, and 10% acetic acid for more than 10 min and destained with water.


The authors are grateful to J. M. M. Caaveiro and S. Nagatoishi (The university of Tokyo) for critical reading of the manuscript.