Immobilization of Biotinylated hGBP1 in a Defined ... - ACS Publications

Mar 29, 2012 - Guanylate binding proteins (GBPs) belong to the dynamin superfamily ... In the case of human guanylate binding protein 1 (hGBP1) homodi...
0 downloads 0 Views 2MB Size
Download PDF
Article pubs.acs.org/Langmuir

Immobilization of Biotinylated hGBP1 in a Defined Orientation on Surfaces Is Crucial for Uniform Interaction with Analyte Proteins and Catalytic Activity Adrian Syguda,† Andreas Kerstan,† Tatjana Ladnorg,‡ Florian Stüben,† Christof Wöll,‡ and Christian Herrmann*,† †

Department of Physical Chemistry I, University of Bochum, 44801 Bochum, Germany Institute of Functional Interfaces, Karlsruhe Institute of Technology, 76344 Karlsruhe, Germany

‡

ABSTRACT: Guanylate binding proteins (GBPs) belong to the dynamin superfamily of large GTP binding proteins. A biochemical feature common to these proteins is guanosine-triphosphate (GTP) binding leading to selfassembly of the proteins, and this in turn results in higher catalytic GTP hydrolysis activity. In the case of human guanylate binding protein 1 (hGBP1) homodimer formation is observed after binding of nonhydrolyzable GTP analogs like GppNHp. hGBP1 is one of seven GBP isoforms identified in human. While cellular studies suggest heterocomplex formation of various isoforms biochemical binding studies in quantitative terms are lacking. In this work we established a method to study hGBP1 interactions by attaching this protein in a defined orientation to a surface allowing for interaction with molecules from the solution. Briefly, specifically biotinylated hGBP1 is attached to a streptavidin layer on a self-assembled monolayer (SAM) surface allowing for characterization of the packing density of the immobilized protein by surface plasmon resonance (SPR) technology and atomic force microscopy (AFM), respectively. In addition, the enzymatic activity of immobilized hGBP1 and the kinetics of interaction with binding partners in solution are quantified. We present a procedure for attaching an enzyme in a defined orientation to a surface which exposes its active end, the GTPase domain to the solution resulting in a homogeneous population of this enzyme in terms of enzymatic activity and of interaction with soluble proteins.

■

INTRODUCTION Human guanylate binding protein 1 (hGBP1) is a member of a family of GTPases, which play a role in interferon-dependent defense activities of the human organism.1−4 Among other recent findings about the biological function antiviral and antibacterial effects have been reported.5−8 Furthermore elevated levels of hGBP1 in the cerebrospinal fluid were found in patients with bacterial meningitis.9 All proteins of this superfamily consist of a similar structure with a large GTPase (LG) domain (∼300 amino acids) at the N-terminus, an αhelical middle-domain (∼150−200 amino acids) and an αhelical C-terminal domain (100 amino acids), see Figure 1. hGBP1 with a mass of about 67 kDa binds GTP (guanosinetriphosphate), GDP and GMP with similar affinities10 and it catalyzes GTP hydrolysis leading to GDP and GMP.11−14 It could be demonstrated by X-ray crystallography that the structural arrangement of the LG-domain of hGBP1 is very similar to the canonical structure of Ras.15 The elongate shape of hGBP1 is a result of the extension caused by the two helical domains.4 The formation of dimers after binding of the nonhydrolyzable GTP analogue 5′-[β,γ-imido]triphosphate (GppNHp) was shown by size exclusion chromatography and other techniques.16 The nucleotide-free, the GDP and the GMP bound states are monomeric and do not lead to any self© 2012 American Chemical Society

Figure 1. Crystal structure of hGBP1 in complex with GppNHp (red) and Mg2+ (sphere) in ribbon presentation (PDB 1DG3) with the LGdomain colored in blue, the middle domain in orange and the Cterminal domain in green; the anchor position 485 for biotinylation is indicated.

assembly of the protein. By mutational analysis and crystal structures one could identify the LG-domain to be responsible for dimer formation and the interface area is clearly defined.16 While a crystal structure of the full length hGBP1 dimer is not available, the comparison of the structure of the nucleotide-free full length protein and the GppNHp bound LG-domain dimer allows to suggest a model of the full length dimer. It turns out Received: February 27, 2012 Revised: March 29, 2012 Published: March 29, 2012 6411

dx.doi.org/10.1021/la3008359 | Langmuir 2012, 28, 6411−6418

Langmuir

Article

biotinylated hGBP1 (1 μM). Buffer A contains 50 mM TrisHCl (pH 8) and 5 mM MgCl2, and the incubation times were 15 min for each step. Catalytic Activity of hGBP1 in Solution and on the Surface. Activity measurements of the hGBP1 mutant K485C were carried out as described before.25 First the activity was measured in solution at RT (25 °C). The concentrations were 0.5 μM for the protein and 50 μM for GTP. This nucleotide concentration allows precise detection of concentration changes and results in reasonable incubation times of not more than a few hours. After defined incubation times 20 μL aliquots were taken and analyzed for nucleotide concentrations by reversed phase chromatography using a Chromolith Performance RP-18 end-capped column (Merck). The same analysis was used for activity measurements on the surface. As soon as the hGBP1 covered silicon wafer 78.5 cm2 in size was bedewed with a volume of 3 mL of GTP solution at 50 μM the measurement was timed, and 20 μL aliquots were taken for HPLC analysis. SPR Measurements. To get information about binding kinetics, total adsorption and molar ratios of the used protein multicomponent system real-time measurements were performed in a surface plasmon resonance (SPR) system (SR7000DC, Xantec bioanalytics, Germany). Using the biotinylated glass wafer installed in the SPR instrument, first the adsorption of streptavidin (Invitrogen,) at 0.4 μM was monitored which is a constitutive homotetramer with a molecular mass of 53 kDa. This is followed by 0.4 μM biotinylated hGBP1K485C and different concentrations of hGBP1 without anchor in presence of 50 μM GppNHp or GDP. All kinetic measurements were carried out at 25 °C with running buffer containing 10 mM HEPES-NaOH (pH 8), 5 mM MgCl2, 150 mM NaCl, 50 μM nucleotide (GppNHp or GDP), and according to the same protocol: First a buffer flow (5 μL/min) over the surface is applied for 5 min, then the flow of protein solution is started followed after 45 min by laundering for 20 min with buffer. In the case of streptavidin, the incubation time is reduced to 20 min because the binding to the surface is rapidly saturated. Atomic Force Microscopy (AFM). For AFM measurements, the gold-coated silicon wafers were further functionalized by microcontact printing to establish defined surface areas of biotinylation. Height differences between regions, stamped with protein-resistant OEG-thiol,27−30 and regions functionalized with a mixture of biotin-thiol and OH-thiol17,18 can be measured precisely by AFM. The oligoethylenglycol-thiolsolution (OEG(6)-thiol, Prochimnia) was prepared in ultrapure ethanol at a concentration of 1 mM. The gold wafer was cut into 10 mm × 10 mm pieces, rinsed with ultrapure ethanol and dried under a stream of nitrogen. The poly dimethylsiloxan stamp with 3 μm × 3 μm squares was also rinsed with ultrapure ethanol and dried under a stream of nitrogen before being loaded with OEG(6)-thiol by incubation with a small amount of 1 mM solution (80 μL) for 1 min. The stamp was dried in a stream of nitrogen and stepped gently on the gold surface left standing for 2 min. After removing the stamp from the gold surface, the sample was first cleaned with ethanol and dried under a stream of nitrogen. Such laterally structured SAM surfaces were subsequently incubated with streptavidin, biotinylated hGBP1K485C, and hGBP1 without anchor in the presence of GppNHp. After each incubation step AFM images were recorded.

to be head to head attached through the LG domains leading to a highly elongated structure. Here, we set out to characterize the immobilization of hGBP1 at dense packing on a surface. We employed biotindoped self-assembled monolayers (SAM) of mercaptoundecanol covered with a dense layer of streptavidin.17,18 The biotin-streptavidin interaction is exceptionally tight and the four biotin binding sites of the constitutive streptavidin tetramer allow to build multicomponent protein systems.19−22 Here, we bind specifically biotinylated hGBP1 to this surface and quantify the structural and biochemical characteristics of the resulting hGBP1 surface.

■

MATERIALS AND METHODS Preparation of hGBP1 and Biotinylation. HGBP1 was expressed from a pQE80L vector (Qiagen, Germany) in Escherichia coli strain BL21 (DE3). Proteins were purified as described.10 For specific biotinylation of hGBP1, we took advantage of a previously established mutant of hGBP1, which has the following point mutations: C12A, C270A, C311S, C396A, and C589S. Thus, all cysteine residues at the protein surface of hGBP1 accessible for biotinylation are removed. Biochemical properties of this mutant are almost identical to the wild type (wt) as we characterized in previous work.23 In the following this protein is referred to as hGBP1Cys5. The DNA of this mutant was used as a template for introducing K485C mutation by a PCR reaction according to standard protocols (Quik Change, Stratagene). This mutant termed hGBP1K485C was biotinylated according to the instructions of the Maleimide-PEG11-Biotin Biotinylation Kit (Pierce, Rockford, IL). The protein at a concentration of 200 μM was incubated on ice for twelve hours. The number of biotin molecules covalently linked to hGBP1K485C was quantitatively assayed by the HABA biotin quantitation kit (Pierce, Rockford, IL) according to the manufacturerś protocol. Surface Preparation. The dimensions of the D263 thin glasses (Schott) were 10 mm × 10 mm × 0.3 mm. Before preparation of the surface they were rinsed with absolute ethanol (Riedel-de Haën), dried in a nitrogen stream and then layered in a Leypold Inficon XTC/2 metal evaporator. For improving the stability of the 485 Å thick gold layer (deposition rate = 15 Å s−1) a thin titanium layer of about 12 Å (deposition rate = 1 Å s−1) was deposited beforehand onto the glass. Evaporation was performed at room temperature at a pressure of approximately 10−7 mbar. The gold-coated glass slides were incubated in a biotin thiol solution (1 mM, 10% biotin thiol, 90% mercaptoundecan-1-ol) for 2−3 h and then dried in a nitrogen stream. The package of the SAM was checked by Xray photoelectron spectroscopy and IR-spectroscopy as described earlier.25 The coated glass wafers could be directly installed into the SPR spectrometer (SR7000DC) for further surface covering. For measurements of enzymatic activity of hGBP1 on the surface titanium with a thickness of 80 Å was deposited on a polished silicon wafer with a [100] surface termination, followed by a 1200 Å layer of gold.17 This wafer with an area of 78.5 cm2 was placed in a round plastic frame and fixed so that a solution on top of the surface does not leak out. The wafer was then preincubated with ethanolic solution of 10% biotin thiol and 90% mercaptoundecan-1-ol and afterward washed with copious amounts of ethanol and dried in a stream of nitrogen. This was followed by subsequent incubation with aqueous solutions of streptavidin (0.4 μM), buffer A only, and 6412

dx.doi.org/10.1021/la3008359 | Langmuir 2012, 28, 6411−6418

Langmuir

Article

AFM measurements were performed in the liquid cell of a MultiMode NanoScope IIIa AFM (Digital Instruments) fitted with a commercial Si3N4 cantilever of a normal spring constant of 0.12 N/m by approaching the tip to the surface of the sample and scanning. The microscope was operating in a constant force mode where the tip was scanned back and forth at 90° along the horizontal line in a scan range of 15 μm. For all scans the tapping mode was used to avoid disturbance of the damageable soft protein layers. The AFM was operated with a 5 μm z-range and 150 μm x- and y-range scanner (type J Digital Instruments). Topographic images were recorded without removing the liquid cell from the AFM Scanner. The height profile can be directly determined by analyzing the data in SPIP (Scanning Probe Image Processor Software). The measurements were performed in buffer A solution.

■

RESULTS Immobilization of hGBP1K485C. Position 485 in hGBP1 was selected for biotinylation as it is located at the end of this elongate protein opposite to the large GTP binding (LG) domain (Figure 1). In our previous work, we generated multiple mutations of hGBP1 leading to removal of five cysteine residues at the surface of the protein and we demonstrated unaltered biochemical and enzymatic properties of this mutant, hGBP1Cys5.23,24 The introduction of a cysteine residue at position 485 allows specific biotinylation at this position. By incubation with biotinmaleimide the protein obtains its biotin anchor ready for attachment to streptavidin. Using the HABA biotin quantitation kit (Pierce, Rockford, IL) a degree of 97% ± 5% biotinylation of hGBP1K485C was determined. As a control hGBP1 Cys5 yielded only 8% biotinylation. This means hGBP1K485C is most specifically and almost completely biotinylated at position 485. In a second step biotinylated hGBP1K485C was run over the streptavidin covered surface recorded by the SPR instrument. Thus, the amount of this protein adsorbed to the surface could be quantified to 127 ng/cm2. From this a molar ratio of hGBP1K485C and streptavidin of 0.75 is revealed meaning that on average 3 out of 4 streptavidin molecules will carry one hGBP1 molecule each. In a third step the surface can be incubated with soluble analyte molecules like hGBP1wt (without biotin anchor). It is known that hGBP1 forms homodimers after GTP binding. More precisely, the nucleotide GppNHp which serves as a nonhydrolyzable GTP analog induces hGBP1 homodimer formation while GTP binding followed by catalyzed hydrolysis leads to even larger oligomers of hGBP1. This is a typical biochemical feature of large GTP binding proteins. As shown in Figure 2 the SPR instrument reports binding of hGBP1wt to the immobilized hGBP1K485C after starting a flow of 20 μM hGBP1wt in the presence of 50 μM GppNHp. The obtained density of 124 ng/cm2 suggests 1:1 binding of hGBP1wt to immobilized hGBP1K485C. The binding kinetics are addressed below. Only a minute increase of the SPR signal is observed in the control experiment where GDP instead of GppNHp is used. We wanted to compare these findings with the observations made earlier with two other variants of hGBP1, one of which biotinylated at the opposite side of hGBP1 as compared to hGBP1K485C, namely at residue 577 close to the LG-domain end, and the other one double-biotinylated at both ends of hGBP1 at positions 485 and 577. The amount of binding to the streptavidin surface derived from SPR experiments (data not

Figure 2. Subsequent binding of proteins onto a biotinylated SAM surface recorded by SPR. First streptavidin is injected showing rapid saturation. After 50 min biotinylated hGBP1K485C is injected and after 85 min hGBP1WT in the presence of GppNHp. The calculated densities of the proteins on the surface are indicated.

shown) is similar for both single-biotinylated proteins, hGBP1K485C and hGBP1Q577C, but about three times as much compared to the double-biotinylated variant, hGBP1K485C/Q577C. We have reported earlier only little binding of soluble hGBP1wt to immobilized hGBP1Q577C as well as to hGBP1K485C/Q577C.25 We did these experiments again and compared the data directly to the results described above for hGBP1K485C in the presence of GppNHp. In contrast to immobilized hGBP1K485C, for hGBP1Q577C and hGBP1K485C/Q577C the flow of soluble hGBP1wt leads only to a small increase of the SPR signal indicating little binding of hGBP1wt to these two mutants on the surface. Enzymatic Activity. While 1:1 binding of immobilized and soluble hGBP1 strongly suggest full integrity of the protein on the surface it is nevertheless important to check the enzymatic activity. To this end we compare the GTP turnover rate of biotinylated surface attached hGBP1K485C to the rate of biotinylated hGBP1K485C in solution. Figure 3 shows the time courses of GTP hydrolysis at 25 °C catalyzed by these two variants. For estimation of the amount of immobilized hGBP1 mutant on the surface the mass of adsorbed protein calculated from SPR measurements is employed as described above. For the surface area of 78.5 cm2 a hGBP1K485C quantity of 10.0 μg is calculated corresponding to 0.150 nmol. The surface is completely covered with 3 mL buffer A solution containing 50 μM GTP. In the control experiment 0.5 μM soluble hGBP1 and 50 μM GTP were used. The rates were obtained from the slopes in Figure 3 and the specific enzymatic activity was calculated by using the protein quantities on the surface and in solution, respectively. Values of 23 min−1 and 1.33 min−1 are calculated for hGBP1 in solution and bound to the surface, respectively, corresponding to a 17-fold difference. As visible from Figure 3 the incubation time is 5 h at complete surface coverage by hGBP1. Therefore, it was not possible to address the activity of surface immobilized hGBP1 at significantly lower hGBP1 densitiy on the surface as this protein is not stable for days at 25 °C. An experiment at 2-fold lower hGBP1 density on the surface resulted in 2-fold less GTP turnover meaning a similar specific enzyme activity as above but with a larger experimental error. A 10-fold reduced density yielded noisy data points indicating hardly GTP hydrolysis over 6413

dx.doi.org/10.1021/la3008359 | Langmuir 2012, 28, 6411−6418

Langmuir

Article

Figure 3. Specific GTPase activity of biotinylated hGBP1K485C in solution (left) and attached to surface (right). The measurements in solution were performed with 0.5 μM hGBP1 and 50 μM GTP. For calculation of the amount of immobilized hGBP1 on the surface (78.5 cm2) the density obtained in SPR experiments was used (127 ng/cm2) yielding 0.15 nmol. The surface was covered with 3 mL of GTP at 50 μM corresponding to 150 nmol.

represent regions filled with biotin thiol and stepwise covered with streptavidin, biotinylated hGBP1K485C, and finally, hGBP1wt without anchor in the presence of GppNHp. For all protein layers the images indicate a close and consistent layer. By using the SPIP software an average height profile of a selected area (white rectangle in the image) is obtained. In the case of streptavidin a height of (4.05 ± 0.5) nm is observed, which agrees with the dimensions of that protein in its crystal structure (4.0−4.3 nm). For the double-layer of streptavidin and biotinylated hGBP1K485C the height amounts to (15.97 ± 0.5) nm. After subtraction of the height found for streptavidin a value of (11.92 ± 0.5) nm is calculated for the hGBP1 layer. This value agrees well with the length of the hGBP1 protein of 12.9 nm as revealed by the X-ray structure. In the last step unmodified hGBP1wt in the presence of GppNHp was added and the image shows a further increase in height. For the three layers of streptavidin, biotinylated hGBP1K485C and hGBP1wt a height of (18.8 ± 0.5) nm was found. Thus, the binding of the soluble hGBP1 protein to the immobilized one increases the height by almost 3 nm. Binding Kinetics and Binding Constant. The ultimate aim for the hGBP1 layer on the streptavidin surface is its use in the characterization of the interaction between specifically immobilized hGBP1K485C and soluble hGBP1wt or other proteins. Therefore binding kinetics and saturation of binding were addressed in further SPR experiments. The chip surface was layered with streptavidin and biotinylated hGBP1K485C as described above. In the presence of 50 μM GppNHp the binding kinetics of soluble hGBP1wt were examined at various concentrations as shown in Figure 5a. For referencing the hGBP1wt/GppNHp solution was run in parallel over a Streptavidin terminated chip surface. The application for 10 min was sufficient to reach an equilibrium value. The dissociation of hGBP1wt was initiated by switching to running buffer (including GppNHp), resulting in a complete dissociation of residual proteins within a time period of 5−10 min. This kind of experiments was repeated several times yielding reproducible results indicating the immobilized hGBP1K485C was left behind in good shape after each regeneration procedure. For evaluation of the data a single exponential equation was fitted to the association phase and the resulting observed rate

hours. Unfortunately, it is not possible either to address the GTPase activity of immobilized hGBP1 in complex with soluble hGBP1 as this requires a concentration of soluble hGBP1 (>5 μM) which will hydrolyze all GTP within a minute. Nevertheless, another interesting observation was made as to the change of the product ratio GMP/GDP which is 59/41 for this hGBP1 mutant in solution at 25 °C. In contrast, after immobilization of hGBP1 the GMP production is much smaller and the GMP/GDP ratio drops to 23/77. This suggests that the second step of catalyzed hydrolysis is not so efficient when hGBP1 is fixed to the surface in this manner. Thickness of the Protein Layers. To further characterize the surface stepwise modified with defined protein layers as pointed out above, we employed atomic force microscopy (AFM). In Figure 4, the results for patterned SAM surfaces obtained by tapping mode AFM are shown.

Figure 4. Topographic height images after incubation of biotinylated SAM one after another with streptavidin (left), biotinylated hGBP1K485C (middle), and hGBP1wt in the presence of GppNHp (right). The dark bridges indicate the wafer areas, where proteinresistant OEG (6)-thiol was stamped, the bright stripes show the regions loaded with biotin-thiol (10%) SAM, streptavidin, K485C and hGBP1 wt, respectively. The obtained height profiles are shown below the respective panel.

The tapping mode was used to avoid compression of the protein layers as it was reported in earlier studies.25 For microcontact printing of the surface highly protein-resistant OEG-thiol was used to load the 3 × 3 μm stamp regions so that such regular areas on the wafer are free of any unspecific protein adsorption. Thus, the bright stripes in Figure 4 6414

dx.doi.org/10.1021/la3008359 | Langmuir 2012, 28, 6411−6418

Langmuir

Article

Figure 5. (a) Typical SPR sensogramms report binding of hGBP1wt to biotinylated hGBP1K485C on a streptavidin surface after reference subtraction. The running buffer including 50 μM GppNHp contains increasing concentrations of hGBP1wt from bottom to top: 0.3125, 0.625, 1.25, 2.5, 5, and 10 μM. (b) The observed rate constants obtained from exponential fits to the binding curves in panel a are plotted versus the concentration of hGBP1wt yielding the value of kon from the linear slope and koff from the intercept. (c) The maximum values from panel a are plotted as a function of hGBP1wt concentration. A hyperbolic fit to these data yields a KD value of 1.5 μM.

are collated in table 1. Compared to the hGBP1wt measurement almost 10-fold larger KD values were obtained for each of the

constant kobs was plotted versus the concentration of hGBP1wt in Figure 5b. According to eq 1 the rate constants for association and dissociation, kon and koff, can be obtained from the slope of the linear fit and the intercept, respectively. kobs = kon·[hGBP1wt ] + koff

Table 1. Rate Constants and Dissociation Constants Obtained from SPR Experiments for hGBP1wt and Three hGBP1 Interface Mutants Binding to Immobilized hGBP1K485C.a

(1) −1

The intercept value equals koff = 0.28 s , and the linear slope kon = 0.24 μM−1 s−1, while the ratio of these two kinetic constants corresponds to the dissociation constant of the dimer hGBP1 complex, KDkin = 1.17 μM. An exponential fit to the dissociation phase in Figure 5a yields the value for koff = 0.034 s−1 which does not agree very well with the value obtained from the intercept as described above. The 8fold smaller dissociation rate constant obtained in the dissociation phase may reflect rebinding processes to the densely covered hGBP1 surface. Another option for evaluation of the SPR data is making use of the concentration dependence of the equilibrium binding level obtained at the end of the association phase which may be taken as a measure of the fraction of hGBP1wt/hGBP1K485C dimer complex formed. The plot of the response value at the end of the association phase versus hGBP1wt concentration in Figure 5c results in a typical saturation curve which is fitted by eq 2, yielding the equilibrium dissociation constant KDeq of the dimer complex. R eq = R max ·[hGBP1wt ](KDeq + [hGBP1wt ])

wt R240A R245A R244A

koff (s−1)

kon (μM−1 s−1)

KDkin (μM)

KDeq (μM)

0.28 0.92 0.92 0.86

0.24 0.085 0.082 0.079

1.17 10.8 10.8 10.9

1.5 9.8 9.8 10.5

a

KDkin values equal the ratio of the koff and kon values, whereas KDeq is obtained from the fit to the equilibrium level according to eq 2.

three mutants indicating decreased affinity. Basically, this is caused by a 3fold increase of the dissociation rate constant and a 3fold decrease of the association rate constant for all three interface mutants of hGBP1. Again, comparison of the KDeq and KDkin values shows very good agreement of the two results each.

■

DISCUSSION Members of the dynamin superfamily of GTP binding proteins share the ability to self-assemble in dependence of nucleotides. More specifically, hGBP1 has been shown in previous work to form homodimers and larger oligomers induced by GTP binding and GTP hydrolysis, respectively. Without nucleotide or in complex with GDP or GMP it remains in the monomer state. From the crystal structure it is known that the LG domain plays a crucial role in dimer formation by undergoing nucleotide controlled, small structural change.16 The motivation of this work was to establish a stabile hGBP1 surface for SPR assay, also defined in structural terms, for the investigation of hGBP1 interactions including kinetic aspects in order to gain access to self-assembly of hGBP1 as well as to heterotypic interactions with other GBP isoforms and other potential binding partners. The measurements of protein density on the surface by SPR and the measurements of protein layer heights by AFM suggest a dense packing of hGBP1 on the surface, presumably the maximum possible. As hGBP1 is specifically biotinylated at one extreme end of the elongate protein, the one which is opposite the LG domain, it may potentially lay on the surface alongside claiming a footprint area of ∼50 nm2. Alternatively, it may stand upright on the surface directing the LG domain toward

(2)

The obtained value KDeq = 1.5 μM agrees well with the value calculated from the kinetic constants, KDkin. In addition, this agreement indicates that the binding kinetics are not perturbed by mass transport effects and therefore the binding rate is not limited by diffusion. Binding Assay for Hetero Measurements. To further validate the hGBP1K485C surface setup established in this work we tested binding partners with different affinities. Three hGBP1 mutants were selected which have a point mutation each located in the contact area of the hGBP1 dimer. These mutants show reduced dimer complex stability as demonstrated indirectly by the effect on the catalytic activity of hGBP1.31 The chip surface was prepared as described above for immobilization of biotinylated hGBP1K485C and the hGBP1 mutants R240A, R244A, and R245A were added to the running buffer (including 50 μM GppNHp) at various concentrations up to 15 μM used in the SPR experiments (data not shown). The data were evaluated as described for hGBP1WT above and the results 6415

dx.doi.org/10.1021/la3008359 | Langmuir 2012, 28, 6411−6418

Langmuir

Article

have demonstrated a 10 to 20-fold increase of catalytic activity when the equilibrium is shifted from the monomer to the dimer state of hGBP1. It was also clearly shown that dimer formation is established through contacts located in the LG domain leading to enhanced GTPase activity. Our results here show high activity for 0.5 μM hGBP1 in solution where the dimer can be formed easily. In contrast, we observe 17-fold lower activity of hGBP1 on the surface despite dense packing which should favor dimer formation. Earlier kinetic data report a 16-fold increase comparing GTPase activity of the monomer and the dimer.12,26 It does not make sense to explain the lower activity in this work by destruction of the protein through surface immobilization since we observe 1:1 binding of soluble hGBP1 to the immobilized one. This suggests full integrity of hGBP1 on the surface. Rather, we explain this observation by impairment of dimer formation of immobilized hGBP1. We suggest that hGBP1 is so strictly oriented that the required head to head orientation of the protein cannot occur. In fact, the structural model of the dimer demands the LG domains facing each other such that the opposite ends of the proteins which carry the biotin anchor point away from each other leading to a distance of 25 nm from each other. This scenario is not conceivable for two hGBP1 molecules immobilized through that end to the surface. Interestingly, the activity reported here is even lower compared to hGBP1 attached to the surface at the LG end (hGBP1Q577C) or at both ends (hGBP1K485C/577C) leading to alongside orientation in our earlier study. Apparently, these two variants are enabled to form dimers on the surface to some extent whereas immobilization at position 485 of hGBP1 at dense packing leads to complete abolishment of dimer formation and a pure monomer enzyme activity. Further, we have studied the protein binding characteristics of hGBP1K485C on the surface. The binding of soluble hGBP1 is demonstrated most clearly, moreover it is shown that this interaction occurs only in the presence of saturating concentrations of the GTP analogue GppNHp. Another important result is given by the nearly perfect 1:1 ratio of soluble hGBP1 binding to the immobilized one. As mentioned above this can be taken as an indication of completely intact hGBP1K485C on the surface. Strikingly, while the increase in mass of soluble hGBP1 is close to 100% in reference to the immobilized hGBP1K485C the increase in height is only 3 nm which corresponds to the width of hGBP1 and not to the length. In Figure 7 we suggest a model how this might be arranged. In fact, the model of the full length dimer mentioned in the introduction is used in this picture.

the solution, thereby covering a surface area similar to streptavidin (see Figure 6) as estimated from X-ray structure

Figure 6. Structures of hGBP1 (blue, PDB 1DG3) and the streptavidin homotetramer (red, PDB 1SWB) to demonstrate the relative footprint areas supposed hGBP1 is oriented in an upright position on the streptavidin surface. While hGBP1 is an elongate protein streptavidin is almost spherically shaped. On the right-hand side the two proteins are viewed from the top.

dimensions (PDB 1DG3). The increase of 12 nm in protein layer height observed in AFM strongly argues for an upright orientation as the length of hGBP1 equals 12.9 nm while its width ranges between 3.5 and 4.5 nm. Streptavidin covered the surface to a maximum extent as evidenced by the SPR data. A footprint area of 24 nm2 can be estimated from the streptavidin dimension which is similar to the footprint of hGBP1 in the upright orientation on the surface, see Figure 6 for comparison. Taking the observed molar ratio of streptavidin/hGBP1 = 4:3 into account, again, upright orientation of hGBP1 is strongly favored. In fact, giving the observed protein density hGBP1 cannot be accommodated in the alongside orientation as the necessary area of 50 nm2 is not available. Concluding, hGBP1 biotinylated at position 485 will be oriented upright on the surface, not necessarily perpendicular to the surface as there is room for some tilting. Finally it is worthwhile to remark that the LG domain will be homogeneously located at the surface of this protein layer, oriented uniformly toward the solution that is not self-evident for proteins conventionally immobilized on a surface. Concerning the issue of hGBP1 orientation our observation on the enzymatic activity of hGBP1 on the surface as compared to the solubilized hGBP1 is most intriguing. Our previous studies of hGBP1 concentration dependent GTPase activity

Figure 7. Model of hGBP1 dimer immobilized on a streptavidincoated surface. 6416

dx.doi.org/10.1021/la3008359 | Langmuir 2012, 28, 6411−6418

Langmuir

■

The observed rate constants are in the usual range of values for protein/protein association and dissociation rates, respectively, which admittedly cover wide ranges each. Our result about the binding strength of the dimer complex is more interesting as there are no data so far originating from direct interaction experiments. Rather an apparent KD value was reported by us earlier which is obtained from the concentration dependence of the GTPase activity. This value of 0.6 μM15 compares very well to the result obtained here from direct binding studies by SPR, namely, KD = 1.3 μM. In addition we could demonstrate the strength of SPR experiments using hGBP1K485C on the streptavidin surface by addressing the interaction with hGBP1 mutants yielding consistent results. The previous characterization of these mutants demonstrated a 100-fold decrease of the apparent dimer affinity as assayed by the mentioned concentration dependent GTPase activity of these mutants. Note, in such an assay we talk about a homotypic interaction between two mutant proteins while in our study we measure the interaction between the interface mutant in solution and hGBP1K485C on the surface which does not have the mutation in the binding interface. This may explain the difference between the 100-fold decrease in affinity for the complex with two residues missing in the interface and the 10-fold decrease in the cases reported here where only one residue is missing in the dimer interface. There is another finding worthwhile to mention concerning the homogeneity of the hGBP1K485C layer. The mentioned mutants hGBP1Q577C and hGBP1K485C/Q577C were reported to be competent for soluble hGBP1 binding only to a small extent while we report here almost 100% binding competence of hGBP1K485C. This means that this mutant results in a homogeneous population after immobilization which will favor single phase kinetics in binding studies and lead to binding constants which are not averaged over a heterogeneous sample. Altogether we suggest that the setup of SPR experiments established in this study is well suited for the characterization of hGBP1 interactions through the LG domain. Alternatively, when interaction partners are addressed which bind to the helical part of hGBP1 rather than to the LG domain, hGBP1 biotinylated near the LG domain like position 577 may be employed. Summarizing, hGBP1 immobilized at position 485 is homogeneously oriented on the surface presenting the LG domain toward the binding partners in solution. At the same time this defined orientation results in uniform biochemical behavior in terms of enzymatic activity since interaction of the immobilized proteins with each other is prohibited. Therefore, this setup can be used to study in a clean manner the interaction with other binding partners of hGBP1.

■

Article

REFERENCES

(1) Cheng, Y. S.; Colonno, R. J.; Yin, F. H. Interferon induction of fibroblast proteins with guanylate-binding activity. J. Biol. Chem. 1983, 258, 7746−7750. (2) Lipnik, K.; Naschberger., E.; Gonin-Laurent, N.; Kodajova, P.; Petznek, H.; Rungaldier, S.; Astigiano, S.; Ferrini, S.; Stürzl, M.; Hohenadl, C. Interferon γ-induced human guanylate binding protein 1 inhibits mammary tumor growth in mice. Mol. Med. 2010, 16 (5−6), 177−187. (3) Tietzel, I.; El-Haibi, C.; Carabeo, R. A. Human guanylate binding proteins potentiate the anti-chlamydia effects of interferon-γ. PLoS One 2009, 4 (8), No. e6499. (4) Prakash, B.; Praefcke, G. J.; Renault, L.; Wittinghofer, A.; Herrmann, C. Structure of human guanylate-binding protein 1 representing a unique class of GTP-binding proteins. Nature 2000, 403 (6769), 567−71. (5) Anderson, S. L. C.; J., M.; Lou, J.; Xing, L.; Rubin, B. Y. Interferon-induced guanylate binding protein-1 (GBP-1) mediates an antiviral effect against vesicular stomatitis virus and encephalomyocarditis virus. Virology 1999, 256 (1), 8−14. (6) Itsui, Y.; Kurosaki, M.; Kanazawa, N.; Tanabe, Y.; Koyama, T.; Takeda, Y.; Nakagawa, M.; Kakinuma, S.; Sekine, Y.; Maekawa, S.; Enomoto, N.; Watanabe, M. Expressional screening of interferonstimulated genes for antiviral activity against hepatitis C virus replication. J. Viral Hepat. 2006, 13 (10), 690−700. (7) Itsui, Y.; Sakamoto, N.; Kakinuma, S.; Nakagawa, M.; SekineOsajima, Y.; Tasaka-Fujita, M.; Nishimura-Sakurai, Y.; Suda, G.; Karakama, Y.; Mishima, K.; Yamamoto, M.; Watanabe, T.; Ueyama, M.; Funaoka, Y.; Azuma, S.; Watanabe, M. Antiviral effects of the interferon-induced protein guanylate binding protein 1 and its interaction with the hepatitis C virus NS5B protein. Hepatology 2009, 50 (6), 1727−37. (8) Bae-Hoon, Kim.; Shenoy, A. R.; Kumar, P.; Das, R.; Tiwari, S.; MacMicking, J. D. A Family of IFN-γ-inducible 65-kD GTPases Protects against Bacterial Infection. Science 2011, 332, 717. (9) Naschberger, E.; Lubeseder-Martellato, C.; Meyer, N.; Gessner, R.; Kremmer, E.; Gessner, A.; Stürzl, M. Human guanylate binding protein-1 is a secreted GTPase present in increased concentrations in the cerebrospinal fluid of patients with bacterial meningitis. Am. J. Pathol. 2006, 169 (3), 1088−99. (10) Praefcke, G. J.; Geyer, M.; Schwemmle, M.; Kalbitzer, R.; Herrmann, C. Nucleotide-binding characteristics of human guanylatebinding protein 1 (hGBP1) and identification of the third GTPbinding motif. J. Mol. Biol. 1999, 292 (2), 321−32. (11) Cheng, Y. S.; Patterson, C. E.; Staeheli, P. Interferon-induced guanylate-binding proteins lack an N(T)KXD consensus motif and bind GMP in addition to GDP and GTP. Mol. Cell. Biol. 1991, 11 (9), 4717−25. (12) Prakash, B.; Renault, L.; Praefcke, G. J.; Herrmann, C.; Wittinghofer, A. Triphosphate structure of guanylate-binding protein 1 and implications for nucleotide binding and GTPase mechanism. Embo J. 2000, 19 (17), 4555−64. (13) Kunzelmann, S.; Praefcke, G. J.; Herrmann, C. Transient kinetic investigation of GTP hydrolysis catalyzed by interferon-gammainduced hGBP1 (human guanylate binding protein 1). J. Biol. Chem. 2006, 281 (39), 28627−35. (14) Schwemmle, M.; Staeheli, P. The interferon-induced 67-kDa guanylate-binding protein (hGBP1) is a GTPase that converts GTP to GMP. J. Biol. Chem. 1994, 269 (15), 11299−305. (15) Vetter, I. R.; Wittinghofer, A. The guanine nucleotide-binding switch in three dimensions. Science 2001, 294 (5545), 1299−304. (16) Praefcke, G. J.; Kloep, S.; Benscheid, U.; Lilie, H.; Prakash, B.; Herrmann, C. Identification of residues in the human guanylatebinding protein 1 critical for nucleotide binding and cooperative GTP hydrolysis. J. Mol. Biol. 2004, 344 (1), 257−69. (17) Chelmowski, R.; Prekelt, A.; Grunwald, C.; Wöll, C. A case study on biological activity in a surface-bound multicomponent system: The biotin−streptavidin−peroxidase system. J. Phys. Chem. A. 2007, 111, 12295−12303.

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Phone: +49 234 3224173. Fax: +49 234 3214785. Author Contributions

A.S. and A.K. contributed equally to this work Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS Financial support by a grant from Deutsche Forschungsgemeinschaft (DFG) is gratefully acknowledged 6417

dx.doi.org/10.1021/la3008359 | Langmuir 2012, 28, 6411−6418

Langmuir

Article

(18) Grunwald, C.; et al. Fabrication of 2D-protein arrays using biotinylated thiols: results from fluorescence microscopy and atomic force microscopy. Phys. Chem. Chem. Phys. 2004, 6 (17), 4358−4362. (19) Willner, I.; Katz, E. Bioelectronics: From Theory to Applications, 1st ed.; Wiley-VCH: Weinheim, Germany, 2005; Chapter 2, pp 63. (20) Willner, I.; Katz, E. Bioelectronics: From Theory to Applications, 1st ed.; Wiley-VCH: Weinheim, Germany, 2005; Chapter 3. (21) Wilchek, M.; Bayer, E. A. Biotin-binding proteins: overview and prospects. Methods Enzymol. 1990, 184, 49−51. (22) Wilchek, M.; Bayer, E. A. Introduction to avidin-biotin technology. Methods Enzymol. 1990, 184, 5−13. (23) Vö pel, T.; Kunzelmann, S.; Herrmann, C. Nucleotide dependent cystein reactivity of hGBP1 uncovers a domain movement during GTP hydrolysis. FEBS Lett. 2009, 583, 1923−1927. (24) Vöpel, T.; Syguda, A.; Britzen-Laurent, N.; Kunzelmann, S.; Lüdemann, M. B.; Dovengerds, C.; Stürzl, M.; Herrmann, C. Mechanism of GTPase-activity-induced self-assembly of human guanylate binding protein 1. J. Mol. Biol. 2010, 400 (1), 63−70. (25) Kerstan, A.; Ladnorg, T.; Grunwald, C.; Vöpel, T.; Zacher, D.; Herrmann, C.; Wöll, C. Human guanylate-binding protein 1 as a model system investigated by several surface techniques. Biointerphases 2010, 5 (4), 131−8. (26) Kunzelmann, S.; Praefcke, G. J.; Herrmann, C. Nucleotide binding and self-stimulated GTPase activity of human guanylatebinding protein 1 (hGBP1). Methods Enzymol. 2005, 404, 512−27. (27) Herrwerth, S.; Eck, W.; Reinhardt, S.; Grunze, M. Factors that determine the protein resistance of oligoether self-assembled monolayersInternal hydrophilicity, terminal hydrophilicity, and lateral packing density. J. Am. Chem. Soc. 2003, 125 (31), 9359−9366. (28) Pertsin, A. J.; Hayashi, T.; Grunze, M. Grand canonical Monte Carlo Simulations of the hydration interaction between oligo(ethylene glycol)-terminated alkanethiol self-assembled monolayers. J. Phys. Chem. B 2002, 106 (47), 12274−12281. (29) Pertsin, A. J.; Grunze, M. Computer simulation of water near the surface of oligo(ethylene glycol)-terminated alkanethiol selfassembled monolayers. Langmuir 2000, 16 (23), 8829−8841. (30) Prime, K. L.; Whitesides, G. M. Adsorption of proteins onto surfaces containing end-attached oligo(ethylene oxide)A model system using self-assembled monolayers. J. Am. Chem. Soc. 1993, 115 (23), 10714−10721. (31) Wehner, M.; Kunzelmann, S.; Herrmann, C. The guanine cap of hGBP1 is responsible for dimerization and self-activation of GTP hydrolysis. FEBS J. 2011, DOI: 10.1111/j.1742-4658.2011.08415.x.

6418

dx.doi.org/10.1021/la3008359 | Langmuir 2012, 28, 6411−6418