ARTICLE pubs.acs.org/ac
Comparative Assessment of Different Histidine-Tags for Immobilization of Protein onto Surface Plasmon Resonance Sensorchips Marcus Fischer,† Andrew P. Leech,‡ and Roderick E. Hubbard*,†,§,|| YSBL, Department of Chemistry, ‡Technology Facility, Department of Biology (Area 15), and §HYMS, University of York, York, YO10 5DD United Kingdom Vernalis Ltd., Granta Park, Abington, Cambridge, CB21 6GB United Kingdom
)
†
ABSTRACT: Surface plasmon resonance (SPR) is widely used to assess the kinetics and thermodynamics of binding of two molecules. The major challenge is immobilization of one molecule onto the sensorchip for robust detection of binding of the other molecule. We have compared a number of immobilization strategies for noncovalent attachment of an example protein (the substrate binding protein SiaP) by hexa-histidine (His), deca-His, and double-His tags to a nickel-nitrilotriacetic acid (NTA) surface. The stability of immobilization was assessed, and the binding of two low molecular weight ligands, Neu5Ac and 2-keto-3-deoxy-D-glycero-D-galacto-nononic acid (KDN), at different temperatures studied. The hexa-His tagged SiaP washed off from the surface too rapidly for ligand binding to be measured reliably. Systematic variation of chip loading identified conditions under which the deca-His tagged SiaP could generate reliable results. The double-His tagged protein performed as well as covalently attached deca-His tagged protein at 15, 25, and 35 °C. The observed ligand binding kinetics were comparable for all immobilization strategies, and thermodynamic values calculated from SPR are in agreement with solution-based isothermal titration calorimetry measurements. Extended trials suggest that covalent attachment is preferable for screening campaigns, whereas the double-His-tag strategy allows rapid regeneration of the chip, for example, when tight binding compounds are assessed.
S
urface plasmon resonance (SPR) has developed into a widely used label-free approach to monitor binding events and quantify the kinetics and binding constants for the interaction between two molecules. The central principle is to immobilize one molecule onto a sensorchip surface and to measure the change in the refractive index near the surface as another molecule binds. The resulting SPR signal is a function of the immobilization density and the mass of the coupled ligand, accessibility and activity of the protein on the surface, and stoichiometry and mass of the analyte compounds. The last of these is the limiting factor in measuring the binding of small molecules where in most cases the protein is immobilized (for an exception see ref 1) since the refractive index change upon fragment binding monitored in the final sensorgram is dependent on mass coverage and therefore proportional to molecular weight for a given stoichiometry.2 An important assumption in SPR is that tethering of one molecule should not restrict its rotational freedom, diffusional properties, occlude the binding site, or influence other parameters in a way that distorts the information gained on reaction thermodynamics and binding kinetics in comparison to solutionbased methods.3 The major advantages over other solutionbased measurements are the dynamic range of interactions that can be measured (from millimolar to nanomolar), the ready access to kinetic (ka and kd) data, and the small amount of sample r 2011 American Chemical Society
that is required. The major experimental requirement is to immobilize one molecule effectively on the chip surface. The method was first commercialized by Biacore in 1990. Since then, a variety of chip types have been developed (covalent, noncovalent) and there have been continuing improvements in the sensitivity and throughput of the instrumentation. This makes it now routinely applicable for screening and characterizing the binding of small molecules to proteins. A number of different strategies have been developed for attachment of a protein to a surface (for example, see the review by Wong et al.4). For SPR, the covalent attachment methods through surface residues such as lysine have the disadvantage of mixed orientations of the protein on the surface and the variation this will introduce in ligand binding characteristics. Attachment through specific biotin labeling can provide some directionality, but the resulting streptavidin-biotin complex is so strong that the chip cannot be regenerated. An alternative approach is to capture molecules labeled at either the N or C-terminus with oligo histidine (His) on sensorchips that consist of a carboxymethylated dextran matrix preimmobilized with nitrilotriacetic acid (NTA). The capture Received: December 2, 2010 Accepted: January 18, 2011 Published: February 11, 2011 1800
dx.doi.org/10.1021/ac103168q | Anal. Chem. 2011, 83, 1800–1807
Analytical Chemistry via Ni2þ NTA chelation allows the attachment to be carried out without significantly changing the pH or the ionic strength during immobilization. Side-chains of Cys, Tyr, Trp, and Lys on the surface of the protein may in principle also participate in binding to the chelated metal although these affinities are comparatively insignificant. The surface can be readily regenerated by stripping nickel from the surface using a high concentration of ethylenediaminetetraacetic acid (EDTA) which enables the convenient reuse of NTA chips. Alternately, it is possible to use the histidinetag (His-tag) to localize the protein to the surface and then to complete a covalent attachment. In this case, the surface cannot be regenerated. A number of groups have extended the hexa-His-tag paradigm for noncovalent attachment and suggested a strategy where oligoHis sequences that are separated by a number of amino acids are placed on one of the termini. Although few details are presented, Williamson et al. used this strategy to characterize binding of compounds to the protein Hsp70.5 The Taussig group used two hexa-histidines separated by an 11-amino acid spacer among others.6 They also showed that this provided a more robust attachment in comparison to single, triple, and dual N- and C-terminal His-tagged proteins for a variety of Ni-NTA surfacebased applications.7 It is significant that the affinity is not necessarily increased simply by augmenting the number of tags.8 Although each of these immobilization strategies have been reported before, there has not been a detailed comparison of the effectiveness of the different strategies for a single target. The work presented here provides a description of the experimental protocols and data analysis to confirm that similar thermodynamics of the binding of the ligand is observed to differently labeled proteins and that the affinity values obtained correlate with an independent measurement by isothermal titration calorimetry. In addition, we assess the suitability of the different immobilization strategies for use in thermodynamic or kinetic measurements for particular ligands or in the screening of large compound libraries. The system we have used to make this comparison is the binding of ligands to the substrate binding protein SiaP. SiaP forms part of a tripartite ATP-independent periplasmic (TRAP) transporter system in the gram-negative bacterium Haemophilus influenzae.9 Its role is to deliver sialic acid to a membrane transporter for uptake and subsequent decoration of surface lipopolysaccharides, required for efficient immune evasion. A number of different His-tags were used for attachment of SiaP10 to a Ni-NTA chip surface, both noncovalently and covalently. The strategies were assessed in terms of baseline drift and the observed binding constants for 2-keto-3-deoxy-D-glycero-D-galacto-nononic acid (KDN) and Neu5Ac at different temperatures. These ligands were previously reported to bind to SiaP with a KD value of 42 μM and 120 nM, respectively, from stopped-flow fluorescence spectroscopy.11
’ MATERIALS AND METHODS The double-His tag construct was designed on the basis of the deca-His SiaP plasmid.10 It was amplified by polymerase chain reaction (PCR) using forward primer 50 -TCTACTGCTAGCCATCACCATCACCACCATCATCACCATCATGGTGGAG-30 and reverse primer 50 -GTCAATGAGCTCGGATTTTCATTATGGATTGATTGC-30 followed by a directional ligation into pET28a vector using NheI and SacI restriction sites. Therefore the TEV protease cleavage site that is carried within the
ARTICLE
Figure 1. Overview of different His-tagged constructs used within this study. Amino acids for the regions flanking either the N0 or C0 -terminus of the protein sequence are presented as one-letter-code, with, e.g., H6 representing a hexa-His-tag. A TEV cleavage site is indicated by an inverted triangle.
pET28a vector is used as part of the spacer that separates the deca-His-tag of the amplification product from the hexa-His-tag provided by the plasmid. As depicted in Figure 1, the C-terminal hexa-His-tag is directly attached to the SiaP sequence while the N-terminal deca-His-tag is separated from the SiaP sequence by an octapeptide linker of the sequence GENLYFQG. The TEV cleavage site enables cleavage of the tag by TEV protease, e.g., for subsequent crystallization. The double-His-tag construct contains an additional hexa-His separated by a tridecapeptide spacer (SSGLVPRGSHMAS) from the N-terminal to the deca-His-tag. After transformation into Escherichia coli BL21 (DE3) pLysS cells, all three constructs were overexpressed analogous to a procedure described previously.10,11 For the subsequent one-step purification, the sample was dialyzed into buffer A (300 mM NaCl, 20 mM Tris, pH 7.5) and loaded onto a 5 mL His-Trap Ni-Sepharose column, pre-equilibrated with buffer A containing 12 mM imidazole. Each protein was eluted in a single symmetric elution peak, with a high degree of purity, as confirmed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). While a step to an imidazole concentration of 200 mM was sufficient to elute hexa- and deca-His-tagged SiaP, the double-His-tagged protein required a higher concentration of 400-700 mM imidazole to elute. All SPR experiments were conducted using a Biacore T100 biosensor instrument with a Series S NTA sensorchip bound to a dextran support. Experiments were performed at 25 °C unless otherwise specified. The T100 biosensor requires a 10-fold higher concentration of Surfactant P20 (0.05%) than older Biacore instruments to reduce sample loss caused by adsorption of hydrophobic molecules to the surfaces of the flow system. The following SPR buffers and solutions were 0.22 μm sterile filtered and degassed: (1) running buffer, 0.01 M HEPES, 0.15 M NaCl, 0.05 mM EDTA, 0.05% surfactant P20, pH 7.4; (2) nickel solution, 0.5 mM NiCl2 in running buffer; (3) regeneration solution, 0.01 M HEPES, 0.15 M NaCl, 0.35 M EDTA, 0.05% surfactant P20, pH 8.3; (4) normalization solution, 70% glycerol in running buffer. The addition of 50 μM EDTA to the running buffer helped to reduce the bleed-off probably by chelating contaminating metal ions that may be present in components used to prepare the buffers.12 The EDTA solution was sufficiently weak not to strip the nickel from the surface and conversely had a beneficial effect on the immobilization of SiaP and was therefore routinely included in all subsequent runs. Two of the four flow cells (Fc) of the sensorchip were used, one (Fc 1) to monitor nonspecific binding and to provide background corrections for the analysis flow cell (Fc 2) that bears the immobilized protein. After extensive washing with regeneration buffer (10 μL/min for 180 s) and running buffer (20 μL/min until a stable baseline is established), the two flow cells were loaded with nickel solution (20 μL/min for 60 s) in order to saturate the NTA surface with Ni2þ. 1801
dx.doi.org/10.1021/ac103168q |Anal. Chem. 2011, 83, 1800–1807
Analytical Chemistry
ARTICLE
Figure 2. Stability of His-tagged protein on NTA sensorchip surface: (A) hexa-, (B) deca-, and (C) double-His tagged protein was noncovalently immobilized on an NTA sensorchip at three different ratios (the top curve corresponds to the highest coverage) and baseline drift was monitored as absolute response units (RU) over time (seconds).
Each His-tagged protein in running buffer was injected at a flow-rate of 5 μL/min into Fc 2 in order to achieve a target immobilization response of 2000, 4000, and 6000 resonance units (RUs; 1 RU corresponds approximately to 1 pg/mm2 of protein on the sensorchip) as shown in Figure 2. The range was chosen to assess the trade-off between inadequate stability at high RU levels of immobilized protein and insufficient analyte binding response at low RUs. A useful guide to assess the expected change in refractive index upon binding is eq 1: Rmax ¼
MW A RL S MW L
ð1Þ
Rmax is the maximum binding capacity (RU) for the analyte, RL is the response level (RU) of immobilized ligand, MWA/ MWL is the molecular weight of analyte/ligand, respectively, and S is the number of binding sites per ligand. An Rmax value of 30 RU can therefore be expected in an ideal case for an analyte of 350 Da, such as a typical sialic acid, binding to an equivalent of 3000 RU of SiaP (35 kDa). Similarly, lower molecular weight fragments (100-250 Da) are expected to deliver an Rmax between 5 and 20 RU. For comparison, the long-term drift specification of the T100 instrument is 6 representative concentrations chosen so that the lowest gives a measurable response, while the highest reaches steady-state. A range of 10-fold above and below the expected KD was measured in order to fit a statistically valid affinity model to the data. To produce data for kinetic measurements, flow rates of >30 μL/min
were used in order to minimize mass transport effects. The setup was automated using the Biacore software. All injections were performed in duplicate with three zero-concentration blanks, one each at the start and end of the experiment and one in the middle separating an ascending from a descending concentration series. The analytes showed some nonspecific binding to the reference flow cell 1 of the Ni2þ-NTA-chip in the absence of the ligand, which was corrected for by subtracting the sensor reference signal from the binding data observed in flow cell 2. The method of processing the data obtained from an SPR experiment is explained in the legend of Figure 3 for the example of KDN.
’ RESULTS AND DISCUSSION Stability Assessment of Noncovalent Protein Immobilization on the Surface Using Different His-Tags. All strategies
employed a Ni2þ NTA sensorchip for immobilization of SiaP protein with the following tags: C-terminal hexa-His-tag, N-terminal deca-His tag (for noncovalent and covalent attachment), and N-terminal double His-tag (Figure 1). The N- and C-termini are equally suitable attachment points for SiaP as both are remote from the binding site and neither restrict the conformational rearrangement that SiaP undergoes upon ligand binding as seen from crystallography and mechanistic studies.9,10 Proteins were immobilized at three different coverages to test for mass transport and rebinding effects at different concentrations (Figure 2) and the expectation that an unsaturated sensorchip surface may reduce the steric crowding of distributed protein. The first experiments (Figure 2A) used hexa-His tagged protein (available from structural studies of SiaP10). The exponentially decaying baseline indicates that the hexa-His tagged protein cannot be stably immobilized even at a lower concentration of the protein and lower flow-rates. Regardless of the starting concentration of SiaP on the surface, practically all the protein was washed off the chip after about 30 min. The baseline drift was 900 RU/min in the initial rapid phase and around 40 RU/min toward the end of the measurement. This is higher than the predicted response of the analyte (eq 1) which precludes analysis of compounds with low molecular weight (below 500 Da). The fragility of the immobilization also became apparent as the addition of small molecules resulted in detachment of the complex from the surface (data not shown). The first attempts at using the deca-His tagged protein at high concentrations also resulted in too much baseline drift, even after adjusting various parameters such as flow-rates, contact times, and buffer composition, similar to the approach described by 1802
dx.doi.org/10.1021/ac103168q |Anal. Chem. 2011, 83, 1800–1807
Analytical Chemistry
ARTICLE
Figure 3. Processing of SPR raw data into a baseline-corrected sensorgram and affinity plot for the example of KDN (cf. chemical structure) binding to deca-His SiaP (A). The blank subtraction accounts for the drifting baseline (∼6 RU/min) due to the bleeding of protein from the Ni2þ-NTA-chip (B). Kinetic association (ka) and dissociation rate constants (kd) are estimated from a model (black lines) that best describes the experimental data at different ligand concentrations (colored lines). The quotient of the two is the estimated kinetic KD of the reaction. Residuals shown below the plot indicate the deviation and statistic reliability of the observation (D). For 1:1 Langmuir steady-state affinity analysis, the equilibrium rate (plateau of individual analyte concentrations) at which the association constant equals the rate of compound dissociating from its interaction with immobilized protein is used to estimate the KD (C).
Figure 4. Baseline-corrected sensorgrams (left) and affinity plots (right) of Neu5Ac (cf. chemical structure) binding to noncovalently immobilized deca-His tagged SiaP at three different temperatures. Data in the sensorgrams (left) is overlaid with a fit of the kinetic model (black lines) for 15 and 25 °C. At 35 °C, the high bleed-off rate makes it impossible to compensate sufficiently for baseline drifts to use the data for kinetic analysis; however, the affinity plots (right) still produce usable data. For 15 and 25 °C, the duplicate runs are depicted in the same diagram, at 35 °C the two halves for the complete range of concentrations for Neu5Ac were analyzed individually, and the affinity plot of the second half of the data is presented as an inset (see text for details). The sensorgram of Figure 4B has been used for the depiction of a typical SPR result in a review by Fischer et al.17
Nieba et al.13 After an initial phase of rapid bleeding of protein from the chip, the decay flattens off for the highest protein concentration
(Figure 2B). Consequently, lower concentrations were tested and found to result in a more stable baseline possibly due to the 1803
dx.doi.org/10.1021/ac103168q |Anal. Chem. 2011, 83, 1800–1807
Analytical Chemistry
1804
(330 ( 28 nM) 223 ( 59 nM 20 ( 8 μM KD average
(31 ( 31 μM)
(126 ( 48 nM)
23 ( 1 nM
123 ( 49 nM
90 ( 26 nM
2.0 10 1.1 106)
9.2 10 1.1 10 2.8 10 double-His SiaP
1.1 106)
7.0 10 ) 2.2 10-8 (= 2.4 10-2/ -7
1.4 10 ) n.d.
a
KD values given in molarity at different temperatures; (kd/ka may 6¼ KD due to rounding errors). KD average values in parentheses are not trustworthy as described in the text. n.d. = not uniquely determined; underlined values are the estimated KD which is outside the range that can be measured; values in italics are kinetic constants approaching limits that can be measured by the instrument.
1.4 106)
7.6 105) 3.5 10-7 (= 4.9 10-1/ -7
7.1 10 ) 7.2 10-8 (= 7.6 10-2/ -8
5 5
1.8 10-7 4
5.3 10-5 (= 7.1 10-1/ 2.1 10-5 covalent Deca-His SiaP
6.0 104)
1.2 10 Deca-His SiaP
8.6 10
2.4 10-8 (= 1.7 10-2/
1.8 10-7
1.4 107)
1.2 10
9.8 10 (= 2.9 10 /
1.3 106)
2.3 10 (= 5.1 10 /
8.8 10
-8 -5
affinity KD
7.8 10-8 (= 5.6 10-2/
2.9 10-7
n.d. (= 1.7 10 /
1.8 10
-7
affinity
-0
kinetics
-7 -8
affinity
-2 -8
kinetics kinetics
-6
-1
affinity
15 °C 25 °C temperature
-5
3.1 10-7 (= 2.4 10-1/
kinetics 35 °C 25 °C
Neu5Ac
increased probability that His-tagged molecules rebind to a neighboring Ni-NTA group by diffusional transport when not all sites are fully occupied. The immobilization of an equivalent of about 3000-4000 RUs (Figure 3) resulted in a sufficiently stable setup to conduct binding experiments with the test compounds (see below and Table 1). The bleeding from the surface for this lower concentration of protein was about 6 RU/min (Figure 3A). However, it is necessary to check that the rate constants of analyte binding to the protein are not influenced by other events such as association and dissociation of rebinding protein as it diffuses across the surface13 or at least is giving a systematic error. In order to distinguish between saturation (capacity of the surface) and equilibrium binding responses, care was taken that the data satisfies eq 1. Finally the immobilization stability of N-terminally double His-tagged protein was assessed. The double His-tag can potentially bind to more than one NTA site. This will reduce the steric hindrance of proximal protein but also reduce the amount of protein that can be immobilized on the chip and consequently Rmax (eq 1), which is critical to detect low molecular weight compounds above the background noise. Stability is reached even at the highest level of surface saturation (Figure 2C) where the rate of bleeding off was ∼10 RUs/min, which provides sufficient stability for detailed kinetics measurements (see below). However, after 2700 RUs of protein is loaded onto a chip, the amount remaining after 10 h of screening was 1600 RUs, corresponding to an average bleed-off of
