Molecular-Level Insights into Orientation-Dependent Changes in the

May 14, 2015 - Andrea ArsiccioJames McCartyRoberto PisanoJoan-Emma Shea. The Journal of Physical .... Shuai Wei , Logan S. Ahlstrom , Charles L. Brook...
0 downloads 0 Views 3MB Size
Download PDF
Article pubs.acs.org/Langmuir

Molecular-Level Insights into Orientation-Dependent Changes in the Thermal Stability of Enzymes Covalently Immobilized on Surfaces Tadeusz L. Ogorzalek,† Shuai Wei,† Yuwei Liu,† Quiming Wang,† Charles L. Brooks, III,†,§ Zhan Chen,†,‡ and E. Neil G. Marsh*,†,∥ †

Department of Chemistry, University of Michigan, 930 North University Avenue, Ann Arbor, Michigan 48109, United States Department of Macromolecular Science and Engineering, University of Michigan, 2300 Hayward Street, Ann Arbor, Michigan 48109, United States § Department of Biophysics, University of Michigan, 930 North University Avenue, Ann Arbor, Michigan 48109, United States ∥ Department of Biological Chemistry, University of Michigan, 1150 West Medical Center Drive, Ann Arbor, Michigan 48109, United States ‡

S Supporting Information *

ABSTRACT: Surface-immobilized enzymes are important for a wide range of technological applications, including industrial catalysis, drug delivery, medical diagnosis, and biosensors; however, our understanding of how enzymes and proteins interact with abiological surfaces on the molecular level remains extremely limited. We have compared the structure, activity, and thermal stability of two variants of a β-galactosidase attached to a chemically well-defined maleimide-terminated self-assembled monolayer surface through a unique cysteinyl residue. In one case the enzyme is attached through an α helix and in the other case through an adjacent loop. Both enzymes exhibit similar specific activities and adopt similar orientations with respect to the surface normal, as determined by sum-frequency generation and attenuated total reflectance FT-IR spectroscopies. Surprisingly, however, the loop-tethered enzyme exhibits a thermal stability 10 °C lower than the helix-tethered enzyme and 13 °C lower than the enzyme in free solution. Using coarse-grain models, molecular dynamics simulations of the thermal unfolding of the surfacetethered enzymes were able to reproduce these differences in stability. Thus, revealing that tethering through the more flexible loop position provides more opportunity for surface residues on the protein to interact with the surface and undergo surfaceinduced unfolding. These observations point to the importance of the location of the attachment point in determining the performance of surface-supported biocatalysts and suggest strategies for optimizing their activity and thermal stability through molecular simulations.

■

INTRODUCTION The immobilization of enzymes on abiological surfaces plays a central role in a wide range of important technological applications, including industrial catalysis, drug delivery, medical diagnosis, and biosensors.1,2 Depending upon the application, immobilization may prolong the useful lifetime of the enzyme or facilitate its removal from the reaction and reuse. The attachment of enzymes to surfaces is known to significantly affect both enzyme activity and thermal stability, issues that play an important role determining the economic feasibility of using enzymes in biotechnological processes.1−4 Despite these important applications, our understanding of how enzymes and proteins interact with abiological surfaces on the molecular level remains a challenging problem.5 In part, this reflects the approaches that have been traditionally used to © XXXX American Chemical Society

immobilize enzymes, which have relied on nonspecific adsorption through electrostatic or hydrophobic interactions or nonspecific covalent cross-linking through the amino groups of surface lysine residues.3,6,7 Such methods are simple to employ but result in poorly defined, heterogeneous mixtures of proteins that are attached in different orientations and may be partly unfolded or constrained in inactive conformations or orientations. This has spurred numerous investigations into the interaction of enzymes with different types surface substratum that have employed a wide range of strategies to attach proteins to the surface in better defined manner and used a wide range Received: March 5, 2015

A

DOI: 10.1021/acs.langmuir.5b01735 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

Figure 1. (A) Structure of maleimide-terminated SAM on silica surface used in these studies to facilitate covalent attachment of proteins. (B) Structure of β-gal, with active site loop indicated. The two surface residues, E147 (helix) and V152 (loop), that were mutated to cysteine to facilitate surface attachment are highlighted in gold.

(ATR) FT-IR spectroscopy21 indicates that the loop-tethered enzyme and the helix-tethered enzyme have similar orientations at room temperature; however, for the loop-tethered enzyme the range of possible orientations deduced from spectroscopy is larger, indicating that it is likely more mobile and thus more likely to suffer surface-induced unfolding than the helixtethered enzyme, in accord with the coarse-grain simulations.

of experimental techniques to probe the interaction of the protein with surface.8−16 In our studies we have used engineered enzymes that contain unique cysteinyl residues introduced at the desired attachment point on the protein’s surface.17,18 This allows the enzyme to be covalently attached to a suitable maleimide-functionalized surface, in a chemically well-defined manner. This approach yields a far more homogeneous population of surface-tethered protein molecules, making it possible to examine how changing the tethering site alters the interaction between protein and surface and its effects on structure, activity and stability. Here we report studies on 6-phospho-β-galactosidase (βGal) from Lactobacillus lactis,19 a representative of the class of glyco-hydrolases that have important uses in a variety of technological applications.20 We examined the properties of two enzyme variants tethered to a surface formed by an (ethylene glycol)4-maleimide-terminated self-assembled monolayer (EG4-maleimide SAM, Figure 1). In one case the enzyme was tethered through a flexible surface loop, in the other, through a more rigid α-helical element. The two immobilized enzymes have similar activities at room temperature; however, their thermal stabilities differ significantly. Whereas the enzyme tethered at the α-helix position has thermal stability similar to β-Gal in solution, the thermal stability of loop-tethered enzyme is significantly lower. Coarse-grain molecular dynamics simulations of the surface-tethered enzymes were able to recapitulate the experimentally determined thermal inactivation curves and facilitate a more detailed analysis of the orientation and fluctuations of the proteins as a complement to our spectroscopic analysis. Comparison of the unfolding trajectories for the tethered enzyme with the enzyme in free solution demonstrates the important role of surface−protein interactions in the unfolded state. Further characterization of the enzymes using the surface-sensitive techniques sum-frequency generation (SFG) spectroscopy and attenuated total reflectance

■

MATERIALS AND METHODS

Design and Expression of Modified β-Galactosidase Constructs. The design of synthetic gene, codon-optimized for expression in Escherichia coli, encoding a variant of 6-phospho-β-galactoside (βGal) from Lactobacillus lactis (PDB entry 2PBG)19 in which all of the native cysteine residues have been replaced by alanine, together with its expression and purification from E. coli BL21(DE3) cells, has been previously described.18 To introduce a unique cysteine into a surface exposed loop, we mutated Val-152 to cysteine by standard methods; similarly, to introduce a unique cysteine into a surface-exposed α-helix, we then mutated Glu-147 to cysteine. These proteins were expressed and purified in the same way as the “no cysteine” β-Gal variant.18 Enzymes were stored at concentrations of 50−100 μM at −80 °C. Functionalization of Glass Beads for Enzyme Assay. 75 μM diameter acid-washed glass beads (Sigma) were incubated overnight in 1 mM of maleimide-PEG4 − silane (Creative PEG Works, Winston, Salem, NC) in toluene. The beads were then washed with toluene, followed by methanol and 100 mM potassium phosphate before being vacuum-dried. 200 mg aliquots of dry maleimide-functionalized beads were incubated in 1 mL of 100 mM potassium phosphate buffer, pH 7.6, 0.1 mM TCEP, containing 5 μM of β-GalV152C or β-GalE147C respectively, prereduced with 1 mM TCEP, for 4 h at room temperature with gentle shaking to attach enzymes. The beads were then washed with 3 × 1 mL of 100 mM potassium phosphate buffer and used immediately. Enzyme Assay. Enzyme activity was determined using the fluorogenic substrate resorufin-β-galactopyranoside (Life Technologies, Grand Island, NY). All assays were performed in 1 mL of buffer containing 100 mM potassium phosphate, pH 7.6, 1% DMSO, and 10 pmol of β-Gal. Assays were initiated by the addition of resorufin-βB

DOI: 10.1021/acs.langmuir.5b01735 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir galactopyranoside to a final concentration of 50 μM. The formation of resorufin was measured using fluorescence, with excitation at 571 nm and recording emission at 584 nm. For enzyme tethered to beads, 18− 20 mg of beads was suspended in 1 mL of buffer in a 1.5 mL Eppendorf tube with shaking. After 1 min of shaking, the beads were allowed to settle before a 750 μL aliquot was transferred to a cuvette, and the fluorescence was measured. The aliquot was then transferred back to the Eppendorf tube and shaken for a further 1 min, and the process was repeated. Typically 10 time points were recorded for each rate measurement. The surface density of enzyme attached to glass beads was quantified using the bicinchoninic acid assay previously discussed.8 Thermal Stability of β-Gal. 100 μL aliquots of a solution containing 100 nM β-gal in 100 mM potassium phosphate buffer, pH 7.6, were heated at temperatures ranging from 25−60 °C for 10 min using a thermocycler, followed by a rapid cooling to room temperature to examine the thermal stability of β-Gal variants in free solution. The enzyme solution was diluted to 10 nM prior to the assaying for residue enzyme activity at 25−60 °C. 18−20 mg aliquots of enzyme-functionalized beads, corresponding to 10 pmol of enzyme, were suspended in 100 μL of 100 mM potassium phosphate pH 7.6 and were heated and cooled as described to examine the thermal stability of β-gal variants tethered to glass beads. The bead suspension was diluted with 900 μL of roomtemperature buffer added prior to assay for residual enzyme activity. To estimate Tm and the slope of the curve at Tm, we fit thermal stability data to eq 1 as previously described22,23

y = A2 +

with the surface. The s- and p-polarized ATR-FTIR spectra of the enzyme covalently tethered to the SAM were recorded after the system reached equilibrium and used for orientation analysis, as previously described.18,25 Comparison of Protein Surface Coverage. The protein coverage of β-Gal immobilized on SFG prisms and on glass beads was compared by XPS. Measurements were performed on a Kratos Axis Ultra XPS under vacuum at ∼1e−9 Torr. A monochromatic Al Xray source was generated from aluminum anode operating at the emission voltage at 14 keV and 6 mA. Spectra were analyzed by CASAXPS. The spectra were referenced by setting C 1s peak to 284.5 eV to compensate for residual charging effects. The N 1s peak was used to compare protein coverage on prisms and beads (as the EG-4 maleimide SAM contributes significantly to the C 1s and O 1s peaks) and normalized with respect to the Si 2p peak (data not shown). The surface coverage of β-Gal was determined to be very similar between both surfaces: bead/prism N 1s intensity ratio = 0.94. Coarse-Grain Molecular Dynamics Simulations. Enzyme simulations used the Karanicolas and Brooks’ structure centric (Go) protein model.26,27 This model describes each residue by one site placed at the Cα position of the residue. Native contacts are defined in this model based on the hydrogen bonding between backbone atoms or side-chain/side-chain interactions.26 It has been shown that using this model the protein folding free energy surface and the folding mechanisms are consistently reproduced.26−29 The simulations described here used a recently developed coarsegrain model of protein−surface interactions30 based on and incorporating the force field of Karanicolas and Brooks.26,27 The potential function is represented by eq 2, in which the first three terms of the potential function between the protein and the surface successfully capture the adsorption well and the energy barrier features as observed in many experimental works.13,16 Furthermore, the twothird power terms were added to the function to account for hydrophobic effects of different self-assembled monolayer (SAM) surfaces and different residues in a protein or peptide by using the hydrophobic index of the surfaces χs and amino acids χp. The five-term model was well-parametrized, and parameters used in this study are the same as previously published.30

A1 − A 2

(

1 + exp

Tm − T slope

)

(1)

where A1 and A2 are the upper and lower asymptotes, respectively, of the enzyme activity and Tm is the temperature at which 50% of the initial activity remains. Sample Preparation for SFG and ATR FTIR Vibrational Spectroscopic Analyses. Right-angle CaF2 prisms (Altos Photonics, Bozeman, MT) coated with 100 nm SiO2 were reacted with 1 mM maleimide-EG4-silane in anhydrous toluene for 24 h at room temperature to produce EG4-maleimide-terminated SAM surface for protein attachment, as previously described. SFG theory and applications have been extensively previously published18,21,24 and are therefore not detailed here. SFG spectra were recorded on a custom-made apparatus purchased from EKSPLA, Vilnius, Lithuania; details of the experimental setup have been previously described.18 In this study, near-total-reflection geometry was used with the EG4-maleimide-functionalized right-angle CaF2 prisms. Proteins were covalently attached to the prisms by immersing the surface in a 2 mL reservoir containing 5 mM pH 7.2 phosphate buffer (PB) and 0.1 mM TCEP. The appropriate volume of an enzyme stock solution, prereduced with 1 mM TCEP at room temperature for 2 h to reduce any potential disulfide bonds, was added to the reservoir to a final concentration of 4 μM. After the system was equilibrated, SFG spectra with a polarization combination of ssp (s-polarized sum frequency output, s-polarized visible input, and p-polarized infrared input) and ppp were collected and used for orientation analysis, as previously described.18,25 All SFG spectra were normalized according to the intensities of the input IR and visible beams. ATR-FTIR spectra were recorded using a Nicolet Magna 550 FTIR spectrometer. Experiments were carried out using ZnSe total internal reflection crystal (Crystran, Dorset, England) deposited with a 50 nm layer of SiO2 and functionalized with EG4-maleimide SAM as previously described. 1.6 mL of 5 mM phosphate buffer, pD 7.2, containing 0.1 mM TCEP in D2O was added to the trough above the SAM-functionalized crystal; D2O was used to avoid possible signal confusion between the O−H bending mode and the peptide amide I mode and to ensure a better S/N ratio in the peptide amide I band region. After recording background spectra, the appropriate volume of an enzyme stock solution, prereduced with 1 mM TCEP at room temperature for 2 h to reduce any potential disulfide bonds, was added to the reservoir to a final concentration of 4 μM and allowed to react

Vsurface =

N

⎧ ⎪

i

⎩

⎡ ⎛ ⎞9 ⎛ σ ⎞7 ⎛ σ ⎞3 σi ⎟ − θ2⎜ i ⎟ + θ3⎜ i ⎟ ⎢⎣ ⎝ zis ⎠ ⎝ zis ⎠ ⎝ zis ⎠

∑ ⎨⎪πρσi3ϵi⎢θ1⎜

⎛ σ ⎞3⎤⎫ ⎪ − (θs(χs − 4.5) + θpχp )⎜ i ⎟ ⎥⎬ ⎪ ⎝ zis ⎠ ⎥⎦⎭

(2)

The hydrophobicity of the maleimide SAM surface, which is the SAM surface examined in this work, is set to be moderately hydrophilic with a χs of 1.5 to represent the measured contact angle of 65° from previous experimental work.31 The bond between the maleimide surface and the cysteine thiol is simulated with a harmonic restraint with an interaction potential of the form Urestraint =

1 k r(r − req)2 2

(3)

where kr = 10 kcal/mol is the parameter describing the strength of the tethering restraint, r is the distance of the tethering site from the origin of the surface (0, 0, 0), and req = 5.8 Å is the equilibrium distance from the tethering site to the surface origin. The tethering length of 5.8 Å approximates the distance between the maleimide surface and the Cα of the cysteine residue at the tethering site. β-Gal was simulated using the Go-like model previously described starting with the previously determined X-ray structure PDB ID 2PBG. β-Gal was tethered to the moderately hydrophilic surface with a cysteine mutated at either residue 147 or residue 152. Both locations for the cysteine residue, when tethered to the maleimide monolayer, orient the active site toward the bulk solution. Multiple folded and unfolded samplings are required in the simulation to obtain protein thermal stability data in the bulk and on the maleimide surface with different tethering sites. This was C

DOI: 10.1021/acs.langmuir.5b01735 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir Table 1. Temperatures of Replicas Used in REMD Simulations replica T (K) replica T (K)

1 270 13 317.5

2 280 14 320

3 285 15 322.5

4 290 16 325

5 295 17 327.5

6 300 18 330

achieved by using the replica exchange molecular dynamics (REMD) method.32−35 24 replicas with temperatures starting at 270 K and increasing to 360 K were used in all three cases (in the bulk and on the surface with site 147 and site 152). Swaps were attempted every 2000 steps, and temperature increments between adjacent replicas ranged from 2.5 to 10°; see Table 1. The smaller increments were used close to the melting temperature and the larger increments further away. The canonical ensemble was used for each replica, and the temperature was maintained by the Nosé−Hoover−Chain integration method with three thermostates of mass 10−26 kg Å2.36−38 Each simulation was performed with 10 million steps of equilibrium and 30 million steps of production with the time step of 1 fs/step. A small step size is used to avoid residues from moving beyond the surface. For the coarse-grained model force field used here, the absolute energy of the residue−residue interaction is arbitrarily assigned so that the model does not reproduce an accurate melting temperature. The force field of the Go-like model used in this work was parametrized so that most proteins have a melting temperature around 350 K; the melting temperature of a protein can be shifted to any value by linearly scaling the energy of native contacts defined in this model. For the present work, the Tm values calculated for β-gal in bulk aqueous solution and the enzyme tethered to surfaces were all linearly scaled by 5 K to facilitate comparison between experiment and computation. This aligns the calculated and experimental values of Tm for the enzyme in bulk aqueous solution but does not change the calculated differences in Tm for the surface-tethered enzymes. Calculation of Thermodynamic Quantities. The metrics used to quantify stability were calculated from simulation data using standard methods from statistical mechanics. The melting point, Tm, is determined as the temperature at which only 50% native contacts are present, which will be shown as the transition point of the fractional nativeness curve. The instantaneous fractional nativeness, Q, is the ratio of the number of native contacts formed at a particular instance relative to the total number of native contacts possible. From the simulations, the average of the fractional nativeness, Q, can be calculated using eq 4 Q (T ) = ⟨Q ⟩T =

8 305 20 335

9 307.5 21 337.5

10 310 22 340

11 312.5 23 350

12 315 24 360

of its activity and its spectroscopically deduced orientation with respect to the surface is consistent with its attachment point. As previously discussed,8 the surface of a 75 μm bead is assumed to be atomically flat and physically identical to a flat surface when compared with the nanometer-scale dimensions of a protein. In this study we wanted to compare the effects on structure, activity, and stability of tethering an enzyme through a flexible element, such as a loop, with tethering through a rigid structural element, such as an α-helix. This could potentially alter activity through noncovalent interactions between the enzyme and surface or possibly affect large-scale, low-frequency vibrational modes that have been shown to be important for catalysis in a number of enzymes.43,44 A previously described variant, β-Gal V152C, allowed the enzyme to be tethered through a flexible loop, and by introducing a cysteine residue at Glu-147 (β-Gal E147C variant) we facilitated attachment at an adjacent surfaceexposed α-helical position (Figure 1). The close proximity of these two attachment points was intended to preserve the orientation of the enzyme with respect to the surface. Steadystate kinetic analysis of these β-Gal variants in free solution, using 2-nitrophenol-galactose as a substrate,18 indicated that the introduction of cysteine at these positions has no effect on enzyme activity. For β-Gal V152C, kcat = 0.18 ± 0.02 s−1 and KM = 0.17 ± 0.02 mM; for β-Gal E147C, kcat = 0.16 ± 0.02 s−1 and KM = 0.21 ± 0.02 mM. These values are within error of the values for the wild-type enzyme40 and a previously engineered “no cysteine” variant.18 Both β-Gal variants were covalently coupled through their respective cysteinyl residues to 75 μm glass beads that had been functionalized with EG4-maleimide-terminated SAM (Figure 1A). The variants coupled with similar efficiency; typically 0.03 μg protein/mg of beads was incorporated, corresponding to an approximate surface coverage of 2 molecules of β-Gal per 100 nm2. Assuming β-Gal to be roughly spherical with a diameter of 5 nm, this corresponds to ∼38% of the surface covered by protein. This value is similar to that expected for a monolayer of protein, assuming the beads to be uniform spheres and the glass surface to be atomically flat The specific activity of the surface immobilized enzymes was determined using the more sensitive fluorogenic substrate resorufin-β-galactopyranoside (rbg). In solution, the β-Gal variants exhibited similar specific activities: 2.7 ± 0.2 nmol/ min/mg protein for β-Gal V152C and 2.6 ± 0.3 nmol/min/mg protein for β-Gal E147C. Upon coupling to EG4-maleimideSAM, the specific activity of the enzymes decreased slightly: 2.0 nmol/min/mg protein for surface-tethered β-Gal V152C and 1.8 nmol/min/mg protein for surface-tethered β-Gal E147C. Thermal Stability of Immobilized Enzymes. Immobilized enzymes often display quite different thermal stabilities than those in free solution. Aliquots of enzyme, either in free solution or covalently tethered to glass beads, were heated at temperatures ranging from 24−60 °C for 10 min and then cooled to room temperature using a thermocycler to determine how tethering β-Gal to the SAM surface affected the thermal stability. The enzymatic activity remaining was measured and

∑U Q (U )Ω(U )e−βU ∑U Ω(U )e−βU

7 302.5 19 332.5

(4)

where β = 1/kBT and kB is Boltzmann’s constant and T is the temperature. U is the potential energy, and the ’s denote the average of the corresponding quantities. The key quantity needed to evaluate eq 3 is the density of states, Ω(U), which is calculated using the weighted histogram analysis method (WHAM)39 from the data obtained from the replica exchange simulations.

■

RESULTS 6-Phospho-β-galactoside (β-Gal) from Lactobacillus lactis serves as a useful model enzyme with which to examine the effects of surface attachment on activity and structure. It is a stable, monomeric enzyme for which the crystal structure has been determined and a simple and sensitive assay is available.19,40 Furthermore, the parallel orientation of the α-helices in the protein gives rise to a strong SFG signal for the backbone carbonyl groups, the polarization-dependent signal of which can be used to determine the orientation of the protein with respect to the surface normal.21,41,42 We have previously validated this approach using the β-Gal V152C variant covalently tethered to an EG4-maleimide-functionalized silica surface.18 Our results indicated that the tethered enzyme retains a significant fraction D

DOI: 10.1021/acs.langmuir.5b01735 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir normalized to activity at 25 °C. Under these conditions thermal unfolding of β-Gal is irreversible and the fraction of activity remaining reflects the population of enzymes that remain folded at a given temperature. In solution both β-Gal-V152C and β-Gal-E147C exhibit sharp thermal unfolding curves that are identical within error (Figure 2). Unexpectedly though, the surface-immobilized

Table 2. Comparison of Experimental T1/2 Values and Calculated Tm Values for Thermal Unfolding of β-Gal in Free Solution and Tethered to EG4-Maleimide-Terminated SAM Surface

β-Gal (solution) β-Gal E147C β-Gal V152C

T1/2 (expt, °C)

Tm (calc, °C)

T1/2 − Tm (°C)

ΔΔGfold (calc relative to solution, kJ mol−1)

51 48 38

48 40 34

3 8 4

0.0 4.6 5.9

Orientation of Tethered Enzymes with Respect to Surface. The close proximity of the tethering points engineered into β-Gal-V152C and β-Gal-E147C was intended to orient both enzymes similarly with respect to the surface. We were interested to determine whether attaching the enzyme through a flexible loop allowed the enzyme to access a wider range of orientations with respect to the surface (defined by tilt and twist angles, shown in Figure 3A) than attachment through a more rigid α-helix. Therefore, we experimentally determined the orientation of the tethered β-Gal variants relative to the surface normal by employing a combination of surface-sensitive SFG and ATR-FTIR spectroscopies under different polarization conditions and complemented these measurements with simulations of the surface-tethered proteins.27 SFG and ATR-FTIR spectra were recorded from immobilized β-Gal-E147C as described in the Materials and Methods. SFG spectra were collected under both ssp and ppp polarization combinations (Figure 3B). The spectra are dominated by peaks at ∼1650 and ∼1635 cm−1, which are contributed by the backbone amide stretches for α-helical and β-sheet residues, respectively. After fitting the spectra, a dichroic ratio χppp/χssp = 1.71 was deduced at 1650 cm−1, after correcting for the Fresnel coefficients, χzzz/χxxz = 1.92. Previously we have developed a computer software package to calculate the SFG responses as a function of protein orientation.45 We also developed a method using heat map plots to show all of the possible orientations (with nonzero score values) for which the calculated SFG signal ratios fall within ±10% of the measured χzzz/χxxz ratio.45 These plots display a score assigned based on how close the calculated value of each protein orientation was to the experimental measured value.45 Such a heap map showing the tilt and twist angles that satisfy the experimentally measured χzzz/χxxz value is shown in Figure 3C. Similarly, ATR-FTIR spectra were collected in both s- and p-polarized modes (Figure 3D) and were fitted to determine the dichroic ratio for the α-helical residues, RATR = 1.76 at 1655 cm−1. A method using heat map plot to show the most possible protein orientations according to ATR-FTIR measurements has also been developed.17 The tilt and twist angles that satisfy the experimental RATR value are shown in Figure 3E. The possible orientations that satisfy the constraints deduced from both the SFG and ATR-FTIR spectroscopies, obtained by overlapping the heat maps in Figure 3C,E are shown in Figure 3F,G. Orientations with tilt angles larger than 90° were excluded as physically unrealistic because these orientations would not permit the cysteine residue to form a bond with the SAM surface. The analysis indicates that the possible orientation of β-Gal-E147C is within a range of tilt angles between ∼13 and ∼45° and twist angles between ∼60 and ∼120° with respect to the surface normal. This range of

Figure 2. Comparison of experimental and computationally determined thermal stability for β-gal in solution and tethered to EG4-maleimide SAM surface. (A) Thermal stability curves (fractional activity remaining after heating at given temperature for 10 min) for βGal V152C (blue triangles) and β-Gal E147C (green squares) in free solution. (B) Thermal stability curve for β-Gal E147C tethered to EG4-maleimide SAM surface (green squares). (C) Thermal stability curve for β-Gal V152C tethered to EG4-maleimide SAM surface (blue squares). In each panel, the red line is the computationally determined thermal unfolding curve (fraction nativeness).

enzymes differed significantly in their thermal stabilities. Whereas the β-Gal-E147C variant exhibited only a small decrease in T1/2 from 50.5 ± 1.0 °C in solution to 48 ± 1.0 °C on the surface, for β-Gal-V152C, T1/2 decreased substantially from 50.6 ± 1.0 °C in solution to 38 ± 1.0 °C on the surface and the thermal unfolding curve was significantly broadened, as measured by the slope at T1/2. These data, summarized in Table 2, suggest that interactions of the enzyme with the maleimideterminated SAM surface that depend upon the specific site of covalent attachment may be responsible for the observed changes in thermal stability. E

DOI: 10.1021/acs.langmuir.5b01735 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

Figure 3. Determination of surface orientation of β-Gal-E147C. (A) Definition of twist (ψ) and tilt (θ) angles using an Euler rotation. ϕ is the azimuthal angle, which is averaged out in the orientation determination. (B) SFG spectra of β-Gal-E147C tethered to EG4-maleimide SAM on silica substrate. (C) Heat map showing possible orientation angle regions deduced from spectra in panel B. (D) ATR-FTIR spectra of β-Gal-E147C tethered to EG4-maleimide SAM on silica substrate. (E) Heat map showing possible orientation angle regions deduced from spectra in panel D. (F) Heat map showing possible orientation angle regions consistent both the SFG and ATR-FTIR measurements. Colors indicate the quality of the match. (G) Plots of possible orientation angles with probability ≥90% β-Gal-E147C in blue and comparison with previously determined data26 for βGal-V152C in red. Further details regarding the construction of heat maps from SFG and ATR-FTIR measurements can be found in ref 45.

= 1.67.18 We previously analyzed the orientation of the β-GalV152C variant18 and showed that the deduced range of tilt and twist angles relative to the surface normal is consistent with its attachment through the loop position 152. As expected, the orientation angles are similar for the two constructs; however,

orientational motion is physically reasonable based on the covalent tethering of the protein through Cys-147. We compared the possible range of orientations deduced for β-Gal-E147C with those deduced for β-Gal-V152C18 (Figure 3G). We reported that the measured χzzz/χxxz = 1.91 and RATR F

DOI: 10.1021/acs.langmuir.5b01735 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

Figure 4. Coarse-grain simulations of thermal unfolding of β-Gal. Representative structures are shown after 3 ns of simulation at three different temperatures corresponding to the Tm of each enzyme form; warm colors represent α-helices, cool colors represent β-sheets and loop regions. (A) βGal in free solution. (B) β-Gal tethered to the surface through position 147 (helix). (C) β-Gal tethered to the surface through position 152 (loop). The structure corresponding the Tm of each enzyme form is indicated. The attachment point for each surface-tethered enzyme is indicated by an arrow.

the possible combinations of twist and tilt angles for β-GalV152C span a wider range than those for β-Gal-E147C. In addition, there is a small possible orientation area with large tilt angles for β-Gal-V152C (Figure 3G). This difference is consistent with residue 152 residing on a flexible loop that allows a great freedom of orientation, whereas the more rigid, helical location may place more constraints on the orientation of the enzyme. This also suggests that the β-Gal-V152Ctethered enzyme is more likely to come into contact with the surface through a large exclusion in tilt angle than is the enzyme tethered through residue 147. As we discuss later, such protein−surface interactions are likely to be destabilizing. Simulation of Protein Folding in Free Solution and on Surfaces. To obtain insight into the differences in thermal stabilities and orientation of the β-Gal variants we undertook coarse-grain replica exchange molecular dynamics (REMD) simulations of these enzymes in bulk solution and when tethered to the maleimide-terminated SAM surface. These simulations used a recently developed coarse-grain model of protein−surface interactions30 based on a well-established Golike protein model.26,27 Figure 2 shows the thermal folding curves for β-Gal in free solution and tethered to the EG4maleimide SAM surface at either position 147 or position 152 as determined from our computations. We modeled the progress in folding/unfolding by monitoring the average fraction of native interactions made at a given temperature relative to interactions present in the natively golden protein. The melting temperature, Tm, is defined as the temperature at which, on average, 50% of the total native contacts are formed. The experimentally determined T1/2 is compared with the computationally determined Tm for each enzyme form in Table 2. From Figure 2 and Table 2 it can be seen that the simulations model the thermal folding/inactivation curves for β-Gal quite closely. The simulations correctly predict the order of enzyme stabilities, that is, that β-Gal in free solution has the

highest Tm, that tethering at position 147 lowers Tm somewhat, and that tethering at position 152 decreases Tm more significantly. The simulations further indicate that the differences in T1/2 observed when β-Gal is tethered to the surface arise from differential interactions of the enzyme with the maleimideterminated SAM surface. Movies of the trajectories for the enzyme at different temperatures in free solution and tethered to surfaces at positions 147 and 152 (which are available as Supporting Information) suggest that tethering on a flexible loop (position 152) allows the protein to adopt a wider range of orientations than does tethering through the less flexible αhelix (position 147), which is in accord with the wider range of possible orientations for β-Gal-V152C determined spectroscopically. This provides more opportunities to make hydrophobic interactions with the surface that are destabilizing to structure. The temperature-dependent simulations allowed the difference in the free energy of folding, ΔΔGfold, to be calculated at the melting temperature of the enzyme in free solution (Table 2). The surface was found to destabilize the folded state by 5 to 6 kJ mol−1 under the parameters used for the simulations. In principle, analysis of these “sticky patches” between the tethered protein and the surface could be used to re-engineer the protein surface to reduce this affect, while maintaining the flexible nature of the tethering and thereby enhance the enzyme’s stability. Structures for β-Gal in free solution, immobilized β-GalV152C, and immobilized β-Gal-E147C calculated after 3.0 ns of simulation are shown in Figure 4 at three temperatures, 48, 40, and 34 °C, corresponding to the calculated Tm values for the three enzymes. As expected, at temperatures below Tm, the structures remain largely folded, with the exception of the active site loop, which is more labile. At Tm and above, the structures become extensively unfolded, although some secondary and tertiary structure remains. For the surface-tethered enzyme the influence of the surface in stabilizing the unfolded structure is G

DOI: 10.1021/acs.langmuir.5b01735 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

identify those that are likely to result in biomaterials with high thermal stability appears to be a promising avenue for optimizing surface-supported enzyme catalysts.

evident; the protein chain extends across the surface with many residues making interactions with the surface.

■

DISCUSSION The immobilization of proteins on abiotic surfaces is important in a wide range of applications including medical implants, drug delivery, sensors and diagnostic testing, bioseparation technologies, and bioreactors.1,3,7,46 Whereas it is well-established that the interaction of proteins with surfaces has a profound influence on their structure and activity,8−16 we currently lack the detailed understanding of these interactions necessary to engineer immobilized proteins for optimal performance. To dissect this complex problem, we have used chemically welldefined SAM as surfaces and engineered enzymes to allow their attachment at precisely defined positions on the protein surface. This results in a population of surface-immobilized enzymes that are sufficiently uniform to permit detailed characterization by spectroscopic and computational methods. We initially chose positions 147 and 152 in β-Gal as surface attachment points to explore the effect of secondary structural context−rigid helix versus flexible loop-on enzyme activity. We reasoned that the choice of attachment point might affect largescale, low-frequency vibrational modes that have been shown to be important for catalysis in a number of enzymes.43,44 In practice, the immobilized enzymes were found to possess very similar specific activities, suggesting that the precise position of the tethering point is not critical for activity. Both of the tethered enzymes adopt a similar orientation with respect to the surface, consistent with the position of the tethering points; however, the β-Gal-V152C construct appears to traverse a greater range of orientations, as determined by analysis of their SFG and ATR-FTIR vibrational spectra (Figure 3). Given the similarity of the two surface tethered enzymes, we were surprised to find a marked difference in their thermal stabilities; however, valuable insights into the origin of these differences have been provided by coarse-grain simulations that replicate with reasonable accuracy the experimentally determined thermal stability curves (Figure 2). Attachment through a flexible loop appears to provide more opportunity for hydrophobic residues, transiently exposed during localized unfolding of the protein, to form favorable interactions with the surface. Moreover, this interpretation is supported by spectroscopic analyses that indicate the loop-tethered enzyme samples a wider range of tilt angles than the enzyme attached through the adjacent helix. Although the SAM surfaces employed in these studies are less complex than the solid supports, for example, polystyrene beads, commonly used in biotechnological applications, we believe these results do have significant implications for the rational design of solid-phase supported biocatalysts. First, they demonstrate the importance of the choice of attachment point on the stability of the immobilized enzyme. Furthermore, they imply that the nonspecific covalent cross-linking and noncovalent physic-adsorption methods commonly used to immobilize enzymes likely result in biomaterials that are suboptimal with respect to their activity and thermal stability. Our results suggest that by better understanding the interactions between enzyme and surface it should be possible to design catalysts with enhanced thermal stability and improved activity and potentially expand the range of enzymes that can be used in industrial processes. In particular, the use of computational methods to systematically screen multiple attachment sites on an enzyme of interest in silico and thereby

■

ASSOCIATED CONTENT

■

AUTHOR INFORMATION

S Supporting Information *

Simulated thermal unfolding of β-galactoside in solution and tethered to surfaces: movies (mpeg files) derived from coarsegrain molecular dynamics simulations showing the unfolding of β-Gal in solution and attached to a surface through position 147 and through position 152. The unfolding trajectories are presented at three simulation temperatures: 48, 40, and 34 °C, corresponding to the calculated Tm values for the three enzymes. The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/ acs.langmuir.5b01735. Corresponding Author

*Phone: 734 763 6096. E-mail [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work is supported by the Defense Threat Reduction Agency (HDTRA1-11-1-0019) and the Army Research Office (W911NF-11-1-0251).

■

REFERENCES

(1) Ansari, S. A.; Husain, Q. Potential Applications of Enzymes Immobilized on/in Nano Materials: A Review. Biotechnol. Adv. 2011, 30, 512−523. (2) Brady, D.; Jordaan, J. Advances in Enzyme Immobilisation. Biotechnol. Lett. 2009, 31, 1639−1650. (3) Frasconi, M.; Mazzei, F.; Ferri, T. Protein Immobilization at Gold-Thiol Surfaces and Potential for Biosensing. Anal. Bioanal. Chem. 2010, 398, 1545−1564. (4) Lee, S. Y.; Lee, J.; Chang, J. H.; Lee, J. H. Inorganic Nanomaterial-Based Biocatalysts. BMB Rep. 2011, 44, 77−86. (5) Sassolas, A.; Blum, L. J.; Leca-Bouvier, B. D. Immobilization Strategies to Develop Enzymatic Biosensors. Biotechnol. Adv. 2012, 30, 489−511. (6) Houseman, B. T.; Gawalt, E. S.; Mrksich, M. MaleimideFunctionalized Self-Assembled Monolayers for the Preparation of Peptide and Carbohydrate Biochips. Langmuir 2003, 19, 1522−1531. (7) Jonkheijm, P.; Weinrich, D.; Schroder, H.; Niemeyer, C. M.; Waldmann, H. Chemical Strategies for Generating Protein Biochips. Angew. Chem. Int. Ed. 2008, 47, 9618−9647. (8) Swadling, J. B.; Wright, D. W.; Suter, J. L.; Coveney, P. V. Structure, Dynamics, and Function of the Hammerhead Ribozyme in Bulk Water and at a Clay Mineral Surface from Replica Exchange Molecular Dynamics. Langmuir 2015, 31, 2493−2501. (9) Nishida, H.; Kajisa, T.; Miyazawa, Y.; Tabuse, Y.; Yoda, T.; Takeyama, H.; Kambara, H.; Sakata, T. Self-Oriented Immobilization of DNA Polymerase Tagged by Titanium-Binding Peptide Motif. Langmuir 2015, 31, 732−740. (10) Grimaldi, J.; Radhakrishna, M.; Kumar, S. K.; Belfort, G. Stability of Proteins on Hydrophilic Surfaces. Langmuir 2015, 31, 1005−1010. (11) Shrivastava, S.; McCallum, S. A.; Nuffer, J. H.; Qian, X.; Siegel, R. W.; Dordick, J. S. Identifying Specific Protein Residues That Guide Surface Interactions and Orientation on Silica Nanoparticles. Langmuir 2013, 29, 10841−10849. (12) Thyparambil, A. A.; Wei, Y.; Latour, R. A. Determination of Peptide−Surface Adsorption Free Energy for Material Surfaces Not Conducive to Spr or Qcm Using Afm. Langmuir 2012, 28, 5687−5694. H

DOI: 10.1021/acs.langmuir.5b01735 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

(35) Hansmann, U. H. E.; Okamoto, Y. New Monte Carlo Algorithms for Protein Folding. Curr. Opin. Struct. Biol. 1999, 9, 177−183. (36) Hoover, W. G. Canonical Dynamics - Equilibrium Phase-Space Distributions. Phys. Rev. A 1985, 31, 1695−1697. (37) Nose, S. A Unified Formulation of the Constant Temperature Molecular-Dynamics Methods. J. Chem. Phys. 1984, 81, 511−519. (38) Nose, S. Constant-Temperature Molecular-Dynamics. J. Phys., Condens. Matter 1990, 2, SA115−SA119. (39) Kumar, S.; Bouzida, D.; Swendsen, R. H.; Kollman, P. A.; Rosenberg, J. M. The Weighted Histogram Analysis Method for FreeEnergy Calculations on Biomolecules 0.1. The Method. J. Comput.l Chem. 1992, 13, 1011−1021. (40) Schulte, D.; Hengstenberg, W. Engineering the Active Center of the 6-Phospho-Beta-Galactosidase from Lactococcus Lactis. Protein Eng. 2000, 13, 515−518. (41) Wang, J.; Buck, S. M.; Even, M. A.; Chen, Z. Molecular Responses of Proteins at Different Interfacial Environments Detected by Sum Frequency Generation Vibrational Spectroscopy. J. Am. Chem. Soc. 2002, 124, 13302−13305. (42) Chen, X.; Boughton, A. P.; Tesmer, J. J.; Chen, Z. In Situ Investigation of Heterotrimeric G Protein Betagamma Subunit Binding and Orientation on Membrane Bilayers. J. Am. Chem. Soc. 2007, 129, 12658−12659. (43) Knapp, M. J.; Klinman, J. P. Environmentally Coupled Hydrogen Tunneling - Linking Catalysis to Dynamics. Eur. J. Biochem. 2002, 269, 3113−3121. (44) Schramm, V. L. Enzymatic Transition States, Transition-State Analogs, Dynamics, Thermodynamics, and Lifetimes. Annu. Rev. Biochem. 2011, 80, 703−732. (45) Boughton, A. P.; Yang, P.; Tesmer, V. M.; Ding, B.; Tesmer, J. J.; Chen, Z. Heterotrimeric G Protein Beta1gamma2 Subunits Change Orientation Upon Complex Formation with G Protein-Coupled Receptor Kinase 2 (Grk2) on a Model Membrane. Proc. Natl. Acad. Sci. U.S.A. 2011, 108, E667−673. (46) Jahr, J. S.; Akha, A. S.; Holtby, R. J. Crosslinked, Polymerized, and Peg-Conjugated Hemoglobin-Based Oxygen Carriers: Clinical Safety and Efficacy of Recent and Current Products. Curr. Drug Discovery Technol. 2011, 9, 158−165.

(13) Wei, Y.; Latour, R. A. Correlation between Desorption Force Measured by Atomic Force Microscopy and Adsorption Free Energy Measured by Surface Plasmon Resonance Spectroscopy for PeptideSurface Interactions. Langmuir 2010, 26, 18852−18861. (14) Sala, A.; Ehrbar, M.; Trentin, D.; Schoenmakers, R. G.; Vörös, J.; Weber, F. E. Enzyme Mediated Site-Specific Surface Modification. Langmuir 2010, 26, 11127−11134. (15) Qian, Z.; Khan, M. A.; Mikkelsen, S.; Chen, P. Improved Enzyme Immobilization on an Ionic-Complementary PeptideModified Electrode for Biomolecular Sensing. Langmuir 2010, 26, 2176−2180. (16) Wei, Y.; Latour, R. A. Benchmark Experimental Data Set and Assessment of Adsorption Free Energy for Peptide-Surface Interactions. Langmuir 2009, 25, 5637−5646. (17) Shen, L.; Schroeder, M.; Ogorzalek, T. L.; Yang, P.; Wu, F.-G.; Marsh, E. N. G.; Chen, Z. Surface Orientation Control of SiteSpecifically Immobilized Nitro-Reductase. Langmuir 2014, 30, 5930− 5938. (18) Liu, Y.; Ogorzalek, T. L.; Yang, P.; Schroeder, M. M.; Marsh, E. N.; Chen, Z. Molecular Orientation of Enzymes Attached to Surfaces through Defined Chemical Linkages at the Solid-Liquid Interface. J. Am. Chem. Soc. 2013, 135, 12660−12669. (19) Wiesmann, C.; Hengstenberg, W.; Schulz, G. E. Crystal Structures and Mechanism of 6-Phospho-Beta-Galactosidase from Lactococcus Lactis. J. Mol. Biol. 1997, 269, 851−860. (20) Husain, Q. Beta Galactosidases and Their Potential Applications: A Review. Crit. Rev. Biotechnol. 2010, 30, 41−62. (21) Wang, J.; Paszti, Z.; Even, M. A.; Chen, Z. Interpretation of Sum Frequency Generation Vibrational Spectra of Interfacial Proteins by the Thin Film Model. J. Phys. Chem. B 2004, 108, 3625−3632. (22) Xie, Y.; An, J.; Yang, G.; Wu, G.; Zhang, Y.; Cui, L.; Feng, Y. Enhanced Enzyme Kinetic Stability by Increasing Rigidity within the Active Site. J. Biol. Chem. 2014, 289, 7994−8006. (23) Duy, C.; Fitter, J. Thermostability of Irreversible Unfolding Alpha-Amylases Analyzed by Unfolding Kinetics. J. Biol. Chem. 2005, 280, 37360−37365. (24) Chen, X. Y.; Clarke, M. L.; Wang, J.; Chen, Z. Sum Frequency Generation Vibrational Spectroscopy Studies on Molecular Conformation and Orientation of Biological Molecules at Interfaces. Int. J. Mod. Phys. B 2005, 19, 691−713. (25) Yang, P.; Boughton, A.; Homan, K. T.; Tesmer, J. J.; Chen, Z. Membrane Orientation of Ga1b1g2 and Gb1g2 Determined Via Combined Vibrational Spectroscopic Studies. J. Am. Chem. Soc. 2013, 135, 5044−5051. (26) Karanicolas, J.; Brooks, C. L. The Structural Basis for Biphasic Kinetics in the Folding of the Ww Domain from a Formin-Binding Protein: Lessons for Protein Design? Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 3954−3959. (27) Karanicolas, J.; Brooks, C. L. Integrating Folding Kinetics and Protein Function: Biphasic Kinetics and Dual Binding Specificity in a WW Domain. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 3432−3437. (28) Hills, R. D.; Brooks, C. L. Insights from Coarse-Grained Go Models for Protein Folding and Dynamics. Int. J. Mol. Sci. 2009, 10, 889−905. (29) Schmitt, T. J.; Clark, J. E.; Knotts, T. A. Thermal and Mechanical Multistate Folding of Ribonuclease H. J. Chem. Phys. 2009, 131. (30) Wei, S.; Knotts, T. A. A Coarse Grain Model for Protein-Surface Interactions. J. Chem. Phys. 2013, 139. (31) 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, 9359−9366. (32) Swendsen, R. H.; Wang, J.-S. Replica Monte Carlo Simulation of Spin-Glasses. Phys. Rev. Lett. 1986, 57, 2607−2609. (33) Hukushima, K. Exchange Monte Carlo Method and Application to Spin Glass Simulations. J. Phys. Soc. Jpn. 1996, 65, 1604−1608. (34) Sugita, Y.; Okamoto, Y. Replica-Exchange Molecular Dynamics Method for Protein Folding. Chem. Phys. Lett. 1999, 314, 141−151. I

DOI: 10.1021/acs.langmuir.5b01735 Langmuir XXXX, XXX, XXX−XXX