Effect of Ionic Liquids on the Solution Structure of Human Serum

Feb 18, 2011 - HSA, the most abundant protein in human blood, is able to bind and transport multiple fatty acids (FAs). ... Citation data is made avai...
0 downloads 0 Views 3MB Size
Download PDF
ARTICLE pubs.acs.org/Biomac

Effect of Ionic Liquids on the Solution Structure of Human Serum Albumin Yasar Akdogan, Matthias J. N. Junk,§ and Dariush Hinderberger* Max Planck Institute for Polymer Research, Ackermannweg 10, 55128 Mainz, Germany

bS Supporting Information ABSTRACT: The effect of several ionic liquids (ILs) on the solution structure of human serum albumin (HSA) is revealed by continuous wave electron paramagnetic resonance (EPR) spectroscopy and nanoscale distance measurements with double electron-electron resonance (DEER) spectroscopy. HSA, the most abundant protein in human blood, is able to bind and transport multiple fatty acids (FAs). Using spinlabeled FA, the uptake of the FA by the protein and their spatial distribution in the protein can be monitored. The FA distribution provides an indirect yet effective way to characterize the structure of the protein in solution. Addition of imidazolium-based ILs to an aqueous solution of HSA/FA conjugates is accompanied by significant destabilization and unfolding of the protein’s tertiary structure. In contrast, HSA maintains its tertiary structure when choline dihydrogenphosphate (dhp) is added. The comparison of FA distance distributions in HSA with and without choline dhp surprisingly revealed that with this IL, the FA anchoring units are in better agreement with the crystallographic data. Furthermore, the FA entry point distribution appears widened and more asymmetric than in pure buffer. These results indicate that choline dhp as a cosolvent may selectively stabilize HSA conformations closer to the crystal structure out of the overall conformational ensemble.

’ INTRODUCTION Ionic liquids (ILs) have attracted extensive attention during recent years because of their potential in various biological and pharmaceutical applications. Specifically, their ability to solubilize and stabilize proteins in vitro for extended times was extensively studied.1-6 The unique features of ILs, including high thermal and chemical stability, negligible vapor pressure, low degree of flammability, and low toxicity, have made them alternative solvents to traditional organic solvents. ILs are often referred to as “designer solvents” because of the fact that their physical properties such as polarity, viscosity, miscibility, and density can be finely tuned to fulfill specific demands by the selection of suitable anions and cations.7-9 Although proteins are, in general, unstable and inactive in many ILs, selected biocompatible ILs were shown to stabilize some proteins. Fujita et al. reported a new family of biocompatible ILs based on dihydrogen phosphate anions that are able to maintain the protein structure and activity of cytochrome c even at higher temperatures.4 In other studies, the protein monellin was shown to be thermally stabilized in 1-butyl-1-methylpyrrolidinium bis(trifluoromethane sulfonyl) imide up to temperatures as high as 105 °C compared with only 40 °C in bulk water.3 Furthermore, a long-term stabilization against aggregation and hydrolysis was reported for lysozyme in ethylammonium nitrate.10 Imidazolium-based systems, which are among the most commonly used ILs, lead to a significant destabilization of the lysozyme tertiary structure. This reduction in stability increases with the hydrophobicity of the imidazolium cations.11 The influence of solvents on the structural changes of several proteins have been analyzed by circular dichroism spectroscopy, r 2011 American Chemical Society

differential scanning calorimetry, fluorescence spectroscopy, Fourier transform infrared spectroscopy, UV-vis spectroscopy, and small-angle neutron scattering.4-6,10,12-17 In this Article, we introduce electron paramagnetic resonance (EPR) spectroscopy as an alternative technique to characterize the effects of ILs on the functional structure of human serum albumin (HSA). EPR spectroscopy in combination with site-directed spin labeling has been increasingly used during the last decades to reveal the molecular and global structure of proteins and their local dynamics.18-22 Labeling of the protein or a substrate in a protein-substrate complex with radicals causes only minimal structural perturbation and allows site-specific characterization of the biomacromolecules. Solvent accessibility and label mobility can be obtained by continuous wave (CW) EPR to determine the local structure of protein within distances up to 2 nm. Pulse EPR techniques, such as double electron-electron resonance (DEER) spectroscopy can access distances up to 6 nm (8 nm under completely optimized conditions) between the unpaired electron spins. In recent years, DEER has been increasingly used to study the global structure of biomacromolecules such as (membrane) proteins, peptides, and nucleic acids.23-27 HSA is the most abundant protein in human blood plasma and serves as a carrier of fatty acids (FAs) and a diverse range of metabolites from the bloodstream to target cells.28,29 In addition, it helps to maintain the osmotic blood pressure and contributes Received: November 25, 2010 Revised: January 18, 2011 Published: February 18, 2011 1072

dx.doi.org/10.1021/bm1014156 | Biomacromolecules 2011, 12, 1072–1079

Biomacromolecules

ARTICLE

Figure 1. (A) Crystal structure (PDB 1e7i) of HSA cocrystallized with seven fatty acid molecules (ref 31). The oxygen atoms of the FA carboxylic acid head groups are displayed in red. (B) Chemical structures of the EPR-active FAs, 16-doxylstearic acid (16-DSA). and 5-doxylstearic acid (5-DSA). (C) Chemical structures of the ILs choline dihydrogenphosphate (dhp) and the 1-alkyl-3-methylimidazolium tetrafluoroborate derivatives, EmimBF4, BmimBF4, and HmimBF4.

to bodily detoxification by binding poisonous materials. The rich binding properties of HSA, especially for FAs, make it one of the most intensely studied proteins. Crystallographic analyses revealed seven distinct binding sites for long-chain FAs,30,31 and NMR spectroscopy provided complementary information on their relative affinities for FAs.32,33 Three binding sites (2, 4, and 5) were reported to exhibit high affinity for FAs, and four binding sites (1, 3, 6, and 7) were found to exhibit a somewhat lower affinity (Figure 1A). Most binding sites are composed of positively charged anchoring units, which strongly interact with the carboxylic acid headgroup of the FAs, and long, hydrophobic pockets surrounding the methylene tail. Although X-ray crystallography and NMR spectroscopy are commonly used to determine protein structures, both methods have some limitations. X-ray crystallography can be applied only to protein crystals with sufficient long-ranged order. However, the crystallization conditions can generate biologically irrelevant conformations due to, for example, lack of flexibility.34 NMR spectroscopy is restricted to small- and medium-sized proteins with molecular weights of 95%, Calbiochem), spin-labeled FAs 5- and 16-DSA (Aldrich), the ILs 1-butyl-3-methylimidazolium tetrafluoroborate (BmimBF4, 98.5%, Sigma-Aldrich), 1-ethyl-3-methylimidazolium tetrafluoroborate (EmimBF4, 99%, Sigma-Aldrich), 1-hexyl-3-methylimidazolium tetrafluoroborate (HmimBF4, 97%, SigmaAldrich), and choline dhp (Roth), and 87 wt % glycerol (Fluka) were used as received. The spin-labeled FAs were partially reduced to EPRinactive hydroxylamines (rDSA) by the addition of phenylhydrazine (97%, Sigma-Aldrich).36 Sample Preparation. Aqueous solutions of 2 mM HSA in 0.11 M phosphate buffer (pH 7.2) and 26 mM DSA and rDSA in 0.1 M KOH were prepared. The combined concentration of DSA and rDSA in final buffered solutions of pH 7.4 was kept constant at 1.5 mM. The molar ratios of DSA and rDSA per protein molecule were varied from 2/0 to 2/4. Different amounts of ILs were added to these aqueous FA/HSA solutions. EmimBF4 and BmimBF4 are water-miscible, and HmimBF4 can be dissolved up to a 1073

dx.doi.org/10.1021/bm1014156 |Biomacromolecules 2011, 12, 1072–1079

Biomacromolecules concentration of ca. 10% (v/v). Choline dhp is water-soluble, but precipitation occurred above a concentration of ca. 25% (v/v) of choline dhp in HSA/buffer solution. Note that for DEER measurements, which are conducted at low temperatures, 15% (v/v) glycerol was added to the aqueous solutions of HSA and FA without any ILs to prevent crystallization upon freezing. This cryoprotectant is known to stabilize the native structure of proteins in frozen aqueous solutions.37 No changes in the CW EPR spectra were observed upon the addition of glycerol. About 100 μL of the final solutions was filled in 3 mm (outer diameter) quartz tubes and shockfrozen in liquid-nitrogen-cooled iso-pentane. EPR Measurements. A Miniscope MS200 (Magnettech, Berlin, Germany) benchtop spectrometer was used for X-band CW EPR measurements at a microwave frequency of ∼9.4 GHz. Measurements were performed at room temperature (293 K) using a modulation amplitude of 0.05 mT. The microwave frequency was recorded with a frequency counter, model 2101 (Racal-Dana). The four pulse DEER sequence π/2(νobs) - τ1 - π(νobs) - (τ1 þ t) (νpump) - (τ2 - t) - π(νobs) - τ2 - echo was used to obtain dipolar time evolution data at X-band frequencies (9.2 to 9.4 GHz) with a Bruker Elexsys 580 spectrometer equipped with a Bruker Flexline split-ring resonator ER4118X_MS3.38,39 The dipolar evolution time t was varied, whereas τ2 = 2.5 μs and τ1 were kept constant. Proton modulation was averaged by the addition of eight time traces of variable τ1, starting with τ1,0 = 200 ns and incrementing by Δτ1 = 8 ns.40 The resonator was overcoupled to Q ≈ 100. The pump frequency, νpump, was set to the maximum of the EPR spectrum. The observer frequency, νobs, was set to νpump þ 61.6 MHz, coinciding with the low field local maximum of the nitroxide spectrum. The observer pulse lengths were 32 ns for both π/2 and π pulses, and the pump pulse length was 12 ns. The temperature was set to 50 K by cooling with a closed cycle cryostat (ARS AF204, customized for pulse EPR, ARS, Macungie, PA). The total measurement time for each sample was around 12 h. The raw time domain DEER data were processed with the program package DeerAnalysis2008.41 Intermolecular contributions were removed by division by an exponential decay with a fractal dimension of d = 3.8. As shown in a previous study, the deviation from d = 3.0 originates from excluded volume effects due to the size of the protein.36 The resulting time traces were normalized to t = 0. Distance distributions were obtained by Tikhonov regularization using a regularization parameter of 1000.

’ RESULTS AND DISCUSSION Effects of Ionic Liquids on the Uptake and Release of Fatty Acids by HSA. Conventional CW EPR spectroscopy on nitr-

oxide spin probes contains information about the mobility of the radical and the polarity of its local environment.42,43 In this study, the CW EPR spectral line shapes of the spin-labeled FAs sensitively depend on their rotational mobility. This mobility is largely restricted when the FAs are placed in tight binding channels of the protein. Hence, CW EPR spectroscopy allows unambiguously retrieving whether FAs are bound to HSA or not. CW EPR spectra of the spin-labeled FA 5-DSA in different HSA solutions are displayed in Figure 2. The FA/HSA ratio was kept constant at 2:1. In a typical room-temperature EPR spectrum of 5-DSA, immobilized FAs (bound) can easily be assigned because of their broad outer hyperfine features stemming from restricted rotational motion. The spectra display sharp three-line signals, signatures of freely tumbling DSA (marked in Figure 2). Figure 2B-E displays the spectra obtained after the addition of the cosolvents ethanol (black) or BmimBF4 (red) at different volume ratios. At concentrations higher than 15% (v/v), the EPR signal intensity of free FAs steadily increases and reaches a maximum at ∼50% (v/v) of ethanol or BmimBF4. At 50% (v/v) concentration, almost all bound FAs are released from HSA and can tumble freely in solution.

ARTICLE

Figure 2. CW EPR spectra of 5-DSA in HSA/buffer solution at a FA/ HSA ratio of 2 upon addition of ethanol (black) and BmimBF4 (red) at different concentrations: (A) 0% (v/v), (B) 15% (v/v), (C) 25% (v/v), (D) 35% (v/v), (E) 50% (v/v), and (F) redilution of sample E with buffer to 10% (v/v). The characteristic signals of bound 5-DSA to albumin are marked by solid lines. The three lines characteristic for freely tumbling 5-DSA are marked by dashed lines.

Ethanol serves as a reference cosolvent because it is wellestablished that the interaction of ethanol with HSA results in the denaturation of the protein.44-46 The secondary and tertiary structure of the protein is partially lost at ethanol concentrations below 30% (v/v), and the protein is completely denatured at 50% (v/v) of ethanol.46 Furthermore, 13C NMR spectroscopy demonstrated ethanol-induced changes of the activity of bovine serum albumin as a consequence of direct binding of ethanol to specific hydrophobic binding sites, for example, the FA binding sites.47 Thus, ethanol competitively binds to the hydrophobic channels, which are essential for the complexation of FAs. These findings are consistent with our CW EPR-derived results. The addition of ethanol leads to a decrease in the fraction of bound FAs. To a certain degree, this is a reversible reaction because a considerable fraction of FAs binds to the protein again after the ethanol-containing solution is rediluted with buffer (Figure 2F). Yet, this solution with 10% (v/v) ethanol contains a distinctly higher fraction of free FAs than the HSA/FA mixture with 15% (v/v) ethanol. This could suggest a partial, irreversible denaturation of the protein by ethanol. Because many FA binding sites are placed at the interfaces of the three main subdomains of HSA,31 one might assume that the tertiary structure is partly lost, that is, that certain linkages between the subdomains are irreversibly destroyed. The protein segments that are connected through H-bonds and electrostatic or hydrophobic interactions may lose their connection if the interaction to the cosolvent is favored (e.g., for entropic reasons). Also, we cannot fully exclude that denatured protein upon ethanol or BmimBF4 addition at a high amount (50% (v/v)) could be aggregated/precipitated to a small degree and that this may also contribute the partial irreversibility. It should also be mentioned that ethanol is a better solvent for FAs than pure buffer and that we cannot exclude a small effect on the uptake by HSA when more ethanol is present. Yet, the binding constants to HSA are so high that the uptake by the protein certainly dominates over potential differences in the solution properties. 1074

dx.doi.org/10.1021/bm1014156 |Biomacromolecules 2011, 12, 1072–1079

Biomacromolecules The binding properties of FAs to HSA with the imidazoliumbased IL BmimBF4 as a cosolvent are almost identical to those in ethanol/buffer mixtures. This suggests that similar to ethanol, BmimBF4 competitively occupies the FA binding sites, partially denatures the protein, or both. To gain more detailed insight into the binding mechanism of imidazolium-based ILs to HSA, we used cations with different alkyl group lengths while keeping [BF4]- as anion and the concentration at 10% (v/v) (Figure 3). As the length of the alkyl chain that is attached to the imidazolium ring increases from ethyl to hexyl, the spectral contribution of freely tumbling FA increases tremendously. The addition of the hexyl-substituted HmimBF4 results in a significantly larger spectral contribution of free FAs, as compared with that found with EmimBF4 or BmimBF4 as cosolvent. This indicates that the imidazolium-based cations and, in particular, their alkyl chains interact with the protein. Considering the competitive uptake of solvent into hydrophobic channels, it is not surprising that the hexyl component is by far most effective in releasing FAs. (See Figure 3.) An increasing length of the IL’s alkyl side chains not only leads to an increased fraction of free FA but also causes changes in the spectral features of the bound FA. The distance of the outer extrema in the spectra (ΔB, marked by solid lines in Figure 3) is very sensitive to the rotational motion of slowly tumbling spin labels, with a larger distance corresponding to a more restricted rotational motion.48 The addition of EmimBF4, BmimBF4, and HmimBF4 leads to a decreasing separation of the outer extrema and hence indicates an increase in the rotational mobility of the bound FA in the protein. Such a gain in mobility is likely due to a higher mobility of the secondary structure elements that interact with FAs (i.e., their binding pockets), which in turn likely stems from changes in the local conformation of these (in the case HSA) self-assembled helices. This implies a more open protein structure with structurally (and dynamically) less restricted segments. Hence, this is a further hint for the partial destruction of those links that are responsible for the formation of a compact and welldefined tertiary protein structure. The influence of hydrophilic ILs on the protein activity and stability usually follows the Hofmeister series (or the kosmotropicity order) when ILs dissociate into individual ions in water.6,17,49 Kosmotropic anions and chaotropic (i.e., nonkosmotropic) cations stabilize protein structures by promoting the water structure around them, whereas kosmotropic cations destabilize protein structures in

Figure 3. CW EPR spectra of 5-DSA in HSA/buffer solutions at a FA/ HSA ratio of 2 without (black) and with the addition of various imidazolium-based ionic liquids at a concentration of 10% (v/v): EmimBF4 (green), BmimBF4 (blue), and HmimBF4 (red). The outer extrema signals are marked by gray lines, and their spectral separation ΔB is indicated by two-headed arrows. The three lines characteristic for freely tumbling 5-DSA are marked by dashed lines.

ARTICLE

solution.6 The kosmotropicity of the used imidazolium-based cations increases in parallel to their hydrophobicity, that is, [Emim]þ < [Bmim]þ < [Hmim]þ. In addition, the anion used in these ILs is tetrafluoroborate [BF4]-, which is a weak kosmotropic anion. Because of the combined ion effects, all three imidazolium-based cosolvents are not tertiary structure-promoting ILs for proteins.15 Our CW EPR results show that BmimBF4 changes the activity of protein like ethanol does (Figures 2 and 3), and hence this IL is used in the following for comparison with structure-promoting ILs. According to the kosmotropicity concept, ILs such as choline dhp should potentially be biocompatible.4,6 The strong chaotropic choline cation has a quaternary ammonium cation that is also commonly found in biological systems. The phosphate anion (found also in the buffer solution), in contrast, has a strong kosmotropicity. Figure 4 illustrates the binding of FA to HSA in buffer solution (A) and upon addition of BmimBF4 (B), ethanol (C), or choline dhp (D) at 25% (v/v). Whereas sharp signals from unbound FAs are observed upon the addition of ethanol or BmimBF4, the amount of bound FAs remains unchanged when choline dhp is added to the buffer solution. Hence, all FAs remain bound to the protein. This suggests that choline dhp does not competitively bind to the hydrophobic binding sites and thus does not replace the FA located therein. One might also assume that the tertiary structure of the protein is preserved when this IL is added to the aqueous solution. The small changes in the spectral line shape may be due to local effects of choline dhp on the FA anchoring group, but analysis thereof is beyond the scope of this study. Distance Distributions of FA Binding Sites in HSA/buffer/ ILs. As shown in the last section, CW EPR spectroscopy is a useful tool to probe sensitively the binding of FAs to HSA in different solvent mixtures. However, CW EPR does not provide quantitative information about the conformational changes of HSA that are induced by the addition of ethanol and the ILs. DEER spectroscopy, a pulse EPR technique, allows measuring of dipolar couplings between unpaired electron spins and provides a means to access distances between paramagnetic moieties in the sample. When spin-labeled FAs are added to the protein, the spatial distribution of the binding sites and hence the structure of the protein in solution can be probed. For spectroscopic reasons, to avoid multispin interactions that may lead to wrong modulation depths and distance peaks, only two EPR-active FAs are used per HSA.36,50 A homogeneous distribution of these two EPR-active FA

Figure 4. CW EPR spectra of 5-DSA and HSA (2/1) in a purely aqueous buffered medium (A) and after addition of 25% (v/v) BmimBF4 (B), ethanol (C), or choline dhp (D). The characteristic signals of 5-DSA bound to HSA are marked by solid lines. The three lines characteristic for freely tumbling 5-DSA are marked by dashed lines. 1075

dx.doi.org/10.1021/bm1014156 |Biomacromolecules 2011, 12, 1072–1079

Biomacromolecules

Figure 5. Intramolecular part of the DEER time-domain data of the spin-labeled FAs in HSA/buffer (black), in HSA/buffer/15% (v/v) BmimBF4 (blue) and in HSA/buffer/20% (v/v) choline dhp (red) solutions. The 5-DSA data are shown in (A), and the 16-DSA data are shown in (B). On average, each protein molecule is loaded with two FA molecules. The modulation depths Δ are marked.

can be achieved by an additional loading of the protein with diamagnetic FAs. Yet, as shown recently, the distance distributions obtained by DEER do not undergo considerable changes when the loading of the protein is varied.36,50 This is also true for the present study. No significant changes of distance distribution were observed when the protein was loaded with additional diamagnetic FAs (Figure S1 in the Supporting Information). Hence, two EPR-active FA are sufficient to explore all seven binding sites simultaneously. Figure 5 illustrates the DEER data of FA/HSA/buffer complexes (2 FAs per HSA) in the presence and absence of BmimBF4 or choline dhp. The FAs are labeled at two different positions (5and 16-DSA) to sample distance information from different positions along the alkyl chain of the FA in the respective binding sites. The EPR-reporter group in 5-DSA is located close to the ionic headgroup of the FA. The DEER data thus describe the spatial distributions of the anchor units of the FA binding sites. The spin-label of 16-DSA is located close to the tail of the FA methylene chain and probes the distribution of the entry points into the binding sites. The background corrected DEER time-domain data are displayed in Figure 5A,B. Whereas the FA/HSA complexes in buffer and buffer/choline dhp solutions yield rather similar DEER data, addition of BmimBF4 causes a strong deviation of the time domain signal from the other two samples. The reliable extraction of distance information in this case is precluded for two reasons. First, the modulation depth of 5-DSA in HSA/buffer/ 15% (v/v) BmimBF4 is significantly smaller with a poor signal-tonoise ratio (at similar measurement time). As indicated in Figure 5A, the modulation depth Δ is the deviation of the plateau time domain value from unity and is directly related to the average number of coupled spins in the protein.51,52 The strong decrease in modulation depth is further evidence of the displacement of FA from the protein (as already witnessed in the CW EPR experiments in Figure 4). Furthermore, from the fact that the mobility of FAs that are still bound to HSA is increased, one may assume that the tertiary structure of the protein is also significantly altered. This could lead to a higher fraction of randomly distributed FA, even within the protein that would not contribute to the intramolecular part of the DEER-signal. One should also note that the background correction used for these measurements is derived from the measurements of HSA in buffer,36 and the addition of BmimBF4 may also lead to a reduced excluded volume and would potentially reduce the background dimensionality closer to the standard case of homogeneously

ARTICLE

distributed spins. (See the Experimental Section.) The applied background would then have a fractal dimension of 3, which in turn would reduce the modulation depth of the intramolecular signal even closer to zero. Second, the decay of the DEER time domain signal is slower when BmimBF4 is added, which qualitatively reflects weaker dipolar couplings and thus longer distances between spin labeled FAs as compared with the case in HSA/buffer and in 20% (v/v) of added choline dhp.39 It should be noted that only 15% (v/v) of BmimBF4 addition significantly decreases the dipolar interactions between two spin labeled FAs despite almost all FAs still being complexed to the protein (Figure 2B). This suggests that BmimBF4 complexed with the protein promotes unfolding of the tertiary structure albumin. In more concentrated BmimBF4 solutions (>15% (v/v)), it is impossible to obtain DEER data because the highly unfolded state diminishes the dipolar interactions of spins below the noise level. This is consistent with the local denaturation of acrylodan labeled HSA in loop 1 of domain I in BmimBF4/2% H2O (v/v).15 Frequency-domain fluorescence spectroscopy revealed that BmimBF4 in this case causes unfolding of the domain I. As the water content increases from 2 to 100%, the unfolded domain I progressively refolds, and the Ac reporter molecule’s motion recouples with domains I, II, and III. As a reference denaturant, ethanol at the same volume content was used to verify the DEER results of labeled FAs in HSA/ buffer/BmimBF4 solution. Again, the DEER signals of 5- and 16DSA in HSA/buffer/ethanol are very weak with a modulation depth at the noise level, preventing extraction of reliable distance distributions (data not shown). In contrast, for a sample containing 20% (v/v) choline dhp, the time-domain DEER signals of 5- and 16-DSA are similar to those in buffered solutions (Figure 5A,B). Because the modulation depths of labeled FAs in HSA/buffer and in HSA/buffer/ choline dhp are very similar, the number of interacting spins per protein appears to not be affected by the addition of choline dhp. However, the decay of the DEER signal is slightly slower in the case of the choline dhp-based sample. Furthermore, the 16-DSA time traces of the HSA/buffer sample exhibit well-defined modulations, which are indicative of one dominating, narrow distance peak. Such a pronounced modulation is absent after the addition of choline dhp. The corresponding DEER distance distributions are displayed in Figure 6A,B. For 5-DSA in HSA/buffer solution, a broad distance distribution is observed that covers a range from 1.5 to 4 nm with a maximum around 2.5 nm. When 20% (v/v) of choline dhp is added, the dominating distance at 2.5 nm increases to 3.1 nm, and a second distance at ∼5 nm is observed. The addition of BmimBF4 causes major changes of the distance distribution. However, the extracted distribution is prone to major artifacts due to the poor signal-to-noise ratio and will not be discussed. For 16-DSA in HSA/ buffer solution, a rather narrow distance distribution is observed with a sharp dominating distance at 3.6 nm and two smaller contributions around at 2.2 and 4.9 nm. By addition of 20% (v/v) of choline dhp, the main distance peak is observed at 3.8 nm. Also, the distance at ∼4.9 nm becomes more intense as compared with buffered solution (Figure 6B). To exclude changes in the distance distribution that may stem from potential protein aggregation induced by choline dhp, we have performed control experiments using 1 FA per protein in the solution of 16-DSA/HSA/buffer/choline dhp (20% (v/v)). These are shown in Figure S2 in the Supporting Information. These DEER data are similar to those found with two or more FAs added, with a strongly reduced modulation depth. Therefore, one can infer 1076

dx.doi.org/10.1021/bm1014156 |Biomacromolecules 2011, 12, 1072–1079

Biomacromolecules

ARTICLE

Figure 6. Comparison of the experimental distance distributions obtained by DEER with the calculated distributions from the crystal structure (blue) assuming that all seven binding sites are occupied by FAs. (A) 5-DSA probes the anchoring points of the fatty acid binding sites in the protein interior. (B) 16-DSA probes entry points of binding sites located at the protein surface. On average, the protein is loaded with 2 equiv of paramagnetic fatty acid derivatives. The experimental DEER data were obtained in purely aqueous buffer solution (black) and in buffer admixed with 20% (v/v) choline dhp (red). Note that distances >6 nm cannot be accessed by DEER with a dipolar evolution time of 2.5 μs.

that choline dhp addition does not cause aggregation of HSA. The noisy time domain signal of the BmimBF4-containing sample again does not allow for the conversion into a reliable distance distribution. The obtained distance distributions of 5- and 16-DSA represent the spatial distributions of the anchoring units and entry points of the FA binding sites, respectively. As we have shown previously, the broad distance distribution of the ionic anchoring units of HSA in aqueous buffer solution is largely in agreement with the distance distribution expected from the crystal structure (Figure 6A).36 The distances between the C-5 (or C-16) atoms of the FAs in all seven binding sites were determined from the crystal structure (Protein Data Base (PDB) entry 1e7i).31,36 In contrast, major discrepancies are observed between the DEER data (buffer solution) and the crystallographic data, which illustrates a major difference in the spatial distribution of the binding sites’ entry points in these two cases (Figure 6B). Whereas the broad distribution of the crystal structure stems from entry points that are asymmetrically distributed over the protein, the occurrence of a dominant, well-defined distance peak in the DEER distributions suggests a more symmetric distribution of the entry points on the protein’s surface.36 Note that distances >6 nm cannot be accessed by DEER with a dipolar evolution time of 2.5 μs. Remarkably, the DEER distance distributions for both 5- and 16-DSA are considerably altered upon the addition of choline dhp. This is apparent, in particular, for the distribution of the anchoring units (5-DSA, Figure 6A), where the broad distribution of distances between 1.5 and 4 nm that is observed in pure buffer is even slightly broadened. Interestingly, after the addition of choline dhp, the DEER-derived distance distribution of the anchoring points is in even better agreement with the crystal structure-derived distribution. The characteristic double peak at ∼2.5 and 3.2 nm present in the crystal structure is almost exactly reproduced by the DEER-derived distribution when choline dhp is added. The appearance of an additional feature at ∼5 nm may be attributed either to a feature that is present in the crystal structure and in pure buffered solution (at ∼4.2 nm) and that has been shifted to larger distances or to a shoulder in the distance peak at ∼5 nm. Figure 6B shows that the experimentally derived distribution of FA entry points (16-DSA) with choline dhp in principle resembles that observed in pure buffer. There are two major differences: a shift of the main distance peak to a slightly larger value of 3.8 nm (from 3.5 nm in pure buffer) and the appearance of a second large peak at ∼5 nm. The latter finding makes the entry point distribution more asymmetric than in pure buffer but unlike

for the anchoring unit distribution (Figure 6A), the entry point distribution is not in better agreement with the crystal structurederived distribution, either. It rather looks like adding choline dhp leads to a “widening” of the entry point distribution as compared to both references (crystal structure and pure buffer). Taken together, one can construct the following picture. The anchoring units for FAs are to a large part buried in the interior of the protein, and their distribution (probed by 5-DSA) when choline dhp is added is in striking resemblance to that found in the crystal structure. This indicates that the inner core of the protein that already in pure buffer showed certain rigidity is even further rigidified, and the structural variation in the solution ensemble is drastically reduced. After addition of choline dhp, the entry point distribution, as probed by 16-DSA, also becomes more asymmetric but additionally widened as compared with that in pure buffer. The larger asymmetry is a signature of partial loss of conformational flexibility that we have found in a prior study36 for the entry points on the HSA surface and which had been attributed to the apparent structural optimization of HSA for rapid uptake and release of FAs. This partial loss of flexibility could be achieved by stronger (e.g., ionic) interactions of HSA amino acid residues with ions of the IL. CW EPR spectra of the 16-DSA/HSA/buffer in the absence and presence of BmimBF4 and choline dhp are shown in Figure S3 in the Supporting Information. Again, 16-DSA molecules are displaced upon the addition of BmimBF4, whereas they are still bound to protein after choline dhp addition. The only change that is observed is that the spectral contribution of 16-DSA bound to less strongly immobilized binding sites (shown by asterisks) is slightly reduced after choline dhp addition as compared with the case in buffer. Because the spin label in 16-DSA is placed at the tail of the FA, this may suggest that some of the 16-DSA molecules that were bound to less immobilized binding pockets are now “rigidified” by the choline dhp-protein interaction. This qualitative finding is also in line with the DEER results. Note that the IL ions may even be taken up by the protein, for example, at the interfaces of several subdomains, which would in addition account for the observed widening of the distance distribution. The entry point and anchoring point distributions can be viewed as fingerprints of a “coarse-grained” functional structure. Our findings, in particular, for the anchoring unit distribution lead to the conclusion that the often described “stabilizing” effect of choline dhp and potentially also of other ILs may in fact be a “crystal structure stabilizing” effect. It seems that out of the large ensemble of protein conformations, choline dhp, at least for the more rigid inner core, stabilizes structures close to that found in the 1077

dx.doi.org/10.1021/bm1014156 |Biomacromolecules 2011, 12, 1072–1079

Biomacromolecules

ARTICLE

crystalline state. In addition, choline dhp ions may interact with HSA, for example, at the interface of secondary structure elements. This may lead to the observed widening in the distance distribution and the larger asymmetry of the entry points, which is indicative of partial loss of the flexibility on the protein’s surface.

support. This work was financially supported by the Deutsche Forschungsgemeinschaft (DFG) under grant number HI 1094/2-1 (DH) by the Foundation of the German chemical industry (FCI) and by the Graduate School of Excellence “Materials Science in Mainz” (MAINZ) (MJNJ).

’ CONCLUSIONS Here we show that a combination of standard CW EPR spectroscopy and nanoscale distance measurements through DEER spectroscopy can be used to characterize the conformational changes induced in HSA when IL are added to buffered solution. Protein-IL interactions have a denaturing effect when imidazolium-based ILs are used and a stabilizing effect when choline dhp is added. Structural information for HSA is obtained from FA binding sites via spin labeled FAs (5- and 16-DSA). CW EPR is used to discriminate bound and free FAs in solution as competitive binding of imidazolium cations into the hydrophobic channels in the protein decreases the number of bound FAs. Similar to the use of ethanol, this effect can only be partially reversed by redilution with buffer . Choline dhp addition has no such effect on the binding affinity of FAs. The distance distributions of bound FAs derived from the DEER measurements show that BmimBF4 promotes HSA unfolding even at 15% (v/v) contents, when CW EPR still shows the signature of a vast majority of bound FAs. In contrast, the binding capacity and hence the tertiary structure of HSA is largely preserved upon choline dhp addition. The distance distribution of FA anchoring units in HSA upon choline dhp addition is in even better agreement with the crystallographic data when compared with that in pure buffered solution. At the same time, choline dhp addition seems to have a widening effect on the protein surface, as evidenced from distance distribution of the entry points of FAs. This may be achieved through ionic interactions and uptake of IL ions into the protein. Our study shows that a self-assembled system of spin-labeled FAs and HSA is a model system well-suited for the quantitative characterization of the effect that solvent/solute-protein interactions have on the tertiary structure of a protein.

’ REFERENCES

’ ASSOCIATED CONTENT

bS

Supporting Information. DEER data of FAs/HSA/ buffer/20% (v/v) choline dhp with different numbers of reduced FAs and two EPR active FAs per protein molecule. DEER data of 16-DSA complexed in HSA/buffer/20% (v/v) choline dhp with different ratios of 16-DSA/HSA: (1:1) and (2:1). CW EPR spectra of 16-DSA/HSA (2/1) in buffered medium and after addition of 20% (v/v) BmimBF4 or choline dhp. This material is available free of charge via the Internet at http://pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected]. Tel: þ49 6131 379 126. Fax: þ49 6131 379 100. Present Addresses †

Department of Chemical Engineering, University of California, Santa Barbara, California.

’ ACKNOWLEDGMENT We thank Prof. Dr. Hans W. Spiess and Dr. G€ulcin Cakan Akdogan for helpful discussions and Christian Bauer for technical

(1) Laszlo, J. A.; Compton, D. L. J. Mol. Catal. B: Enzym. 2002, 18, 109–120. (2) Ohno, H; Suzuki, C; Fukumoto, K.; Yoshizawa, M.; Fujita, K. Chem. Lett. 2003, 32, 450–451. (3) Baker, S. N.; McCleskey, T. M.; Pandey, S.; Baker, G. A. Chem. Commun. 2004, 940–941. (4) Fujita, K.; MacFarlane, D. R.; Forsyth, M. Chem. Commun. 2005, 38, 4804–4806. (5) De Diego, T.; Lozano, P.; Gmouh, S.; Vaultier, M.; Iborra, J. L. Biomacromolecules 2005, 6, 1457–1464. (6) Fujita, K.; MacFarlane, D. R.; Forsyth, M.; Yoshizawa-Fujita, M.; Murata, K.; Nakamura, N.; Ohno, H. Biomacromolecules 2007, 8, 2080– 2086. (7) Wasserscheid, P.; Keim, W. Angew. Chem., Int. Ed. 2000, 39, 3772–3789. (8) P^arvulescu, V. I.; Hardacre, C. Chem. Rev. 2007, 107, 2615–2665. (9) Chen, X.; Liu, J.; Wang., J. Anal. Methods 2010, 2, 122–1226. (10) Byrne, N.; Wang, L.-M.; Belieres, J.-P.; Angell, C. A. Chem. Commun. 2007, 26, 2714–2716. (11) Lange, C.; Patil, G.; Rudolph, R. Protein Sci. 2005, 14, 2693– 2701. (12) De Diego, T.; Lozano, P.; Gmouh, S.; Vaultier, M.; Iborra, J. L. Biotechnol. Bioeng. 2004, 88, 916–924. (13) Constantinescu, D.; Weing€artner, H.; Herrmann, C. Angew. Chem., Int. Ed. 2007, 46, 8887–8889. (14) McCarty, T. A.; Page, P. M.; Baker, G. A.; Bright, F. V. Ind. Eng. Chem. Res. 2008, 47, 560–569. (15) Page, T. A.; Kraut, N. D.; Page, P. M.; Baker, G. A.; Bright, F. V. J. Phys. Chem. B 2009, 113, 12825–12830. (16) Baker, G. A.; Heller, W. T. Chem. Eng. J. 2009, 147, 6–12. (17) Constantinescu, D.; Herrmann, C.; Weing€artner, H. Phys. Chem. Chem. Phys. 2010, 12, 1756–1763. (18) Berliner, L. J., Eaton, G. R., Eaton, S. S. Distance Measurements in Biological Systems by EPR; Biological Magnetic Resonance; Kluwer: New York, 2000; Vol. 19. (19) Jeschke, G.; Polyhach, Y. Phys. Chem. Chem. Phys. 2007, 9, 1895–1910. (20) Fajer, P. G. J. Phys.: Condens. Matter 2005, 17, 1459–1469. (21) Steinhoff, H.-J. Biol. Chem. 2004, 385, 913–920. (22) Hubbell, W. L.; Cafiso, D. S.; Altenbach, C. Nat. Struct. Biol. 2000, 7, 735–739. (23) Hilger, D.; Polyhach, Padan, E.; Jung, H.; Jeschke, G. Biophys. J. 2007, 93, 3675–3683. (24) Smirnova, I.; Kasho, V.; Choe, J.-Y.; Altenbach, C.; Hubbell, W. L.; Kaback., H. R. Proc. Natl. Acad. Sci. U.S.A 2007, 104, 16504– 16509. (25) Bennati, M.; Robblee, J.; Mugnaini, V.; Strubbe, J.; Freed, J.; Borbat, P. J. Am. Chem. Soc. 2005, 127, 15014–15015. (26) Milov, A.; Samoilova, R. I.; Tsvetkov, Y. D.; Formaggio, F.; Toniolo, C.; Raap, J. J. Am. Chem. Soc. 2007, 129, 9260–9261. (27) Schiemann, O.; Piton, N.; Mu, Y.; Stock, G.; Engels, J. W.; Prisner, T. F. J. Am. Chem. Soc. 2004, 126, 5722–5729. (28) Carter, D. C.; Ho, J. X. Adv. Protein Chem. 1994, 45, 153–203. (29) Curry, S. Drug. Metab. Pharmacokinet. 2009, 24, 342–357. (30) Curry, S.; Brick, P.; Franks, N. P. Biochim. Biophys. Acta, Mol. Cell Biol. Lipids 1999, 1441, 131–140. (31) Bhattacharya, A. A.; Gr€une, T.; Curry, S. J. Mol. Biol. 2000, 303, 721–732. (32) Simard, J. R.; Zunszain, P. A.; Hamilton, J. A.; Curry, S. J. Mol. Biol. 2006, 361, 336–351. 1078

dx.doi.org/10.1021/bm1014156 |Biomacromolecules 2011, 12, 1072–1079

Biomacromolecules

ARTICLE

(33) Simard, J. R.; Zunszain, P. A.; Ha, C.-E.; Yang, J. S.; Bhagavan, N. V.; Petitpas, I.; Curry, S.; Hamilton, J. A. Proc. Natl. Acad. Sci. U.S.A 2005, 102, 17958–17963. (34) Kim, M.; Xu, Q.; Murray, D.; Cafiso, D. D. Biochemistry 2008, 47, 670–679. (35) Rao, J. N.; Jao, C. C.; Hegde, B. G.; Langen, R.; Ulmer, T. S. J. Am. Chem. Soc. 2010, 132, 8657–8668. (36) Junk, M. J. N.; Spiess, H. W.; Hinderberger, D. Angew. Chem., Int. Ed. 2010, 49, 8755–8759. (37) Dockter, C.; Volkov, A.; Bauer, C.; Polyhach, Y.; Joly-Lopez, Z.; Jeschke, G.; Paulsen, H. Proc. Natl. Acad. Sci. U.S.A 2009, 106, 18485– 18490. (38) Pannier, M.; Veit, S.; Godt, A.; Jeschke, G.; Spiess, H. W. J. Magn. Reson. 2000, 142, 331–340. (39) Jeschke, G. Pannier, M. Spiess, H. W. In Biological Magnetic Resonance: Distance Measurements in Biological Systems by EPR; Kluwer Academics: New York, 2000; Vol. 19. (40) Jeschke, G.; Bender, A.; Paulsen, H.; Zimmerman, H.; Godt, A. J. Magn. Reson. 2004, 169, 1–12. (41) Jeschke, G.; Chechik, V.; Ionita, P.; Godt, A.; Zimmermann, H.; Banham, J.; Timmel, C. R.; Hilger, D.; Jung, H. Appl. Magn. Reson. 2006, 30, 473–498. (42) Columbus, L.; Hubbell, W. L. Trends Biochem. Sci. 2002, 27, 288–295. (43) Akdogan, Y.; Heller., J.; Zimmermann, H.; Hinderberger, D. Phys. Chem. Chem. Phys. 2010, 12, 7874–7882. (44) Rosenberg, R. M.; Rogers, D. W.; Haebig, J. E.; Steck, T. L. Arch. Biochem. Biophys. 1962, 97, 433–441. (45) Michnik, A. J. Therm. Anal. Calorim. 2007, 87, 91–96. (46) Taboada, P.; Barbosa, S.; Castro, E.; Gutierrez-Pichel, M.; Mosquera, V. Chem. Phys. 2007, 340, 59–68. (47) Avdulov, N. A.; Chochina, S. V.; Daragan, V. A.; Schroeder, F.; Mayo, K. H.; Wood, W. G. Biochemistry 1996, 35, 340–347. (48) Jeschke, G. In Advanced ESR Methods in Polymer Research; Schlick, S., Ed.; Wiley: New York, 2006. (49) Zhao, H. J. Mol.Catal. B: Enzym. 2005, 37, 16–25. (50) Junk, M. J. N.; Spiess, H. W.; Hinderberger, D. submitted, 2010. (51) Milov, A. D.; Ponomarev, A. B.; Tsvetkov, Y. D. Chem. Phys. Lett. 1984, 110, 67–72. (52) Bode, B. E.; Margraf, D.; Plackmeyer, J.; Durner, G.; Prisner, T. F.; Schiemann, O. J. Am. Chem. Soc. 2007, 129, 6736–6745.

1079

dx.doi.org/10.1021/bm1014156 |Biomacromolecules 2011, 12, 1072–1079