Saikat
Pal
a,
Partha
Pyne
a,
Nirnay
Samanta
b,
Simon
Ebbinghaus
*b and
Rajib Kumar
Mitra
*a
aDepartment of Chemical, Biological and Macromolecular Sciences, S N Bose National Centre for Basic Sciences, Block JD, Sector III, Salt Lake, Kolkata 700106, India. E-mail: rajib@bose.res.in
bInstitute for Physical and Theoretical Chemistry, TU Braunschweig, BRICS, Rebenring, 56 D-38106 Braunschweig, Germany. E-mail: S.Ebbinghaus@tu-braunschweig.de
First published on 27th November 2019
Cells are crowded with various cosolutes including salts, osmolytes, nucleic acids, peptides and proteins. These cosolutes modulate the protein folding equilibrium in different ways, however, a unifying concept remains elusive. To elucidate the cosolute size-effect, macromolecular crowders are commonly compared to their monomeric building blocks (e.g. dextran vs. glucose or polyethylene glycol with different degrees of polymerization). To the best of our knowledge, such studies do not exist for protein crowders, raising the question of how single amino acids modulate the folding equilibrium. Therefore, we investigate the effect of glycine, alanine, proline and arginine on the stability of a model globular protein bovine serum albumin (BSA) upon thermal and urea-induced unfolding. We use three complementary techniques, fluorescence spectroscopy (as a local site-specific probe), circular dichroism (as a global probe for α-helical structure) and differential scanning calorimetry (to probe the energetics of unfolding). We find that the amino acids modulate BSA stability and unfolding, however, without following a particular trend with either the hydrophobicity scale or the solvent accessible surface area (SASA) of the added amino acids. Our data rather suggest that solvation effects play a role in understanding the cosolute effect.
Although several experimental and theoretical studies on the protein folding–unfolding equilibrium in the presence of co-solutes are available in the literature, still the exact mechanism remains elusive because of the inherent complexity of such systems. “Counteracting” osmolytes15,16 (e.g. TMAO, glycerophosphorylcholine, betaine, etc.) affect both the stability and functional activity of proteins,17,18 whereas “compatible” osmolytes16 like sucrose and some amino acids affect their stability.15 Further insights were gained from the comparison of polysaccharide macromolecular crowders (such as Ficoll and dextran) to their monomeric building blocks (such as sucrose and dextran). Such experiments showed that both act similarly on the folding equilibrium, questioning ‘excluded volume effects’ to be the main contribution to the observed increase in stability. In fact, spectroscopic and calorimetric experiments showed that the stability increase is mainly governed by enthalpy rather than entropy.9 This seeded further experiments and discussion on the contribution of different crowding effects on protein in-cell stability.19 However, polysaccharides are not good crowding agents to mimic the cellular environment, where proteins and other biomolecules (including smaller amino acids, peptides and metabolites) exert the crowding effect. In this regard amino acids stand as a more suitable candidate as they are the building blocks of proteins and are abundant in the cellular environment. As such their thermodynamic fingerprint is required to elucidate if they modulate the protein folding equilibrium in a colligative manner or if they display amino acid specific behavior. Some earlier studies have shown that amino acids can act as protective osmolytes to stabilize protein structure.20,21 An earlier report by Shiraki et al. concluded that arginine can prevent aggregation of lysozyme.21 Yancey et al.16 observed that amino acids like glycine, alanine, proline, taurine etc. do not significantly perturb the enzymatic activity of pyruvate kinase whereas some basic amino acids like arginine and lysine show significant modulation. Taneja et al.22 studied the effect of a series of amino acids on thermal denaturation of cytochrome c and concluded that amino acids could act as neutral or a stabilizer or a destabilizer depending upon their structures. In a previous study we have investigated the effect of a series of amino acids on their collective hydration dynamics.23 We found that the hydration dynamics change with the hydrophobicity of the amino acids and solvent accessible surface area (SASA). This suggested that amino acids, when added externally, could influence the hydration dynamics and stability of a protein. In this contribution we have studied the effect of four different amino acids, glycine (Gly), alanine (Ala), arginine (Arg) and proline (Pro), on bovine serum albumin (BSA) stability. BSA was chosen as a probe as it is commonly used as a crowder to mimic the densely crowded cytoplasm.19 Further, BSA is a well-studied water soluble globular protein, usually monomeric in physiological conditions and also easily available commercially. Gly is the smallest and the only achiral amino acid, Ala has one carbon atom more than Gly, and Pro consists of a five membered nitrogen containing ring. Arg is a basic amino acid containing hydrophilic as well as hydrophobic groups and possesses structural resemblance with guanidinium hydrochloride.
We investigated the effect of these amino acids on thermal as well as urea mediated unfolding of BSA. We used steady state fluorescence spectroscopy to monitor the local environment of the Trp212 moiety in BSA in the absence and in the presence of amino acids as well as in 4 M urea. Temperature dependent circular dichroism measurements were carried out to monitor the change in the secondary and tertiary structures of the protein and to estimate the associated thermodynamic parameters. Differential scanning calorimetry (DSC) measurements were used to measure the melting temperature (Tm) and enthalpy (ΔH) of unfolding.
Steady state fluorescence measurements of the protein in the presence and in the absence of amino acids and urea were performed using a Fluorolog 3 (Horiba, Jobin Yvon) instrument. The samples were excited at 295 nm in order to avoid any possible fluorescence contribution from the tyrosine moiety of the protein. Far UV (190–260 nm) circular dichroism spectroscopic measurements were performed using a JASCO J-815 spectrometer with a Peltier attachment for the temperature dependent measurements using a 0.1 cm path-length quartz cuvette. The secondary structure of the protein was calculated using CDNN software (http://http//bioinformatik.biochemtech.uni-halle.de/cdnn).
Thermal denaturation of BSA was studied by temperature dependent CD measurements both in the absence and in the presence of amino acids and urea. To eliminate the effect of protein concentration, the CD value was calculated in terms of molar ellipticity units using the following relation:
We assumed a two-state protein folding–unfolding equilibrium model between the native state ‘N’ and the unfolded state ‘U’. At any temperature T the equilibrium constant (K) for this process is given by: where [U] and [N] are the concentrations of the unfolded and the native forms, respectively. The native fraction (φ) present at any temperature T is given by:
(1) |
(2) |
The standard free energy of unfolding (ΔG0u) is obtained by the equation24–26
(3) |
For simplicity we ignore the superscript “0” and subscript “u” in the ΔG0u(T) term and use ΔG(T) throughout the manuscript. For the enthalpy (ΔH) and entropy (ΔS) terms also the same notations are applicable and we ignore the superscript “0” and the subscript “u”. The corresponding van’t Hoff enthalpy (ΔHVF) of unfolding was estimated using the following non-linear equation,9,27
(4) |
The thermal stability of BSA in the absence and in the presence of the amino acids was measured using a MicroCal PEAQ-DSC system (Malvern Panalytical) at a scan rate of 90 °C hour−1 (without feedback mode) in the temperature range of 20–90 °C. The evaporation/boiling of the liquids were prevented by applying a constant pressure over the solution in both the reference and the sample cells (250 μL solution in each). The BSA concentration was kept fixed at 10 μM and the amino acid concentration was 0.08 M for all the measurements. Before each scan of the protein sample several buffer–buffer (or amino acid solution–amino acid solution) scans in the same conditions were performed until reproducibility of the data was achieved and the last data were used for baseline corrections of the protein sample. All the data were analyzed using Microcal PEAQ-DSC software. The calorimetric enthalpy (ΔHcal) was measured as the area under the curve of the excess molar heat capacity (Cp, baseline corrected) of each transition.
(5) |
This is irrespective of any model. The corresponding van’t Hoff enthalpy (ΔHv) was estimated as
(6) |
The MicroCal PEAQ-DSC software uses Levenberg–Marquardt non-linear least-squares methods to fit the Cp(T) data with the following model:
(7) |
We also monitor temperature-induced unfolding–refolding of BSA in the presence of the amino acids using fluorescence measurements. Previous time-resolved fluorescence and FRET measurements show that human serum albumin (HSA) forms a distinct intermediate structure below 338 K and the protein undergoes global unfolding beyond 348 K.34 THz measurements showed that the protein hydration dynamics changes according to the same temperature behavior.35 Since HSA and BSA are structurally much analogous with respect to the amino acid sequence as well as tertiary structure such two-step unfolding could also be seen in the case of BSA. We monitor the emission profile of BSA at different temperatures in the presence of the amino acids (a representative plot for BSA in the presence of Gly is provided in Fig. 2b). For BSA in buffer, the emission intensity is quenched as the temperature is increased; this might be due to the increase in the non-radiative decay channels of Trp36 as we also observe this in bare Trp in buffer solution (Fig. S1, ESI†). Up to 328 K, the emission peak does not undergo noticeable change, however, beyond that it shows a blue shift. Such a blue shift is not apparent in the case of bare Trp in water (Fig. S1, ESI†). Upon temperature-induced unfolding the protein exposes its otherwise buried hydrophobic moieties towards the Trp, and correspondingly the peak shifts to smaller wavelength. A similar behavior is observed in the presence of the amino acids also; the extent of the change in the emission intensity (at the peak), however, is different for different amino acids (Fig. 2b). The slope of the relative intensity (I/I0) as a function of temperature changes significantly beyond 328 K (Fig. 2b, inset). It can be noted here that the temperature induced change in I/I0 of Trp in buffer is relatively high compared to that of Trp in BSA (Fig. S1, ESI†). This suggests that the Trp moiety is not entirely exposed towards the solvent during unfolding. We compare the decrease in the relative intensity in the native (293 K) and unfolded (353 K) states as a function of the amino acid (Fig. 2c). It is found that in the buffer the decrease is ∼76%. In Ala, Arg and Pro the decrease is comparable to that of the buffer. For Gly we observe a marginal however statistically significant (see Fig. S3 and Table S6 in the ESI†) change. This result suggests that Gly mildly stabilizes the local environment of subdomain IIA of BSA. We also monitor the refolding process as the temperature is first raised to 353 K and then cooled to 293 K. We found that both the intensity and the emission peak do not recover to their native values (Fig. 2b, broken lines). In the presence of buffer the intensity is recovered up to 60% while the peak remains mostly blue shifted. A similar behavior is observed in the presence of the different amino acids. We plot the change in relative intensity at 293 K due to refolding as a function of the amino acids (Fig. 2d). We find that the amino acids promote the refolding with decreasing efficiency in the order: Gly > Arg > Pro, whereas Ala does not show any effect.
(8) |
The energetics of thermal unfolding of BSA is calculated by fitting the CD signals at 222 nm using eqn (4). We fit the ΔG(T) vs. T curves near Tm9 and obtain reasonably good fits (Fig. 4a). The fitted parameter ΔHVF and ΔCp values are presented in Table S4 (ESI†) and Fig. 4b (open symbols). For BSA we obtain ΔHVF = 30.8 ± 0.6 kcal mol−1 and ΔCp = 1.01 ± 0.05 kcal K−1 mol−1; these values are in comparable agreement with those for previously reported globular proteins using CD measurements.38 Upon the addition of the amino acids, ΔHVF of unfolding increases for Ala, Pro and Arg while in the presence of Gly it decreases modestly.
Fig. 4 Change of Gibbs free energy of BSA in buffer and in different amino acids in the absence (a) and presence (c) of 4 M urea. The solid lines are non-linear curve fits (see eqn (4)). (b) van’t-Hoff enthalpy (ΔH) at T = Tm and change in heat capacity (ΔCp) of BSA for the buffer and amino acids. ΔH and ΔCp are calculated in kcal mol−1 and kcal K−1 mol−1 units respectively. (d) van’t-Hoff enthalpy (ΔH) at T = Tm and change in heat capacity (ΔCp) of BSA for the buffer and amino acids in the presence of 4 M urea. ΔH and ΔCp are calculated in kcal mol−1 and kcal K−1 mol−1 units respectively. |
Interestingly, when we plot the first derivative of θ222 of BSA in buffer with respect to T (we have smoothed the data using a well-known Savitzky–Golay least squares procedure39) we do not observe a sharp peak as expected for a two-state unfolding model, and instead an additional hump is observed at a higher temperature (Fig. S2a, ESI†). We deconvolute the dθ222/dT curve into two Gaussians and identify two transition temperatures TIm and TIIm corresponding to two unfolding processes. The appearance of two peaks is also evident in the case of the amino acids also (Fig. S2, ESI†), a representative deconvolution for BSA–Arg is shown in Fig. 3c. Such non-two state unfolding for BSA has been reported in earlier studies.40,41 In buffer, we obtain TIm ∼337.3 K and TIIm ∼352.9 K. Both TIm and TIIm increase in the presence of the amino acids except for Ala (TIIm decreases for Ala compared to the buffer) (Fig. 3d); the changes are statistically significant (Fig. S4 and Table S7 in the ESI†) and the trend is comparable to that of Tm (Fig. 3b).
We obtain the corresponding calorimetric enthalpies (ΔHIcal and ΔHIIcal) by fitting the experimental Cpvs. T curves to a non-two state model using Marquardt non-linear least-squares methods (eqn (7)), and deconvoluting the area under the double humped curves of Cpvs. T. The obtained enthalpies are presented in Table S5 (ESI†) and Fig. 5c. We observe that in the presence of Ala ΔHIcal increases by ∼6 kcal mol−1 with respect to that in buffer, while its value is reduced when Gly, Pro and Arg are added individually in the protein solution following the order of ΔHIcal as: Gly ≫ Pro > Arg. We also calculate the change in the enthalpy contribution of unfolding (ΔΔH = ΔHprotein–AA − ΔHprotein–buffer) for each amino acid (Table S5, ESI†). We find that for Gly the ΔΔH value is highly negative while that for Ala is positive. This feature also corroborates the ΔΔH values obtained from the CD measurements (Table S4, ESI†). For process-II the increase in enthalpy upon the addition of the amino acid is quite noticeable for Ala (96.3 kcal mol−1). Pro and Arg exhibit a minor decrease in ΔHIIcal while a significant increase is observed for Gly (16.2 kcal mol−1). We estimate the corresponding van’t Hoff enthalpies using eqn (6) and the values are provided in Table S4 (ESI†). The values of the van’t Hoff enthalpies differ significantly from the corresponding calorimetric enthalpies, implying that the latter count all the contributions from the system including solvent rearrangement while the former do not.45 The ratio ΔHcal:ΔHv allows us to consider the cooperativity of unfolding. If the ratio is close to unity, the unfolding process might grossly be approximated by a two-state model. We found that the ratio is higher than one for process I, which suggests that there are several intermediate processes involved in the folding equilibrium. The ratio is the lowest in Gly (1.4), while it is the highest in Ala (2.3). On the other hand, the ratio is lower than unity for process II. We calculate the calorimetric entropy of unfolding (ΔScal) from the area under the curve of Cp/T vs. T for each transition (Fig. 5d). We calculate the corresponding change in entropy (ΔΔS) i.e. ΔΔS = ΔSprotein–AA − ΔSprotein–buffer (Table S5, ESI†). We observe that ΔΔS is negative for Gly while it is negligible in Ala.
We measure the fluorescence intensity of the Trp residue of BSA in the presence of 4 M urea. We observe that the emission intensity is quenched drastically in 4 M urea solution in comparison to that in buffer (Fig. 2e), indicating that BSA starts to unfold. Addition of the amino acids increases the intensity of the urea-mediated partially unfolded form of the protein. We also plot the relative change of the fluorescence peak intensity as a function of the amino acids (Fig. 2e, inset). The extent of relative change is higher in Ala compared to that in Gly; the Trp peak intensity of the partially unfolded BSA decreases as: Arg > Pro > Ala ≫ Gly.
Further, we perform temperature dependent CD measurements of BSA in the presence of urea and the amino acids. We observe that Ala increases TIm compared to aqueous solutions (Table S3, ESI†). The effect is marginal in the other three amino acids. We also calculate the enthalpy (ΔHVF) and change in heat capacity (ΔCp) for the two-state folding–unfolding equilibrium process by non-linear fitting of ΔG(T) versus T with the help of eqn (4), taking the values from Tm to the near neighborhood of Tm (Fig. 4c and Table S4, ESI†). The ΔHVF value is very small in urea compared to that in buffer (Table S4, ESI†). The value increases in the presence of the amino acids, suggesting their role in stabilizing the protein, the effect being the most prominent in Ala.
The CD measurements show that the amino acids increase the helicity of the protein (Fig. 1a), which could be explained by an excluded volume effect.3 However, the amino acid concentrations used in the experiments are small (0.08 M) and therefore the changes observed are only modest. We obtain the melting temperature by a non-two-state model, as it correlates the melting of two different domains. The results were also confirmed by a two-state (Fig. 3b) model. However, it can be noted that the trends of Tm are identical in both cases. Addition of the amino acids to BSA grossly resists the unfolding of the α-helical content in all the sub-domains, the effect being more pronounced in Pro.
The unfolding thermodynamics obtained by DSC provides information about the changes in the hydration behavior of polar and non-polar amino acids upon unfolding and the changes in internal interactions (van der Waals, hydrogen bonding etc.) and conformational entropy. We observe two transitions of BSA that are attributed to the thermal unfolding of two different domains of the protein. We observe that the TIm values obtained from DSC measurements are a bit lower than those obtained from CD measurements. The dissimilarity arises from the fact that while CD measurements associate changes in the more structured α-helical content only, DSC measurements associate the global unfolding, which involves less structured secondary motifs also. The van’t Hoff enthalpy is smaller compared to the calorimetric one, which illustrates the non-two-state unfolding equilibrium in the protein. The striking difference in both the enthalpies (van’t Hoff enthalpy and calorimetric enthalpy) of Gly and Ala is intriguing; while Gly significantly lowers it, Ala increases it modestly, the effect being more prominent in process II. The CD measurements imply that both Gly and Ala increase TIm while the DSC thermograms reveal a marginal decrease in TIm for Gly. This apparent contrasting behavior can be discussed in the light of their hydration behavior yielding contrasting solvation energy. One more interesting observation from this overall investigation is that the thermodynamic parameters as well as the protein structural perturbations neither follow a linear relationship with the hydrophobicity scale nor the SASA of the amino acids, thus these parameters are not sufficient to explain the experimental data.
The analysis of the amino acid mediated changes in the thermodynamic parameters of the protein reveals a contrasting behavior between Gly and Ala, although they differ in only one additional methyl group in Ala. We find that Ala produces a positive change in the enthalpy contribution (ΔΔH, Table S5, ESI†), while the corresponding change in entropy is only marginal. It can be argued that enthalpy stabilization (positive ΔΔH), as evidenced in Ala, manifests a classical osmolyte like behavior48,49 of this amino acid in which protein stabilization occurs through a change in the hydration layer of the protein mediated by the osmolyte.50 On the other hand, we observe a negative ΔΔH in Gly and a large negative ΔΔS value. The change in the entropic effect in the protein folding–unfolding process in aqueous media principally emanates from two contributions: one is due to conformational entropy, which is in turn related to the changes in the conformation of the protein structure transforming from a native to an unfolded state, and the other one is associated with the hydration of polar and non-polar groups.51 We make a rough estimate of the conformational contribution from the temperature dependent CD measurements. The van’t Hoff entropy (ΔSVF) associated with the unfolding process can be obtained by the relation (Table S4, ESI†). We observe that the corresponding ΔΔSVH of Gly is slightly negative while those for the other amino acids are largely positive. A positive change in the conformational entropy during unfolding is believed to stabilize the unfolded state of the protein while that for the hydration favors the folded state.51,52 The overall negative value of ΔΔS in Gly (Table S5, ESI†) is thus partly favored by the conformation entropy contribution. On the other hand, in Ala and the other amino acids the positive contribution from the conformational entropy is (over)compensated by the negative hydration contribution. It should be noted here that the estimation of the relative contribution is only an approximate one and does not take into consideration any possible direct interaction between the amino acids and the protein towards the overall entropy change.
Gly thus acts as a non-conventional osmolyte and the occurrence of such a negative ΔΔH is comparable to that observed in a protein denaturant molecule urea, which identifies a direct interaction of protein–osmolyte.9 At this point it is interesting to consider the individual hydration behavior of the cosolutes. Our previous THz spectroscopic results, which probe the collective hydrogen bond dynamics of water, have concluded contrasting hydration behavior of Gly as compared to other amino acids; while amino acids in general are water ‘structure makers’, Gly is a water ‘structure breaker’ and in that respect it is comparable to urea, which also destabilizes the hydrogen bonded network of water.53 Our present study therefore invokes the idea that osmolyte induced stabilization/destabilization of the protein and the related energetics are correlated with the associated change in the hydration dynamics of both the protein and the added co-solute(s) taken together. A more detailed systematic investigation involving amino-acid like molecules with varying carbon chain length at different pH conditions is therefore much needed to generalize their effect on protein stability and functionality.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c9cp04887a |
This journal is © the Owner Societies 2020 |