Hydrophilic Association in a Dilute Glutamine Solution Persists

Dec 3, 2015 - Hydrophilic Association in a Dilute Glutamine Solution Persists. Independent of Increasing Temperature. Natasha H. Rhys,. †. Alan K. S...
0 downloads 0 Views 1MB Size
Download PDF
Article pubs.acs.org/JPCB

Hydrophilic Association in a Dilute Glutamine Solution Persists Independent of Increasing Temperature Natasha H. Rhys,† Alan K. Soper,*,‡ and Lorna Dougan† †

School of Physics and Astronomy, University of Leeds, Leeds LS2 9JT, U.K. ISIS Facility, STFC Rutherford Appleton Laboratory, Harwell Oxford, Didcot, Oxon OX11 OQX, U.K.

‡

ABSTRACT: Recent studies suggest that hydrophilic interactions play an important role in controlling self-assembly in biological processes. To explore the effect of temperature on this interaction, we extend our previous work on the glutamine−water system at 24 °C (at a mole ratio of 1 glutamine to 269 water molecules) and present additional neutron diffraction data, at the same concentration, at 37 and 60 °C, using hydrogen/deuterium substitution on the water and glutamine, coupled with further extensive empirical potential structure refinement computer simulations. Taking all the possible hydrophilic couplings between glutamine molecules into account, we find that nearly one-fifth of the glutamines in solution are linked by hydrogen bonds at any one time. This number contrasts strongly with the ∼3−4% fraction found in the same simulation with random packing and no hydrogen bonds. Within the uncertainties imposed by dilute solution statistics, we find no temperature dependence in these values. The clusters are highly transitory, forming and disappearing rapidly as the simulations proceed. Hydrophobic association of the alkyl groups on glutamine without concomitant hydrophilic association of the charged head and side-chain groups is only weakly observed.

1. INTRODUCTION The structure and function of biological molecules is directly linked with their hydration. Understanding the interactions between water and biological molecules and how those interactions influence self-assembly and conformational changes is critical to understanding a wide range of biological processes. In the folding and association of proteins, and the maintenance of lipid bilayers and cell membranes, the hydrophobic interaction is generally accepted to be the driving force and has dominated much of the thinking in the last 70 years concerning biomolecular interactions in solution,1−3 although the molecular details of this interaction remain somewhat controversial. The thermodynamics of the association process is well established; when two nonpolar groups come together in aqueous solution, the free energy change is negative, resulting from a dominant, positive entropy change. Historically, the source of the positive entropic change was attributed to a change in the structure of the solvent in the environment of the nonpolar group, as a result of more ordered water than in the bulk.4 Thus, the change in the entropy of the solvent was considered to be the entropic driving force of the hydrophobic association process. While this model from the 1940s was conceptually very appealing, experimental structural techniques in the last two decades have struggled to find any experimental justification for enhanced atom-scale ordering in water around nonpolar groups. Instead, these experiments suggested other sources for the entropic driving force. For example, neutron diffraction experiments on alcohol and water mixtures (including t-butanol and methanol) have suggested that it is in the water second coordination shell (i.e., the © 2015 American Chemical Society

structure of water molecules around other water molecules) that the entropic driving force for the hydrophobic interactions is to be found.5−10 Furthermore, the same scattering evidence suggests that the negative excess entropy of mixing observed in alcohol−water systems could arise from incomplete mixing at the molecular level, rather than from water restructuring. Indeed, there is clear structural evidence of molecular segregation or microscopic phase separation.6 Due to the advancement of radiation scattering techniques combined with computational modeling, it is now possible to measure liquid structure and changes in hydrogen bonding as a function of solute concentration11−22 and in some cases temperature.23−26 Such studies of solute association have shown rich behavior, which reflects a complex balance of polar and nonpolar interactions. Uncovering the detailed nature of this balance in simple solutions may lead to a better understanding of the forces that control biomolecule structure and self-assembly. In particular, the process by which proteins fold in solution to form their native, folded structure is still not well understood. While the hydrophobic effect is frequently employed to explain this process, hydrogen bonding and charge−charge interactions are also known to play important roles in determining protein stability. A detailed understanding of protein folding will only be possible with further understanding of the delicate interplay between these interactions. In recent work, the relative roles of hydrophobic and hydrophilic Received: July 31, 2015 Revised: November 30, 2015 Published: December 3, 2015 15644

DOI: 10.1021/acs.jpcb.5b07413 J. Phys. Chem. B 2015, 119, 15644−15651

Article

The Journal of Physical Chemistry B

2. METHODS 2.1. Neutron Scattering Experiment. The data reported here was accumulated at the same time as the previous experiment.33 Data were recorded on the Small Angle Neutron Diffractometer for Amorphous and Liquid Samples (SANDALS) at the ISIS pulsed neutron facility. Solutions of glutamine and water were made up in a glutamine:water molar ratio of 1:269. Note that this concentration is approaching the solubility limit of the molecule; ideally, a mole ratio of at least 1:50 would have been preferred if it were available. A total of four solutions were prepared at this concentration, namely, protiated glutamine and light water (H2O), protiated glutamine and heavy water (D2O), glutamined5 and H2O, and glutamine-d5 and D2O. In glutamine-d5, the five nonexchangeable hydrogen atoms are replaced with deuterons. The solutions were injected into standard 1 mm internal thickness TiZr alloy flat plate containers and mounted on the SANDALS sample changer, which was used also to control the temperature at three values, 24, 37, and 60 °C. The TiZr alloy is used for the container because it is both corrosion resistant and has effectively zero coherent scattering for neutrons, thus producing a mostly flat background in the scattering spectra. Neutron scattering data were corrected for background, multiple scattering, container scattering, and selfattenuation and placed on an absolute scale of differential scattering cross section per steradian per atom by comparison with the scattered intensity from a standard 3 mm slab of vanadium. A correction for inelastic neutron scattering was made using the iterative method described in ref 38. 2.2. Data Interpretation. As in the previous report, data interpretation was performed using empirical potential structure refinement (EPSR).39 In this method, an initial computer simulation box of water and glutamine molecules is built at the same mole ratio as in the experiment. The atomic number density for this simulation box was set to 0.1 atoms/Å3 at 24 °C, with lower densities of 0.0997 atoms/Å3 and 0.0987 atoms/Å3 at 37 and 60 °C, respectively, corresponding to the lower densities expected for pure water at these higher temperatures.40 For this simulation the glutamine carbon atoms are labeled Cb, Cm, and Cs; the nitrogen atoms Nb and Ns; the oxygen atoms Ob and Os; and the hydrogen atoms Hb, Hs, and Hm, according to where they occur in the backbone (b), side chain (s), or alkyl chain (m). Ca corresponds to the carbon at the center of the amino acid backbone. Water atoms are labeled Ow and Hw. The structure of an isolated glutamine molecule was initially refined with the semiempirical quantum chemistry program MOPAC-7 using the AM1 Hamiltonian.41 This was then used to construct a tetragonal simulation box of dimension 187.72922 × 93.864609 × 93.864609 Å3, containing 200 glutamine molecules and 53 800 water molecules at 24 °C, with correspondingly increased dimensions for the higher temperatures. The potential parameters were the same as those used in the previous work,33 but a significant difference was that in the current work internal dihedral angles within the glutamine molecules were constrained close the values generated by the MOPAC-7 calculation. This ensured that the carboxylate and amide groups retained their initially planar geometries, something that did not happen consistently in the previous simulations (see Figure 11 of ref 33). Another significant difference in the present work compared to previously reports is that the number of molecules in the

interactions in the process of protein folding were considered by determining the structure of dipeptide fragments in aqueous solution.27 Three dipeptides were considered, each containing different hydrophobic and hydrophilic portions, and their structure and association in solution were obtained using neutron diffraction experiments coupled with computational modeling. Thus, the relative importance of hydrophobic and hydrophilic association could be directly monitored. The results were surprising in that charged sites on the peptides were found to dominate in the association of dipeptides, resulting in the formation of self-assembled, larger structures in solution. Thus, the results indicated that hydrophilic interactions, rather than hydrophobic interactions, are the dominant force in the association of the dipeptides. Therefore, this work strongly suggests that hydrophilic interactions require further understanding to determine their importance in the stability and association of biological molecules in solution. It is also important to discover how hydrophilic interactions change with increasing temperature. While there is much literature on the hydrophobic interaction, including studies at different temperatures,28,29 comparatively less is known about the relative strength of hydrophilic interactions as a function of temperature. Such insight would help in the general understanding of the relative roles of the driving forces in biomolecular stability and association. Furthermore, while protein folding under ambient temperature conditions is well studied, the discovery of organisms thriving under extreme temperature conditions raises interesting questions about the driving force in biomolecule assembly at nonambient temperatures. For example, thermophilic organisms are adapted to exist in environments with high temperatures and exhibit optimum growth in the range of 45−80 °C.30 At these elevated temperatures, proteins from thermophilic organisms maintain their native, folded structure while having the stability and flexibility to complete their biological function.30−32 It is therefore interesting to consider which interactions in the protein contribute to this stability and how the strength of these interactions depend on temperature. Such insight would shed light on the origin of protein stability in thermophilic organisms and could also aid in the design of novel proteins with enhanced stability for industrial applications. The aim of this work was thus to obtain experimental information on the interactions present for a biological molecule in solution at elevated temperatures. Building on previous work, our molecule of choice is the amino acid Lglutamine (hereafter referred to as glutamine), a polar amino acid which contains a backbone (CO2CHNH3) and a side chain (CH2CH2CONH2).33 Glutamine is the most abundant, naturally occurring amino acid in the human body. Studies on proteins containing polyglutamine regions have pointed to the importance of glutamine−glutamine hydrogen bonding in driving protein association and subsequent aggregation.34−37 In the present study we use experimental and computational methods to allow the determination of a complete set of partial radial distribution functions for glutamine molecules in the liquid state. Neutron diffraction is a suitable probe for the structural study of aqueous glutamine because of the large scattering cross section of deuterium and the high contrast achievable using hydrogen/deuterium substitution on specific hydrogen sites in the molecule. The main goal of this work was to obtain structural data of glutamine in water at 24, 37, and 60 °C to allow determination of the association of glutamine with itself, at physiological temperatures. 15645

DOI: 10.1021/acs.jpcb.5b07413 J. Phys. Chem. B 2015, 119, 15644−15651

Article

The Journal of Physical Chemistry B

definition of a hydrogen bond in any of these cases was simply that a hydrogen bond was deemed to occur whenever one of the specified hydrogen atoms on one glutamine molecule was within 2.4 Å of the specified oxygen atom of another glutamine molecule. In other words, the same distance criterion was used for all potential hydrogen bonds, and the distance value was chosen to be close to the average position of the O−H first minimum in all four interactions. As it happens, none of these minor changes makes a significant difference to the primary conclusions of the previous paper.

simulation box is 8 times larger. Because the interest in the present work is to focus on the glutamine−glutamine interactions, these interactions would be sampled with greater efficiency in the much larger box size, while at the same time establishing whether box-size effects are present in these simulations. Hence, the simulations shown in this paper, while on the same data as used previously at 24 °C, are completely new. The EPSR simulation works by first equilibrating the ensemble of molecules using only the supplied potential parameters. This potential is called the reference potential, RP. Once this distribution is stable, an empirical potential, EP, is introduced as a perturbation to the RP, which attempts to drive the structure factors calculated from the simulation box as close as possible to the measured data. The amplitude of the EP is allowed to increase only to the point where the quality of fit, called the “R-factor”, which is actually the mean square deviation between model and data, ceases to get any smaller. Further significant increases in the amplitude of the EP beyond this point can lead to artifacts being generated in the simulation that have nothing to do with the real structure or data but are a result of Fourier transform truncation effects becoming prevalent in the EP. The R-factors obtained, in the region of 1.5 to 1.7 × 10−4 per data point, are better than reported previously, but they relate to the improved method of subtracting the inelastic scattering, rather than to any significant improvement in the structure refinement. The simulations were accumulated for some 16 000 individual molecular configurations, where each configuration is a result of 5 attempted translations and rotations of every molecule in the simulation box. Hence, in the present instance, the total number of attempted Monte Carlo moves is of order 1010. Alongside the structure refinement simulations, two other sets of simulations were run under identical conditions to the first, but either with the EP set to zero and all the effective atom charges set to zero, or with only the effective charges on the glutamine molecules set to zero and with the EP attempting to fit the data as before. The purpose of these additional simulations was to determine the effect of hydrogen bonding on the degree of glutamine clustering. The second set of simulations would have all the packing effects of the first set of simulations, but none of the bonding effects. The third set of simulations, with only the charges on the glutamine molecules set to zero, and with structure refinement as in the first set of simulations, demonstrates the effect to which charge−charge (hydrophilic) interactions between glutamine molecules affect the outcomes. The comparison of the three sets of simulations shows to what extent the glutamine clustering is driven by hydrogen bonding effects and to what extent it is driven by packing effects. In the previous work, glutamine−glutamine clusters were classified into three kinds, namely, backbone−backbone, backbone−side chain, and side chain−side chain, as represented by the four different kinds of hydrogen bonds between Ob and Hb, Ob and Hs, Os and Hb, and Os and Hs, respectively. Each interaction was determined from the position of the first minimum in the corresponding site−site radial distribution function. This method correctly identifies the individual types of interaction but will not include those cases where a cluster of three or more glutamines contains hydrogen bonds of the four different kinds. Therefore, in the present work, we counted clusters where any of the four types of hydrogen bonds could occur. In addition, for simplicity, the

3. RESULTS The EPSR fits to the scattering data at 37 °C are shown in Figure 1 as a function of the wave vector change, Q. Equivalent

Figure 1. EPSR fits (lines) to the measured neutron differential scattering cross sections (circles) for four glutamine−water solutions at a molar ratio of 1:269 and at (a) 24 °C, (b) 37 °C, and (c) 60 °C, each with a different combination of hydrogen/deuterium isotopes. Each data set is shifted upward by 1.0 barns/sr/atom (1 barn = 10−24 cm2).

fits were obtained at the other temperatures; therefore, these are not shown here. For the most part the fits are indistinguishable from the data, except at low Q where it is believed residual inelasticity effects in the data prove hard to remove completely. That they are residual inelasticity effects and not genuine structure can be identified by plotting the individual differential scattering cross sections obtained at different scattering angles (not shown here). In this region, broad peaks can be seen which occur at different Q values for different scattering angles. A genuine structural feature has to appear at the same Q value whatever the scattering angle. Calculations have shown that such broad features can arise from specific inelastic molecular excitations.42 The effect of substituting D for H is very apparent in these neutron data. The deuteron has a positive neutron scattering length (6.67 fm), while the proton has a negative scattering 15646

DOI: 10.1021/acs.jpcb.5b07413 J. Phys. Chem. B 2015, 119, 15644−15651

Article

The Journal of Physical Chemistry B

Figure 2. Site−site radial distribution functions for glutamine−glutamine hydrogen bond interactions (Og−Hg, see eq 1) at 24 °C (a), 37 °C (b), and 60 °C (c), and for the three simulations (shifted vertically for clarity), namely, full charges and full structure refinement (line, top), zero glutamine charges and full structure refinement (dots, middle), and no charges and no structure refinement (dashes, bottom). (d−f) Corresponding Cm−Cm radial distribution functions under the same conditions as in panels a−c. These latter distribution functions represent the hydrophobic interactions between glutamine molecules in solution.

length (−3.74 fm). This means the corresponding H-other correlations are negatively weighted in the H-containing samples compared to the D samples, while the other H−H and nonsubstituted correlations continue to make a positive contribution, resulting in markedly different scattering patterns with isotope. Due to the low concentration, the effect of isotope substitution on the glutamine is barely visible in the scattering data, while the effect on the water scattering is very marked. It was shown previously33 that the water structure found at this dilution is indistinguishable, within the uncertainties, with that found in pure water. The same work also discussed extensively the hydration of the amide and carboxylate groups in solution. Here we concentrate on the glutamine−glutamine interactions. To simplify the presentation we have combined the separate hydrogen bonding pairs, Ob−Hb, Ob−Hs, Os− Hb, and Os−Hs, into a single glutamine−glutamine correlation function between Og and Hg, where Og is either Ob or Os and Hg is either Hb or Hs:

gOgHg (r ) =

1 [6g (r ) + 4gObHs(r ) + 3gOsHb(r ) + 2gOsHs(r )] 15 ObHb (1)

where the weights derive from the relative contribution of each term to the total. In Figure 2a−c we show this radial distribution function for the three temperatures studied and for the three simulations, namely, with full charges and EP, with no charges on glutamine but still full structure refinement (nonzero EP), and with no charges on any atom and no EP (no structure refinement). It can be seen that each of these functions has a distinct and characteristic shape. This shape is apparently only weakly dependent on temperature. The first two well-defined peaks in all three cases are clear evidence of hydrogen bonding between atoms on nearby glutamine molecules. The complete absence of these peaks when the charges on the glutamine are removed but the EP is left intact tells us that the hydrogen bonding interactions between glutamine molecules in these simulations derive only from the assumed charges on the glutamine molecules and are not a result of the structure refinement, a consequence of the very low concentration of the glutamine in these data. 15647

DOI: 10.1021/acs.jpcb.5b07413 J. Phys. Chem. B 2015, 119, 15644−15651

Article

The Journal of Physical Chemistry B

shown are the cluster size distributions when there are no charges in the EPSR reference potential and the EP is set to zero. These latter curves signify the extent of clustering due to packing effects alone. It can be seen that in all cases the cluster size distribution drops away much more rapidly without Hbonding than when it is present, so confirming that most of the glutamine clustering identified here is due to hydrophilic association between the glutamine molecules. Similar results (not shown here) are found for when the glutamine charges are set to zero, but the EP is retained, allowing full structure refinement. It can be seen that with this definition of a glutamine− glutamine hydrogen bond, cluster sizes as large as 5 or 6 glutamine molecules can be found, albeit with very low probability. We note that the fraction of clusters containing just 2 molecules is around 10%, which is close to the value reported previously.33 The fraction of glutamine molecules that occur in clusters of 2 or greater is shown in Table 1 for the combined hydrogen bond condition, Og−Hg. There was a clear trend, as the simulation ran longer, for these values to draw closer together, so that within the uncertainties, and despite the fact some variations with temperature are apparent, the differences between values are probably not significant within the uncertainties. It can be readily seen that the fraction of glutamine molecules in clusters drops substantially by a factor of ∼6 or more when no hydrogen bonds are present between glutamine molecules, which suggests that most of the glutamine clustering is directly a result of hydrogen bonding between glutamine molecules. In addition, the cluster sizes were calculated when Cm−Cm interactions were included in the definition of bonded glutamines, shown in Table 1, corresponding to including hydrophobic clusters in the cluster calculation. In this case, two glutamines were considered bonded if respective Cm atoms on each molecule were less than 5 Å apart, this being considered the maximum that can be allowed for nonsolvent separated pairs. The increase in cluster size when this additional definition is included is marginal, suggesting that hydrophobic clusters, i.e., those that do not involve hydrogen bonds, are relatively rare in these solutions. To highlight the nature of the local order around a glutamine molecule in solution we have calculated the spatial density function of water around a central glutamine molecule in solution, Figure 4a. The same figure shows the most probable orientations of water molecules and other glutamine molecules (panels a and b in Figure 4, respectively) in the first coordination shell. From the spatial density function it is clear that water exists in bands around the central glutamine, both around the hydrophilic end groups and the central hydrophobic groups. Around the hydrophilic groups the water molecules are strongly oriented into either hydrogen bond donating or accepting positions. Around the hydrophobic

Also shown in Figure 2 are the hydrophobic Cm−Cm distributions between dissolved glutamine molecules. These are also seen to change very radically when the charges on the glutamine molecules are set to zero, with far less hydrophobic association when the charges are absent compared to when the charges are present. This highlights a point to be made later that any “hydrophobic” association between these molecules in solution is driven by the charged nature of the hydrophilic end groups and has little to do with the way water might enclathrate the hydrophobic parts of the molecule. Glutamine clusters were identified by means of the O···H intermolecular distance as described in Methods: any pair of glutamine molecules whose O···H distance was less than 2.4 Å was classified as bonded, from which clusters of glutamine molecules could be identified and counted. Figure 3 shows the

Figure 3. Fraction of glutamine clusters as a function of size at 24 °C (blue, solid), 37 °C (black, solid), and 60 °C (red, solid). The error bars show the root-mean-square deviation of individual configurations around the mean values. Also shown (dashed lines) are the cluster size distributions when the charges on glutamine are set to zero, but the empirical potential in the EPSR simulation continues to refine the structure, corresponding to the case when no hydrogen bonding between glutamines is possible. The fraction of clusters is on a logarithmic scale.

cluster size distribution for the case where all the possible O···H distances between glutamine molecules are included in the bond definition. This distribution represents the fraction of all the clusters (including those containing only one molecule) that contain the given number of glutamine molecules. Also

Table 1. Fractions of Glutamine Molecules in a Cluster of 2 or More Molecules as a Function of Temperaturea temperature [°C]

Og−Hg full charges and structure refinement

Og−Hg + Cm−Cm full charges and structure refinement

Og−Hg zero glutamine charges, full structure refinement

Og−Hg zero charges and no structure refinement

24 37 60

0.188 ± 0.027 0.166 ± 0.023 0.168 ± 0.029

0.205 ± 0.027 0.179 ± 0.027 0.181 ± 0.030

0.039 ± 0.016 0.018 ± 0.009 0.023 ± 0.010

0.038 ± 0.013 0.038 ± 0.013 0.031 ± 0.012

a A hydrogen bond is assumed to occur if the corresponding Og···Hg intermolecular distance was ≤2.4 Å. For clusters which included Cm−Cm interactions, the maximum allowed Cm−Cm distance was 5.0 Å. The error bars are the root-mean-square deviation of individual molecular configurations from the mean. Og corresponds to either the Ob or Os atoms, and Hg to either the Hb or Hs atoms, on the glutamine molecule.

15648

DOI: 10.1021/acs.jpcb.5b07413 J. Phys. Chem. B 2015, 119, 15644−15651

Article

The Journal of Physical Chemistry B

Figure 4. Based on a spherical harmonic reconstruction of the full glutamine−glutamine and glutamine−water orientational pair correlation functions,43 the most probable orientations of water molecules around a central glutamine molecule (a) and for other glutamines around glutamine in dilute solution (b). For glutamine−water interactions (a), the orientations are shown near the hydrogen-bonding groups and in two positions in the equatorial plane of the central molecule corresponding to hydrophobic interactions. Pronounced hydrogen bonding is observed for the water molecules near glutamine hydrophilic groups, while in the equatorial region primarily tangential orientations are observed as expected for a hydrophobic entity. The isosurface in this plot (yellow) shows the most probable positions of water molecules around glutamine, averaged over orientations of the water molecules. For the glutamine−glutamine plot (b), the orientations are shown in four directions, directly above and directly below, directly to the left and directly to the right (equatorial plane) of the central molecule. It is found that although hydrophobic associations (Cm−Cm) do occur in the equatorial region of the central molecule, these usually also involve hydrophilic associations at the backbone and sidechain ends of the molecules. Note that these plots are not the same as snapshots taken from the simulation box because they highlight the most probable orientations of neighboring molecules: individual molecules will vary markedly from these most probable orientations in practice.

simulation with realistic forces which are refined against the total scattering data. Because the latter are totally dominated by the water−water correlations, it has to be assumed that the observed glutamine−glutamine clustering derives only from the modeled hydrogen bond forces between these molecules. The only situation that could have changed this outcome, that is where the glutamine−glutamine clustering had a visible impact on the scattering data, would be if the glutamine molecules clustered in a globular fashion, causing a relatively large excluded volume hole in the water structure, which would have then shown up in the water−water correlations and would give rise to a distinct rise in the scattering at low Q, especially for those samples where D was substituted for H. The absence of a significant low Q rise in the scattering data for any of three temperatures studied is therefore clear evidence that globular clustering of the glutamines occurs only weakly, if at all, in these solutions. Nonetheless, our results do suggest that clustering via hydrogen bonds between glutamine molecules is a significant, albeit fleeting interaction. The clusters do not survive many Monte Carlo steps, but on average, approaching 20% of the glutamine molecules (Table 1) are involved in glutamine− glutamine clusters. At the same time there are very few, if any, clusters which involve hydrophobic functional groups being brought together at short-range without also an interaction at the hydrophilic ends of the molecule, Figure 4. This lends support to the notion that the primary interaction between amino acids in solution is via hydrophilic groups, not hydrophobic groups.27 In a simulation in which the hydrogen bond forces were switched off, Table 1 and Figure 3, the fraction of glutamine molecules in a cluster dropped

groups, the water molecules tend to arrange with their dipole moments tangential to the Cm···Ow axis, which is a classic symptom of hydrophobic hydration. For the glutamine− glutamine interactions, it is notable how, although hydrophobic association does occur, it is apparently driven by hydrophilic associations between the charged end groups. This is also seen in the gCmCm(r) radial distribution functions in Figure 2d−f, where a broad hump which is present near 6 Å in these functions when the headgroups are charged disappears when the charges are switched off.

4. DISCUSSION From the outset it must be understood that the results presented here for glutamine−glutamine clustering are a result of EPSR computer simulations which include realistic force fields to represent the likely hydrogen-bonding environment for these molecules as well as being refined against total neutron scattering differential cross section data (Figure 1). However, the weighting of the glutamine−glutamine correlations in the total scattering differential cross sections, which arises from the product of atomic fractions and neutron scattering lengths, is very small because of the low concentration, compared to the water−water and water-glutamine correlations. For example, the relative weighting of the Ob−Hb intermolecular correlation (the strongest of the glutamine−glutamine hydrogen bond correlations) in the total scattering differential cross section data, such as shown in Figure 1, is 6.7 × 10−6 compared to 1.2 × 10−3 and 1.6 × 10−1 for the Ob−Hw and Ow−Hw correlations, respectively. Hence, the scattering data can tell us nothing about the glutamine−glutamine correlation directly. The clustering information is obtained here from the computer 15649

DOI: 10.1021/acs.jpcb.5b07413 J. Phys. Chem. B 2015, 119, 15644−15651

Article

The Journal of Physical Chemistry B dramatically by a factor of ∼6 compared to when hydrogen bond forces were present. The present simulations used a revised protocol compared to33 which constrained the dihedral angles within the glutamine molecule to the values that emerged after the initial molecule structure refinement using MOPAC-7. This does not appear to have significantly affected any of the previous conclusions33 concerning the water structure and glutamine hydration, but means the glutamine molecules adopt structures in the simulation much closer to what might be expected based on first-principles calculations. The attempt here to observe the effect of temperature on this clustering was inconclusive. Due to low concentration, the sampling of the glutamine−glutamine correlations was poor even after several weeks of simulation, so that the glutamine− glutamine radial distribution functions shown in Figure 2 probably have not reached their fully sampled values. There is weak indication of a temperature effect, Figures 2 and 3 and Table 1, but within the uncertainties, any trends with temperature are probably not significant at this stage. Hence, the overall fraction of glutamine molecules in clusters remains almost invariant with temperature, within the uncertainties. Certainly there is no clear decline in this clustering with increasing temperature.

and water−solute correlations. In addition, despite running the simulations for several weeks on six parallel processors, it is questionable whether a full ensemble average of all possible glutamine−glutamine configurations has been achieved here. The radial distribution functions shown in Figure 2 show some small variations with temperature, but these are as likely to be due to sampling issues in the simulations as they are to genuine temperature effects in the real solution. Within these uncertainties the effect of temperature on the degree of glutamine clustering appears to be almost negligible in these solutions. Hence, the lessons learned here will be important when considering analogous studies at low solute concentration. What has been established beyond reasonable doubt is that glutamine molecules bond to each other primarily via hydrogen bonds and only weakly through hydrophobic interactions. The importance of hydrophilic interactions has previously been noted in a study of dipeptide association in solution.27 The present study supports this earlier finding and demonstrates the dominance of hydrophilic interactions in single amino acid association also. Furthermore, it expands upon the earlier study by demonstrating that hydrophilic interactions between glutamines persist at elevated temperatures. This finding is important because it highlights the necessity of considering hydrophilic interactions in the self-assembly and association of biological systems, and not just the more traditional hydrophobic interaction.

5. CONCLUSION The present account has attempted to study the effect of temperature on the degree to which glutamine molecules associate in dilute aqueous solution. It is found here that hydrophilic interactions are dominant in the association of glutamine molecules. At any instant, about 20% of the glutamine molecules are associated via intermolecular bonds and within the uncertainties caused by the low concentration. Furthermore, these hydrophilic interactions are persistent at increased temperature, with the fraction of molecules involved in clustering not changing appreciably over all the temperatures examined in this study. The majority of the clusters involve pairs of glutamine molecules, but occasionally clusters as large as 5 or 6 molecules form, independent of the temperature. At all the temperatures studied, the clusters appear to be shortlived. There is little evidence for larger-scale globular-type clustering of the glutamines under conditions of increased temperature, as has been observed in the case of tertiary butanol.44 Although the experimental conditions of the present work are the same as those of our previous study,33 the data interpretation has been enhanced by the use of dihedral angles on the glutamine molecules, to give a more rigid and reproducible molecular structure, and the use of a much larger simulation box with 8 times the number of molecules. However, these changes only partially alleviate the technical challenges caused by the low concentration of glutamine in these solutions. For example, the clusters are not observed directly in the experimental data, but are inferred from the simulation based on the likely hydrogen bonding in these solutions. Therefore, the results highlight the importance of incorporating realistic forces when attempting to refine the structure of even relatively uncomplicated systems such as single amino acids in solution. If the likely hydrogen bond forces between the glutamine molecules are not specifically modeled, only weak glutamine clustering is found, even with structure refinement (Figure 2 and Table 1), because the scattering data themselves are sensitive to only the water−water

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Web: http://www.isis.stfc.ac. uk/People/alan_soper5044.html. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS L.D. is supported by a grant from the European Research Council (258259-EXTREMEBIOPHYSICS). Beam time was allocated by the ISIS Facility (STFC, UK).

■

REFERENCES

(1) Kauzmann, W. Some Factors in the Interpretation of Protein Denaturation. Adv. Protein Chem. 1959, 14, 1−63. (2) Privalov, P. L.; Gill, S. J. Stability of Protein-Structure and Hydrophobic Interactions. Adv. Protein Chem. 1988, 39, 191−234. (3) van Meer, G.; Voelker, D. R.; Feigenson, G. W. Membrane Lipids: Where They Are and How They Behave. Nat. Rev. Mol. Cell Biol. 2008, 9, 112−124. (4) Frank, H. S.; Evans, M. W. Free Volume and Entropy in Condensed Systems, III Entropy in Binary Liquid Mixtures; Partial Molal Entropy in Dilute Solutions; Structure and Thermodynamics in Aqueous Electrolytes. J. Chem. Phys. 1945, 13, 507−532. (5) Bowron, D. T.; Finney, J. L.; Soper, A. K. Structural Investigation of Solute-Solute Interactions in Aqueous Solutions of Tertiary Butanol. J. Phys. Chem. B 1998, 102, 3551−3563. (6) Dixit, S.; Crain, J.; Poon, W. C.; Finney, J. L.; Soper, A. K. Molecular Segregation Observed in a Concentrated Alcohol-Water Solution. Nature 2002, 416, 829−832. (7) Dixit, S.; Soper, A. K.; Finney, J. L.; Crain, J. Water Structure and Solute Association in Dilute Aqueous Methanol. Europhys. Lett. 2002, 59, 377−383. (8) Dougan, L.; Bates, S. P.; Hargreaves, R.; Fox, J. P.; Crain, J.; Finney, J. L.; Reat, V.; Soper, A. K. Methanol-Water Solutions: A BiPercolating Liquid Mixture. J. Chem. Phys. 2004, 121, 6456−6462.

15650

DOI: 10.1021/acs.jpcb.5b07413 J. Phys. Chem. B 2015, 119, 15644−15651

Article

The Journal of Physical Chemistry B

(31) Hoffmann, T.; Tych, K. M.; Brockwell, D. J.; Dougan, L. SingleMolecule Force Spectroscopy Identifies a Small Cold Shock Protein as Being Mechanically Robust. J. Phys. Chem. B 2013, 117, 1819−1826. (32) Tych, K. M.; Hoffmann, T.; Brockwell, D. J.; Dougan, L. Single Molecule Force Spectroscopy Reveals the Temperature-Dependent Robustness and Malleability of a Hyperthermophilic Protein. Soft Matter 2013, 9, 9016−9025. (33) Rhys, N. H.; Soper, A. K.; Dougan, L. The Hydrogen-Bonding Ability of the Amino Acid Glutamine Revealed by Neutron Diffraction Experiments. J. Phys. Chem. B 2012, 116, 13308−13319. (34) Marchut, A. J.; Hall, C. K. Side-Chain Interactions Determine Amyloid Formation by Model Polyglutamine Peptides in Molecular Dynamics Simulations. Biophys. J. 2006, 90, 4574−4584. (35) Natalello, A.; Frana, A. M.; Relini, A.; Apicella, A.; Invernizzi, G.; Casari, C.; Gliozzi, A.; Doglia, S. M.; Tortora, P.; Regonesi, M. E. A Major Role for Side-Chain Polyglutamine Hydrogen Bonding in Irreversible Ataxin-3 Aggregation. PLoS One 2011, 6, e18789. (36) Plumley, J. A.; Dannenberg, J. J. The Importance of Hydrogen Bonding between the Glutamine Side Chains to the Formation of Amyloid VQIVYK Parallel Beta-Sheets: An ONIOM DFT/AM1 Study. J. Am. Chem. Soc. 2010, 132, 1758−1759. (37) Walters, R. H.; Murphy, R. M. Examining Polyglutamine Peptide Length: A Connection between Collapsed Conformations and Increased Aggregation. J. Mol. Biol. 2009, 393, 978−992. (38) Soper, A. K. The Radial Distribution Functions of Water as Derived from Radiation Total Scattering Experiments: Is There Anything We Can Say for Sure? ISRN Phys. Chem., 2013, Article ID 27963, http://dx/doi.org/10.1155/2013/279463. (39) Soper, A. K. Partial Structure Factors from Disordered Materials Diffraction Data: An Approach Using Empirical Potential Structure Refinement. Phys. Rev. B: Condens. Matter Mater. Phys. 2005, 72, 104204. (40) CRC Handbook of Chemistry and Physics, 57th ed.; CRC Press: Boca Raton, FL, 1976. (41) Stewart, J. MOPAC. http://openmopac.net/ (accessed October 29, 2014). (42) Soper, A. K. Inelasticity Corrections for Time-of-Flight and Fixed Wavelength Neutron Diffraction Experiments. Mol. Phys. 2009, 107, 1667−1684. (43) Gray, C. G.; Gubbins, K. E. Theory of Molecular Fluids; Clarendon Press: Oxford, 1984. (44) Finney, J. L.; Bowron, D. T.; Daniel, R. M.; Timmins, P. A.; Roberts, M. A. Molecular and Mesoscale Structures in Hydrophobically Driven Aqueous Solutions. Biophys. Chem. 2003, 105 (2−3), 391−409.

(9) Finney, J. L.; Bowron, D. T.; Soper, A. K. The Structure of Aqueous Solutions of Tertiary Butanol. J. Phys.: Condens. Matter 2000, 12, A123−A128. (10) Finney, J. L.; Bowron, D. T.; Soper, A. K.; Dixit, S. S. What Really Drives the Hydrophobic Interaction? Abstr. Pap. Am. Chem. Soc. 2002, 224, U317−U317. (11) McLain, S. E.; Soper, A. K.; Luzar, A. Investigations on the Structure of Dimethyl Sulfoxide and Acetone in Aqueous Solution. J. Chem. Phys. 2007, 127, 174515. (12) McLain, S. E.; Soper, A. K.; Watts, A. Water Structure around Dipeptides in Aqueous Solutions. Eur. Biophys. J. 2008, 37, 647−655. (13) Towey, J. J.; Soper, A. K.; Dougan, L. Preference for Isolated Water Molecules in a Concentrated Glycerol-Water Mixture. J. Phys. Chem. B 2011, 115, 7799−7807. (14) Towey, J. J.; Soper, A. K.; Dougan, L. Molecular Insight into the Hydrogen Bonding and Micro-Segregation of a Cryoprotectant Molecule. J. Phys. Chem. B 2012, 116, 13898−13904. (15) Towey, J. J.; Soper, A. K.; Dougan, L. What Happens To the Structure of Water In Cryoprotectant Solutions? Faraday Discuss. 2014, 167, 159−176. (16) Towey, J. J.; Dougan, L. Structural Examination of the Impact of Glycerol on Water Structure. J. Phys. Chem. B 2012, 116, 1633−1641. (17) Pagnotta, S. E.; McLain, S. E.; Soper, A. K.; Bruni, F.; Ricci, M. A. Water and Trehalose: How Much Do They Interact with Each Other? J. Phys. Chem. B 2010, 114, 4904−4908. (18) Mancinelli, R.; Botti, A.; Bruni, F.; Ricci, M. A.; Soper, A. K. Hydration of Sodium, Potassium, and Chloride Ions in Solution and The Concept Of Structure Maker/Breaker. J. Phys. Chem. B 2007, 111, 13570−13577. (19) Botti, A.; Bruni, F.; Imberti, S.; Ricci, M. A.; Soper, A. K. Ions in Water: The Microscopic Structure of Concentrated NaOH Solutions. J. Chem. Phys. 2004, 120, 10154−10162. (20) Mancinelli, R.; Bruni, F.; Ricci, M. A.; Imberti, S. Microscopic Structure of Water in a Water/Oil Emulsion. J. Chem. Phys. 2013, 138, 204503. (21) Bruni, F.; Imberti, S.; Mancinelli, R.; Ricci, M. A. Aqueous Solutions of Divalent Chlorides: Ions Hydration Shell and Water Structure. J. Chem. Phys. 2012, 136, 064520. (22) Busch, S.; Lorenz, C. D.; Taylor, J. W.; Pardo, L. C.; McLain, S. E. On the Short-Range Interactions of Concentrated Proline in Aqueous Solution. J. Phys. Chem. B 2014, 118, 14267−14277. (23) Bowron, D. T.; Soper, A. K.; Finney, J. L. Temperature Dependence of the Structure of a 0.06 Mole Fraction Tertiary Butanol-Water Solution. J. Chem. Phys. 2001, 114, 6203−6219. (24) Chou, S. G.; Soper, A. K.; Khodadadi, S.; Curtis, J. E.; Krueger, S.; Cicerone, M. T.; Fitch, A. N.; Shalaev, E. Y. Pronounced Microheterogeneity in a Sorbitol-Water Mixture Observed through Variable Temperature Neutron Scattering. J. Phys. Chem. B 2012, 116, 4439−4447. (25) Dougan, L.; Hargreaves, R.; Bates, S. P.; Finney, J. L.; Reat, V.; Soper, A. K.; Crain, J. Segregation in Aqueous Methanol Enhanced by Cooling and Compression. J. Chem. Phys. 2005, 122, 174514. (26) Sillren, P.; Swenson, J.; Mattsson, J.; Bowron, D.; Matic, A. The Temperature Dependent Structure of Liquid 1-Propanol as Studied by Neutron Diffraction and EPSR Simulations. J. Chem. Phys. 2013, 138, 214501. (27) McLain, S. E.; Soper, A. K.; Daidone, I.; Smith, J. C.; Watts, A. Charge-Based Interactions between Peptides Observed as the Dominant Force for Association in Aqueous Solution. Angew. Chem., Int. Ed. 2008, 47, 9059−9062. (28) Baldwin, R. L. Temperature Dependence of the Hydrophobic Interaction in Protein Folding. Proc. Natl. Acad. Sci. U. S. A. 1986, 83, 8069−8072. (29) Schellman, J. A. Temperature, Stability and the Hydrophobic Interaction. Biophys. J. 1997, 73, 2960−2964. (30) Rothschild, L. J.; Mancinelli, R. L. Life in Extreme Environments. Nature 2001, 409, 1092−1101. 15651

DOI: 10.1021/acs.jpcb.5b07413 J. Phys. Chem. B 2015, 119, 15644−15651