How to characterize stability of a protein from Trp fluorescence vs [denaturant] curves?

How to characterize stability of a protein from Trp fluorescence vs [denaturant] curves?

We are searching data for your request:

Forums and discussions:
Manuals and reference books:
Data from registers:
Wait the end of the search in all databases.
Upon completion, a link will appear to access the found materials.

A colleague of mine has taken Trp fluorescence measurements from a dimer in combination with various ligands, over a range of denaturant concentrations.

The idea is that ligands which bind more tightly to the dimer will stabilize it and delay changes in fluorescence. This is what our data look like:

Total fluorescence decreases, probably due to quenching from the denaturant. Papers I've seen using this method show it's possible to calculate the fraction of denatured protein by analyzing displacement in peak emission wavelength. However, they use monomeric proteins with few Trp, whereas I have a dimer with several Trp plus some additional variability introduced by the different ligands.

Additionally, we have data from a scintillation proximity assay that indicates ligand affinity is ordered as 1<… <5 and 6<… <11. Ideally, we would like to replicate this same ordering through the analysis conducted on the data above. I would appreciate any suggestions of methods that will allow us to carry out such comparisons.

Papers I've seen using this method show it's possible to calculate the fraction of denatured protein by analyzing displacement in peak emission wavelength.

Have you tried doing that with your data? What do you get if you plot peak emission wavelength as a function of denaturing agent concentration? If these look like sigmoidal curves (you might need a log scale on the X axis) with a clear upper plateau, then you can apply the method mentioned by these papers because the signal you observe is correlated to the fraction of denatured protein and saturates (this part is important: the signal must saturate past a certain concentration of denaturing agent, because all protein molecules are fully denatured and therefore adding more denaturing agent won't give further signal change; if your signal doesn't saturate, something is wrong).

However, they use monomeric proteins with few Trp, whereas I have a dimer with several Trp

Oligomeric state and number of Trp residues should not matter, because peak emission wavelength is only correlated to the fraction of denatured protein. Total intensity will increase with more Trp residues, but you're not using intensity as a measure of the fraction of denatured protein (and if anything, it's good to have more intensity to begin with given the quenching effect you observe with increasing denaturing agent concentration).

plus some additional variability introduced by the different ligands.

To take care of this concern, you need two controls that:

  1. prove that none of the ligands alters the fluorescence of your protein,
  2. prove that none of the ligands emits fluorescence between 300 and 450 nm, or if they do then you need to measure it to be able to subtract it from your total spectra, in order to recover the protein contribution to the total fluorescence.

For 1, record fluorescence emission spectra of the protein without denaturing agent but as a function of ligand concentration (span a few orders of magnitude) and see if the peak emission wavelength changes with ligand binding (it might change, which means the same assay can yield binding curves and Kd values for your ligands, which could also be useful!). Full ligand titration curves would be the most robust control, but you can simply use the same ligand concentrations as used in the denaturation experiments.

For 2, the simplest case would be that your ligands are not fluorescent, but you cannot simply assume that. Therefore, an important control you also need is to record emission spectra for each of your ligands at the concentrations used in the denaturation experiments, and this at each concentration of denaturing agent (or simply at 0 and max concentration: if there's no difference between these two spectra, you don't need to record one at every concentration in between). If the ligands emit fluorescence between 300 and 450 nm, you need to subtract that from your total spectra to recover the fluorescence arising from the protein only.

Chapter Six - Quantum protein folding ☆

To establish a fundamental framework for understanding biomolecules from the first principle of quantum mechanics, the manuscript reviews the work of authors’ group on protein and RNA folding to demonstrate the existence of a common quantum mechanism in the conformational transition of biomolecules. Based on the general equation of the conformation-transitional rate several theoretical results are deduced and compared with experimental data through bioinformatics methods. The main results we obtained are: The temperature dependence and the denaturant concentration dependence of the protein folding rate are deduced and compared with experimental data. The quantitative relation between protein folding rate and torsional mode number (or chain length) is deduced and the obtained formula can be applied to RNA folding as well. The quantum transition theory of two-state protein is successfully generalized to multistate protein folding. Then, how to make direct experimental tests on the quantum property of the conformational transition of biomolecule is discussed, which includes the study of protein photo-folding and the observation of the fluctuation of the fluorescence intensity emitted from the protein folding/unfolding event. The above results show that the quantum mechanics provides a unifying and logically simple theoretical starting point in studying the conformational change of biological macromolecules. The far-reaching results in practical application of the theory are expected.

Protein Folding and Stability Using Denaturants

Measurements of protein folding and thermodynamic stability provide insight into the forces and energetics that determine structure, and can inform on protein domain organization, interdomain interactions, and effects of mutations on structure. This chapter describes methods, theory, and data analysis for the most accessible means to determine the thermodynamics of protein folding: chemical denaturation. Topics include overall features of the folding reaction, advances in instrumentation, optimization of reagent purity, mechanistic models for analysis, and statistical and structural interpretation of fitted thermodynamic parameters. Examples in which stability measurements have provided insight into structure and function will be taken from studies in the author's laboratory on the Notch signaling pathway. It is hoped that this chapter will enable molecular, cell, and structural biologists to make precise measurements of protein stability, and will also provide a strong foundation for biophysics students who wish to undertake experimental studies of protein folding.


Effects of mutations on the conformations of apoCopC

As shown in Figure 1 , folded CopC (apoCopC and various mutants) exhibits a negative CD signal around 210 to 220 nm, which is a characteristic of the β-sheet structure. The far-UV CD spectra of these five proteins are similar, suggesting that the four mutations did not give rise to any significant changes in the secondary structures. Although the protein structure did not undergo major changes, the subtle changes should be analyzed. The data from CD revealed that the β-sheet content (Y79F, Y79W) was reduced by 11 and 8% compared with apoCopC, respectively [ Fig. 1 (A)]. Similarly, the β-sheet content for Y79WW83L exhibited a loss of 5%, whereas that for Y79WW83F increased by 2% compared with Y79W, as shown in Figure 1 (B).

Far-UV CD spectra of protein (25 μM) (A: apoCopC, Y79F, and Y79W) (B: Y79W, Y79WW83L, and Y79WW83F) using 1-mm path length quartz cells in 2 mM Hepes, pH 7.4, at 25ଌ. [Color figure can be viewed in the online issue, which is available at]

Emission fluorescence spectra

To assess the effect of the mutation on the fluorescence characteristics of the protein, we analyzed their emission spectra using fluorescence spectroscopy. Mutant proteins showed a subtle shift in emission wavelength, suggesting that the polarity of the microenvironment around the Trp changed a little, but were still hydrophobic. As shown in Figure 2 (A), the emission maximum of Y79W was 325 nm and that of Y79F was 320 nm. For Y79WW83L and Y79WW83F, as shown in Figure 2 (B), the emission maximums were 331 and 328 nm, respectively. For apoCopC and Y79F, the only Trp was located at the 83 position. However, for Y79WW83L and Y79WW83F, the only Trp was located at the 79 position. The fluorescence parameters of Trp residues are sensitive to the environment (environment-sensitive fluorophores). The emission maximum of the Trp fluorescence spectrum depends on the properties of the environment around Trp residues in proteins. The explanation was that the microenvironment around Trp83 was different from that around Tyr79.

The emission fluorescence spectra (A: apoCopC (1), Y79F (2) and Y79W (3)). (B: apoCopC (1), Y79WW83F (2) and Y79WW83L (3)), excited at 295 nm in 10 mM Hepes, pH 7.4, at 25ଌ.

Effects of mutations on the binding capacity of Cu(II)

The apoCopC protein is a small soluble molecule (10.5 kDa) with a β-barrel structure it features two distinct copper-binding sites that are highly specific for Cu(I) and Cu(II). To explore the effects of mutations on the protein function, an experiment on the proteins combined with copper was carried out. Titration of proteins with Cu(II) quenched the fluorescence intensity linearly until 1 equiv of Cu(II) was bound, as shown in Figure 3 (A). The fluorescence intensity decreased by 60% for Y79F and apoCopC. However, the intensity decreased by 40% for Y79WW83L and Y79WW83F, suggesting that the fluorescence was quenched by Cu(II) in a large degree when the Trp in the protein was located at the 83 position. This result may be due to the fact that copper-binding sites located at the N-terminal and Trp83 were closer to the copper-binding sites than Trp79. Job's plot is shown in Figure 3 (B), in which the fluorescence intensity at 320 nm decreased linearly in two slopes, and the stoichiometry of Cu(II) binding with Y79F was 1:1. When [Cu(II)]/([Cu(II)]+[Y79F]) π.5, fluorescence intensity was observed from the free Y79F and Cu(II)-Y79F. However, when [Cu(II)]/([Cu(II)]+[Y79F]) 𢙐.5, the fluorescence intensity was only observed from Cu(II)-Y79F, and the content of Cu(II)-Y79F decreased with the reduction in the concentration of Y79F.

A: The fluorescence spectra of Y79F (50 µM) at different concentrations of Cu(II) in 10 mM Hepes, pH 7.4, at 25ଌ. The concentration of Cu 2+ from a to i is 0, 7, 14, 21, 28, 35, 42, 49, 60 µM, respectively. B: Job's plot of fluorescence intensity at 320 nm against [Cu(II)]/([Cu(II)]+[Y79F]) in 10 mM Hepes, pH 7.4, at 25ଌ. The sum concentration of Y79F and Cu(II) is 2.0 × 10 𢄦 M and [Cu(II)]/([Cu(II)]+[Y79F]) is 0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, and 1.0, respectively. C: The titration curve of Y79F by Cu(II) (a) or Cu(II)-EDTA (1:0.5) (b).

To calculate the binding constants between Cu(II) and mutants, we used EDTA as a competitive ligand. The fluorescence intensity (F) values at 320 nm were derived from the titration of Y79F with EDTA-Cu(II), and the titration curves were prepared by plotting F versus [Cu(II)] [ Fig. 3 (C)]. Assuming that the decrease in fluorescence intensity at given [Cu(II)] is attributed to the change of Cu(II)-Y79F to Cu(II)-EDTA, the binding constant can be calculated with Formulas (1) to (7) (See Supporting Information). The constants of the other mutants can be obtained through the same method. The binding constants of Cu(II) and proteins are listed in Table I . The data showed that the binding ability of Cu(II) was not affected by mutations.

Table I

The Binding Constants of Cu(II) and Proteins

ProteinK (M 𢄡 )R 2
apoCopC(2.24 ± 0.36) × 10 14 0.97
Y79F(0.19 ± 0.04) × 10 14 0.94
Y79W(1.53 ± 0.16) × 10 14 0.98
Y79WW83L(0.31 ± 0.03) × 10 14 0.98
Y79WW83F(0.28 ± 0.42) × 10 14 0.96

Fluorescence lifetimes of apoCopC and its mutants

Investigating the intrinsic fluorescence from protein is an effective method to study protein conformational dynamics. CopC has a single Trp residue, so it has been studied using a single Trp residue as an intrinsic fluorophore. The indole double ring has a dual physical nature. This ring has a great hydrophobic van der Waals surface area, but possesses a highly polar NH group. For the 𠇎xposed” tryptophyls, the most preferable situation is that where indole is partly buried by its benzol ring and interacts with water through its NH group. According to the model proposed by Burstein, 27 the fluorescence properties of the Trp residues in the proteins suggest the existence of three discrete spectral classes: one buried in the nonpolar regions of the protein (λem = 330� nm, τ = 2.1 ns) and two on the surface. One of the latter is completely exposed to water (λem = 350� nm, τ = 5.4 ns), whereas the other is in limited contact with water, which is probably immobilized by bonding at the macromolecular surface (λem = 340� nm, τ = 4.4 ns). Fluorescence decays of CopC and its mutants in 10 mM Hepes, pH 7.4, at 25ଌ are shown in Figure 4 . Figure 4 (A) shows the decays of apoCopC, Y79F, and Y79W, wherein their lifetimes were different, although the fluorescence spectrum for Y79F was the same as that for apoCopC. Similarly, Figure 4 (B) shows the decays of apoCopC, Y79WW83L, and Y79WW83F. The lifetimes of Y79WW83L and Y79WW83F were larger than that of apoCopC. For apoCopC, the single Trp was located at the 83 position, whereas the single Trp for Y79WW83L and Y79WW83F was located at the 79 position. In other words, the lifetimes of Trp in proteins increase in a manner dependent on the position of the Trp residue. From the lifetime values listed in Table II , we posit that the exposed degree of Tyr79 in a hydrophilic environment is more enhanced compared with that of Trp83. The results coincided with the data from the fluorescence emission spectra. Y79W has two lifetimes, suggesting that the environments around these two Trp residues are different. One residue was located in a hydrophobic environment, whereas the other was located at a relatively hydrophilic environment.

Table II

The Fluorescence Parameters of apoCopC and Its Mutants

ProteinAmplitude (Bi)Lifetime, τi (ns)χ 2
Y79W0.040, 0.0431.67, 4.651.060

Fluorescence decays of [A: apoCopC (a), Y79F (b), and Y79W (c)]. [B: ApoCopC (a), Y79WW83F (b), and Y79WW83L (c)] in 10 mM Hepes, pH 7.4, at 25ଌ.

Acrylamide and KI quenching

Fluorescence quenchers are widely used in studying the relative accessibilities of fluorescent groups in proteins. KI and acrylamide are the universal quenching agents. Acrylamide quenching is very sensitive to the degree of Trp accessibility to the solvent containing the acrylamide. Given that acrylamide can diffuse into the interior of the protein, accessibility to acrylamide may result from the existence of channels leading to the interior of the protein. Iodide ion is hydrated and is limited by its large size and charge to quenching of Trp residues lying at or near the surface of proteins. 28 Quenching of the protein Trp fluorescence by KI involves a collision-quenching mechanism. 29 Eftink and Ghiron 30 reported that KSV derived from the slope of the Stern–Volmer plot, can, in effect, be taken as a crude estimation of the accessibility of Trp residues in proteins. Stern–Volmer plots for the KI quenching of proteins fluorescence, with fluorescence emission monitored at 320 nm, are shown in Figure 5 (A). Parallel quenching experiments were performed with acrylamide, a polar, uncharged water-soluble molecule that can penetrate a protein matrix as a function of protein size and dynamics. 31 The linear Stern–Volmer plots of F0/F versus acrylamide concentration are shown in Figure 5 (B). The quenching parameter analyses obtained from the Stern–Volmer plots are presented in Table III . The data list in Table III reveals that the degree quenched by acrylamide is significantly greater than that of KI. This phenomenon suggests that Trp is buried in the protein deeply, and acrylamide, a neutral quencher, can access Trp easily. Furthermore, the quenched degree for the Trp located at 79 was larger than that located at 83, which is in agreement with the conclusion derived from KI.

Stern–Volmer plots of iodide (A) and acrylamide (B) quenching of apoCopC (х), Y79W (☆), Y79F (•), Y79WW83F (□), and Y79WW83L (▽) in 10 mM Hepes, pH 7.4, at 25ଌ.

Table III

Stern–Volmer Constants of apoCopC and Its Mutants Using Different Quenchers

KSV (M 𢄡 )
apoCopC12.45 ± 0.363.62 ± 0.24
Y79F15.91 ± 0.196.12 ± 0.28
Y79W20.83 ± 0.349.15 ± 0.20
Y79WW83F15.61 ± 0.366.54 ± 0.12
Y79WW83L18.53 ± 0.297.31 ± 0.27

GdnHCl/urea-induced unfolding of apoCopC and its mutants

A denaturant refers to a reagent that decreases the stability of protein and leads to a structural change from its specific, compact, and three-dimensional structure to an unfolded state. From the structural point of view, urea and GdnHCl are very similar but the latter is a positive ion. Therefore, we can surmise that GdnHCl, giving rise to charged species in water, could effectively interfere with favorable electrostatic interactions among the charged groups on the protein surface. The molecular mechanisms of GdnHCl and urea denaturation are different 32 – 36 and the screening of favorable electrostatic interactions on the surface of globular proteins may be the only real difference between the two denaturants.

The thermodynamic stability of apoCopC and its mutants was tested in GdnHCl/urea-induced unfolding experiments in 10 mM Hepes, pH 7.4, at 25ଌ. In proteins, Trp is highly sensitive to the polarity of its surrounding environment. Native apoCopC exhibited an maximum emission at 320 nm. In 3M GdnHCl, the spectra remained similar in shape, but the emission maximum peak shifted from 320 to 350 nm, which corresponds to the fluorescence maximum of Trp in an aqueous solution. The fluorescence spectra of Y79F under different concentrations of GdnHCl are shown in Supporting Information Figure S1. To best assess the shape of the apoCopC transitions, a ratio of fluorescence intensities at 400 and 310 nm (I400 nm/I310 nm) was plotted as a function of denaturant concentration. At low denaturant concentration, the unfolding fraction was 0, indicating that the protein was in the fold state. The unfolding fraction increased with the further denaturant accumulation. When the concentration reached a certain value, the unfolding fraction was 1 and the protein unfolded completely. The unfolding curves of the mutants were obtained through the same method. As shown in Figure 6 (A), the unfolding curves of Y79F, Y79W, and apoCopC were similar, and the unfolding transitions best fitted to a two-state model, the unfolding–transition midpoints for Y79F and Y79W shifted to lower GdnHCl concentrations. The data listed in Table V show that the stabilities of Y79F and Y79W decreased, and Y79F was less stable compared with Y79W. The corresponding urea-unfolding experiments are shown in Figure 6 (B). Similarly, the unfolding of Y79F and apoCopC was the same as above, confirming the two-state processes. Notably, the unfolding of the Y79W mutant induced by urea best fitted to a three-state model. A tiny plateau was present in the transition region from 6 to 6.2M urea, suggesting greater population of the partially folded intermediate within the scope of urea concentration. Equilibrium unfolding data obeyed a three state model with midpoints of 4.6 and 7.1M urea for the first and second transitions, respectively. According to the structure element model, the unfolding curve of Y79W can be fitted by the combination of Y 1 and Y 2 (see analysis of denaturation data in Supporting Information Fig. S2). The thermodynamics parameters of the unfolding of proteins induced by urea and GdnHCl are listed in Tables IV and ​ andV, V , respectively.

GdnHCl (A and C) or urea (B and D) induced unfolding of apoCopC and its mutants. Denaturation of protein was monitored by fluorescence spectroscopy. Samples of a fixed concentration of protein (2 μM) were titrated with GdnHCl from 0 to 3M and were allowed to unfold for 30 min at 25ଌ before the measurement of fluorescence intensity. The unfolding experiment induced by urea was carried under the same condition.

Table IV

Thermodynamics Parameters of apoCopC and Its Mutants from Urea-Induced Unfolding at pH 7.4, 25ଌ

ProteinParametersF (I/U)I (U)G 0 element> (kJ/mol)Δ<ΔG 0 element> (kJ/mol)
apoCopC[D]1/2 (M)5.68
m (kJ/M/mol)4.36 ± 0.05 24.350
ΔG 0 i (kJ/mol)24.35 ± 0.35
Y79XY79F[D]1/2 (M)4.13
m (kJ/M/mol)3.80 ± 0.02 17.84𢄦.51
ΔG 0 i (kJ/mol)17.84 ± 0.82
Y79W[D]1/2 (M)4.67.1
m (kJ/M/mol)4.08 ± 0.046.78 ± 0.1322.32𢄢.03
ΔG 0 i (kJ/mol)18.86 ± 0.1848.61 ± 0.08
Y79W[D]1/2 (M)4.67.1
m (kJ/M/mol)4.08 ± 0.046.78 ± 0.1322.320
ΔG 0 i (kJ/mol)18.86 ± 0.1848.61 ± 0.08
W83LY79WW83L[D]1/2 (M)3.00
m (kJ/M/mol)3.62 ± 0.06 10.88�.44
ΔG 0 i (kJ/mol)10.88 ± 0.21
Y79WW83F[D]1/2 (M)4.27.2
m (kJ/M/mol)4.88 ± 0.024.76 ± 0.0324.141.82
ΔG 0 i (kJ/mol)20.55 ± 0.1033.70 ± 0.18

Table V

Thermodynamics Parameters of apoCopC and Its Mutants from GdnHCl-Induced Unfolding at pH 7.4, 25ଌ

ProteinParametersF (I/U)I (U)G 0 element> (kJ/mol)Δ<ΔG 0 element> (kJ/mol)
apoCopC[D]1/2 (M)1.55
m (kJ/M/mol)16.16 ± 0.22 24.980
ΔG 0 i (kJ/mol)24.98 ± 0.31
Y79XY79F[D]1/2 (M)0.99
m (kJ/M/mol)17.30 ± 0.39 17.04𢄧.94
ΔG 0 i (kJ/mol)17.04 ± 0.38
Y79W[D]1/2 (M)1.33
m (kJ /M/mol)15.13 ± 0.11 20.16𢄤.82
ΔG 0 i (kJ/mol)20.16 ± 0.13
Y79W[D]1/2 (M)1.33
m (kJ/M/mol)15.13 ± 0.11 20.160
ΔG 0 i (kJ/mol)20.16 ± 0.13
m (kJ/M/mol)16.92 ± 1.19 10.75𢄩.41
ΔG 0 i (kJ/mol)10.75 ± 0.74
Y79WW83F[D]1/2 (M)1.161.54
m (kJ/M/mol)17.79 ± 0.2217.04 ± 0.7323.583.42
ΔG 0 i (kJ/mol)20.65 ± 0.2326.27 ± 1.32

Figure 6 (C,D) was obtained using the methods of Figure 6 (A,B). By contrast, the unfolding curve of Y79W was added in Figure 6 (C,D), in which the unfolding of Y79WW83L best fitted to the two-state model induced by urea or GdnHCl. The two state fit unfolding curves yielded midpoints of 0.64M GdnHCl and 3M urea for Y79WW83L [ Fig. 6 (C,D)]. The midpoint of transition for Y79WW83L was much less than that for Y79W. Interestingly, the unfolding experiments with Y79WW83F revealed some differences. The unfolding curves obtained for Y79WW83F seemed to involve two sequential transitions (see analysis of denaturation data in Supporting Information Figs. S3 and S4). This result indicates that the equilibrium–unfolding reactions have three states. The GdnHCl-induced unfolding curve of Y79WW83F showed that the midpoint of the first transition was 1.16M, whereas that of the second transition was 1.54M GdnHCl [ Fig. 6 (C)]. Similar to GdnHCl, the unfolding transition induced by urea had three states, with a transition midpoint of 4.2M urea for the transition from native to intermediate and a midpoint of 7.2M urea for the intermediate to the unfolded transition [ Fig. 6 (D)]. The thermodynamics parameters of the unfolding of proteins induced by urea and GdnHCl are listed in Tables IV and ​ andV, V , respectively.

Figure 6 and Tables IV and ​ andV V show that the stabilities of apoCopC, Y79W, and Y79F followed the series apoCopC>Y79W>Y79F. The difference among these three proteins was only the 79th amine acid. The abilities of the formation of hydrogen bonds decreased successively, which revealed that the ability of the 79th amine acid to form H-bonds has an important role in maintaining the structure of apoCopC. The stabilities of Y79W, Y79WW83L, and Y79WW83F followed the series Y79WW83F>Y 79W>Y79WW83L. The replacement of Trp83 with Phe and Leu induced a completely opposite effect, indicating that the aromatic ring of Trp83 was important in maintaining the hydrophobic core of apoCopC.


To increase the internal void volume of a protein (Fig. 1, Top, Left) (10, 11), 10 variants of a very stable variant of SNase referred to as Δ + PHS (11.8 kcal/mole at 298 K (10), compared to 5.4 kcal/mol for WT SNase), were engineered with substitutions of internal positions to Ala to create additional internal cavities.

Structures of Δ + PHS SNase and cavity-containing variants. Left Panel, from Top to Bottom: Structure of Δ + PHS SNase (3BDC) with the C α positions of the 10 cavity-containing variants indicated with red spheres and with the surface representation of the central cavity in purple. Cavity volume was calculated using a 1.1 Å sphere and McVol (11). Structures of the cavity mutants L125A (3NXW), I92A (3MEH) and V66A (3NQT) with the engineered cavities in green and the mutated residue in red. Right Panel: Estimation of the void density (Top) and hydration density (Bottom) for each C α position of the Δ + PHS reference protein. See SI Materials and Methods in SI Appendix and (9) for the details concerning calculation of void and hydration density. Briefly, the 1,000 configurations resulting from a 10 ns all-atom MD simulation in explicit solvent were submitted to Monte Carlo point oversampling. All points that fell within the structure and not on an atom of solvent or the protein were counted for void density. Water density was calculated as the number of oxygen atoms of water molecules within 5 Å of each C α carbon, normalized to the largest number found for a C α carbon.

Crystal Structures.

The crystal structures of all variants, solved at 1 atm, showed that the substitutions led to the creation of internal cavities in the folded state. The structures were very similar among themselves and to that of Δ + PHS (Fig. 1 and Fig. S1 and Table S1 in SI Appendix), the main difference being either the enlargement of the naturally occurring microcavity or the introduction of a new cavity. No internal water molecules were observed in any of the cavities not even a trace of electron density was found.

Molecular Dynamics Simulations.

Because cavities can be hydrated transiently or filled through dynamic side-chain reorganization, the persistence of the naturally occurring cavity in the Δ + PHS reference protein was investigated using molecular dynamics (MD) simulations in explicit solvent (SI Materials and Methods in SI Appendix). Void and hydration density at each C α (Fig. 1, Right), defined as the number of Monte Carlo points (normalized to the maximum for all C α atoms) that could be inserted within 5 Å of its position and the number of water oxygen atoms (normalized to the maximum number) within 5 Å, respectively, support the notion that even the naturally occurring microcavity observed in the structure of the wild type protein is present and dehydrated, even after dynamic reorganization and transient hydration are taken into consideration.

Equilibrium Unfolding Monitored by Trp Fluorescence.

Pressure unfolding monitored by the fluorescence of Trp-140 (12) (Fig. 2A and Fig. S2 in SI Appendix) revealed that the variants with the engineered cavities all exhibited significantly larger values for the apparent volume change upon unfolding ΔVu( = -ΔVf) than did the Δ + PHS protein (Fig. 2 A and B and Table 1). The midpoints of the unfolding profiles shifted to lower pressures with increasing GuHCl, but the slopes were unaffected, indicating that the denaturant destabilized the protein without any effect on ΔVu.

Pressure unfolding monitored with Trp fluorescence. (A) High-pressure fluorescence average emission wavelength profile for (Left Panel) Δ + PHS variant at 2.0 M (circle), 2.3 M (triangle) and 2.6 M (square) GuHCl and (Right Panel) I92A cavity mutant at 0.8 M (circle), 1.0 M (triangle) and 1.2 M (square) GuHCl. (B) Folding volume change, ΔVf( = -ΔVu) values obtained from analysis of the high-pressure fluorescence unfolding profiles for Δ + PHS and 10 cavity-containing variants assuming a two-state model. The dashed line indicates the value for the Δ + PHS protein used as reference. Measurements were performed in the presence of guanidinium chloride (GuHCl) to ensure complete unfolding in the pressure range of the instrumentation used (< 3 kbar).


Far-UV CD spectra and secondary structure analysis

Far-UV CD spectra of MixH and PrgI (Fig. 2 ​ 2) ) show that they have a similar secondary structure with high helical content. Deconvolution and analysis of the spectra using Dichroweb (Whitmore and Wallace 2004) suggests that that PrgI contains more helix than MxiH (Table 1 ​ 1). ). At 10ଌ, the recombinantly expressed MxiH has a little 㹐% α-helical content and

20% β-sheet, whereas PrgI has almost 70%α-helix and

10%β-sheet. Additionally, MxiH appears to have more random and turn structure than its Salmonella homolog. At temperatures as low as 25ଌ, both proteins seem to lose a significant amount of secondary structure, especially with respect to helical content (Table 1 ​ 1). ). This indicates that both proteins may be relatively unstable (see below).

Table 1.

Summary of results from the deconvoluted far-UV CD spectra into various components of secondary structure in MxiH and PrgI

Reported values are percentages of the total structure with an estimated uncertainty of 2.3%.

Far-UV CD spectra for MxiH and PrgI at 10ଌ at a protein concentration of 50 μM using a 0.01-cmpath-length cell. The circles show the spectrum for MxiH, and the triangles show the spectrum of PrgI. The data shown are an average of greater than or equal to two trials.

Thermally induced unfolding as monitored by CD and second-derivative UV absorbance spectroscopy

To examine the relative stability of MxiH and PrgI, both proteins were examined as a function of temperature. Each demonstrates 㺐% thermal unfolding reversibility after heating to 90ଌ as monitored by CD (Fig. 3 ​ 3) ) and UV absorbance spectroscopy (Figs. 4 ​ 4, , 5 ​ 5). ). Analysis of the CD unfolding curves in Figure 3 ​ 3 yield an approximate Tm of 42ଌ and 37.5ଌ for MxiH and PrgI, respectively. Correspondingly, nearly indistinguishable ΔH values of 22.3 and 23.9 kcal/mol are seen.

Thermal unfolding of MxiH (A) and PrgI (B) as monitored by CD. The solid black traces represent unfolding, and the dashed lines, the unfolding of proteins previously heated to 㺐ଌ and then cooled.

MxiH thermal unfolding as monitored by UV absorbance spectroscopy. Data are derived from second derivatives of the UV spectra as a function of temperature. Peaks 1, 2, and 3 represent contributions from Phe peaks 3, 4, and 5 from Tyr and peaks 5 and 6 from Trp. (Peaks 3 and 5 are combination peaks.) The line in the data for peaks 4 and 5 is the fit of a two-state unfolding model to the data.

PrgI thermal unfolding as monitored by UV absorbance. Data are derived from second derivatives of the UV spectra at various temperatures as described in the legend to Figure 4 ​ 4 .

To examine the effects of thermal stress on the tertiary structure, high-resolution second-derivative UV absorption spectroscopy was employed. This technique has the advantage that the three different types of aromatic residues tend to be dispersed nonuniformly throughout most protein structures, thereby providing a somewhat different view of protein behavior than intrinsic fluorescence spectroscopy, which in this case relies upon changes in a single Trp residue (see below). The high-resolution second-derivative spectra of the needle proteins consist of six peaks (not illustrated). Peaks 1, 2, and 3 represent contributions from Phe 4 and 5, from Tyr and 5 and 6, from Trp (peak 5 is a combination peak) (Figs. 4 ​ 4, , 5 ​ 5). ). The three Phe and the highest wavelength of the Trp peaks show little evidence for well-defined thermal transitions for both proteins. In contrast, Tyr peaks manifest typical thermal unfolding curves. Assuming a two-state unfolding model, the Tyr profiles yielded Tm and corresponding ΔH values that are similar, within error limits, to those observed in the CD studies (Table 2 ​ 2 ).

Table 2.

Summary of the thermodynamic parameters extracted from the analysis of the thermal induced unfolding curves as monitored by CD and second-derivative UV absorbance spectroscopy

MxiH42 ± 0.122.3 ± 2.148.0 ± 2.342.7 ± 2.328.8 ± 3.024.9 ± 1.9
PrgI37.5 ± 0.523.9 ± 0.436.0 ± 3.435.7 ± 4.822.0 ± 7.626.0 ± 1.1

All Tm values reported are in units of ଌ, and the enthalpy values, in kcal/mol. The values shown are ± standard error with n = 2 or 3.

The optical density of the solution was also monitored at 350 nm as a function of temperature to search for macroscopic aggregation. No detectable increase in the optical density was seen as the temperature was increased (not shown). Thus, the thermal transitions observed do not result in detectable aggregation, consistent with their reversibility.

Urea-induced unfolding as monitored by CD

To further access the stability of these proteins, urea unfolding studies were performed using the intrinsic CD signal of the proteins at 222 nm to monitor changes in secondary structure as a function of urea concentration. The unfolding curves, as seen in Figure 6 ​ 6, , show some degree of cooperativity (10ଌ) (Fig. 6A1,A2 ​ 6A1,A2), ), which diminishes with increasing temperature (25ଌ) (Fig. 6B1,B2 ​ 6B1,B2). ). The hyperbolic nature of the unfolding curves at 25ଌ, as opposed to the more sigmoidal curve observed in the lower temperature experiments, indicates a loss in cooperativity. From the urea unfolding data, the intrinsic free energy of unfolding (ΔG°0,un) and their dependence on denaturant concentration (m-values) of the proteins were calculated using a nonlinear least-square analyses for a two-state unfolding transition. These results are reported in Table 3 ​ 3. . The results show that MxiH is slightly more stable than PrgI and also has a higher m-value, indicating that more surface area of MxiH is exposed during unfolding than during PrgI unfolding.

Table 3.

Summary of results from a two-state analysis of the urea-induced unfolding of MxiH and PrgI at 10ଌ and 25ଌ

ProteinΔG°0,un (kcal/mol 1− )m (kcal/mol 𢄡 M 𢄡 )ΔG°0,un (kcal/mol 𢄡 )m (kcal/mol 𢄡 M 𢄡 )
MxiH1.62 ± 0.010.82 ± 0.020.89 ± 0.160.69 ± 0.00
PrgI1.19 ± 0.030.77 ± 0.010.44 ± 0.020.62 ± 0.02

The errors reported are standard errors (with n=2). The free energy of unfolding (ΔG°0,un) is the extrapolated free energy of unfolding back to zero denaturant concentration.

Urea-induced unfolding of MxiH and PrgI as monitored by CD at 222 nm. Panels A1 and A2 show MxiH, and panels B1 and B2 show PrgI. The upper panels (A1,B1) show data acquired at 10ଌ, and the lower panels (A2,B2) show data acquired at 25ଌ.

Intrinsic tryptophan fluorescence

The fluorescence emission spectra of MxiH and PrgI (Fig. 7 ​ 7) ) show only very subtle differences with λmax values of 349.5 and 347.5 nm, respectively. These values indicate extensive exposure of the proteins’s single Trp residue to the solvent. This is consistent with the position of the 290 nm absorption peak (peak 6 in Figs. 4 ​ 4, , 5 ​ 5). ). Both techniques suggest that the MxiH indole is slightly more exposed than the corresponding residue in PrgI.

Typical normalized fluorescence emission spectra of MxiH and PrgI (50 μM) in PBS at pH 7.0. Samples were excited at 295 nm and the emission monitored from 302 nm to 450 nm. PrgI is shown with the solid and MxiH is shown with a dashed line.

Urea-induced unfolding monitored by Trp fluorescence (Fig. 8 ​ 8) ) shows a linear change in Trp fluorescence with increased urea concentration, with the slope of the curves at 0.04 and 0.07 M 𢄡 for MxiH and PrgI, respectively. Model studies with the tryptophan derivative, N-acetyltryptophanamide (NATA), find that there is a linear dependence of indole fluorescence on urea concentration with a slope of 0.07M 𢄡 (Eftink 1994). Thus, our results suggest that the urea-induced changes seen here are simply due to direct effects of urea on the MxiH and the PrgI indole side chains. Fluorescence based thermal unfolding studies show a curvilinear dependence of Trp fluorescence on temperature (Fig. 9 ​ 9). ). Furthermore, the emission peak position does not change with temperature, supporting the conclusion that the Trp environment does not significantly change.

Urea-induced unfolding of MxiH and PrgI at 10ଌ as monitored by Trp fluorescence with excitation at 295 nm. The proteins were examined at 50 μM. Panels A and B represent MxiH and PrgI with slopes 0.05 and 0.07 M 𢄡 , respectively.

Thermal unfolding curves of MxiH and PrgI as monitored by Trp fluorescence. Panels A1 and A2 represent PrgI, while panels B1 and B2 represent MxiH. The upper panels (A1,B1) show the intensity of the fluorescence emission at 340 nm as a function of temperature. The lower panels (A2,B2) show the wavelength of the emission maximum as function of temperature.

Differential scanning calorimetry of PrgI

PrgI produces a distinct DSC endotherm with a Tm of 38ଌ and a ΔH of 27.9 kcal/mol, in excellent agreement with the CD and absorbance results (Fig. 10 ​ 10). ). Surprisingly, for reasons that are unknown at this time, we were unable to obtain a well-defined DSC endotherm for MxiH, even at high protein concentrations.

PrgI DSC endotherm with buffer subtraction and baseline correction at a protein concentration of 1.82 mg/mL. A two-state unfolding model was used to fit the data and the values of Tm=38.1ଌ ± 0.3 and ΔH=27.9 kcal/mol ± 0.1 were found. The solid line is from a representative experiment with the dashed line showing a best fit analysis.

Application of Tryptophan Fluorescence Bandwidth-Maximum Plot in Analysis of Monoclonal Antibody Structure

Monoclonal antibodies have become the fastest growing protein therapeutics in recent years. The stability and heterogeneity pertaining to its physical and chemical structures remain a big challenge. Tryptophan fluorescence has been proven to be a versatile tool to monitor protein tertiary structure. By modeling the tryptophan fluorescence emission envelope with log-normal distribution curves, the quantitative measure can be exercised for the routine characterization of monoclonal antibody overall tertiary structure. Furthermore, the log-normal deconvolution results can be presented as a two-dimensional plot with tryptophan emission bandwidth vs. emission maximum to enhance the resolution when comparing samples or as a function of applied perturbations. We demonstrate this by studying four different monoclonal antibodies, which show the distinction on emission bandwidth-maximum plot despite their similarity in overall amino acid sequences and tertiary structures. This strategy is also used to demonstrate the tertiary structure comparability between different lots manufactured for one of the monoclonal antibodies (mAb2). In addition, in the unfolding transition studies of mAb2 as a function of guanidine hydrochloride concentration, the evolution of the tertiary structure can be clearly traced in the emission bandwidth-maximum plot.

This is a preview of subscription content, access via your institution.


Considerable progress has been made in our understanding of the mechanism of protein folding and aggregation in vitro, providing a foundation to elucidate protein misfolding pathologies such as the neurodegenerative diseases—Alzheimer's, Parkinson's, and other amyloid diseases. 1 However, the complexities of the cellular environment must be considered in order to face the challenges of protein folding and misfolding in vivo. 2 , 3 Folding in vivo occurs in a crowded, spatially organized, heterogeneous cellular environment with a cast of folding assistants this environment is drastically different from the highly dilute solutions and optimized buffers exploited for in vitro folding studies. Theoretical predictions as well as some exploratory experiments suggest that the high concentration of macromolecules (350–400 mg/ml) will markedly affect the conformational dynamics and energetics of folding polypeptides, favoring compact states. 4 , 5 Moreover, proteins emerge vectorially from the ribosome with a rate comparable with the rates observed for fundamental steps in protein folding in vitro the rate of translation in bacteria is ∼10–50 amino acids per second and in eukaryotes, 4–5 amino acids per second. Thus, the N-terminal regions of a growing polypeptide in vivo will explore conformational space before the C-terminal regions have emerged from the ribosome. The sequence code for folding has arisen based on in vivo selection pressures and must necessarily reflect this vectorial conformational search process. Additionally, molecular chaperones interact with a large fraction of nascent and newly synthesized polypeptides and these interactions may influence the energy landscape for folding. 3 Nonetheless, abundant evidence accumulated over the last half century argues persuasively that the intrinsic structural propensity of a protein's primary sequence contains the information for its eventual native fold. What is not yet well understood is how this sequence information is expressed in vivo.

Major technical challenges impede our ability to characterize protein folding in vivo using the wide array of sophisticated approaches that have been so informative in vitro. Observations of the protein of interest must be made at physiological concentrations in the background of all other cellular constituents. Kinetic experiments must monitor a population of synchronized molecules, in a system where the manifold of reactions is huge. Ideally, we seek to observe how structure formation occurs, what intermediate states are visited by a nascent or newly synthesized polypeptide, and what energetic differences exist among the accessible states. We must be able to ‘see’ the protein of interest first and foremost, leading to a need for signals that are observable above the cellular background. The signal must report on the conformational explorations of the labeled protein. And then we must be able to manipulate the system to pose questions about the formation of structure in the protein.

We have used a predominantly β-sheet protein, cellular retinoic acid-binding protein I (CRABP), whose in vitro folding we have explored in detail, 6-9 to study folding and aggregation in vivo. We have incorporated a binding site (tetra-cysteine motif) for the fluorescent dye, 4′,5′-bis(1,3,2-dithioarsolan-2-yl)fluorescein (FlAsH), 10 , 11 into a loop in both wild type CRABP and a slow-folding, aggregation prone mutant, P39A CRABP (Figure 1). 12 FlAsH fluorescence has allowed us to assess the apparent in vivo stability of CRABP, 12 to elucidate the aggregation mechanism of P39A tetra-Cys CRABP in vitro, 13 and to probe the behavior of chimeras of CRABP and exon 1 of huntingtin (Htt), 14 the protein that aggregates in Huntington's disease. Our findings suggest that aggregation of P39A follows a nucleation–polymerization mechanism from a monomeric nucleus. 13 Additionally, the osmolyte proline abrogates this aggregation both in vitro and in vivo. 15 Lastly, expansion of the polyglutamine repeat in the Htt exon 1 destabilizes the flanking CRABP and leads to aggregation that is dominated early on by CRABP and later by the polyglutamine tract. 14 These and other new results are enabling us to move from the test tube to the cell in our exploration of protein folding and aggregation.

Design of tetra-Cys CRABP. (A) Backbone structure of CRABP (PDB code 1CBI) showing the position of the tetra-Cys motif in the highly variable Ω-loop. (B) Fluorescence spectra of FlAsH-labeled tetra-Cys CRABP in its native state (folded) and denatured with 8M urea (unfolded). [Figure adapted from Ignatova, Z. Gierasch, L. M. Proc Natl Acad Sci USA 2004, 101, 523–528.]

In this article, we briefly review the results we have obtained to date and the questions these results raise. Then, we report intriguing new findings that suggest fundamental differences in the way folding occurs in vivo and how we are investigating the reasons for these differences.

Materials and Methods

Production and Purification of SP-BN

The human wild-type propeptide was expressed in E. coli as a fusion protein (MBP-SP-BN) with the Maltose Binding Protein (MBP) and purified after cleaving the fusion with factor Xa as described [13]. The fractions containing the purified SP-BN in 20 mM Tris-HCl buffer pH 7, 500 mM NaCl were dialyzed towards the buffer needed in subsequent experiments. The mean neat charge of the propeptide is defined as the net charge at pH 7 [18] divided by the total number of residues. The hydrophobicity at each sequence position was calculated by the Kite and Doolittle scale [19] using a window size of 5 amino acids and normalized to a scale of 0–1. The mean hydrophobicity is defined as the sum of the normalized hidrophobicities of all residues divided by the number of residues of the propeptide.

Far-UV Circular Dichroism Experiments

Circular dichroism spectra of SP-BN were recorded at 25°C in a Jasco J-715 spectropolarimeter using thermostated quartz cells of 0.1-cm path length, at 50 nm·min -1 (1 s response time) for the far-UV (250–195 nm) spectral range, each spectrum being the accumulation of 5 scans. The spectra were obtained in 200 μL of 5mM acetate, 5 mM MES, 5 mM Tris-HCl buffer (AMT buffer) 150 mM NaCl pH 7 at 0.115 mg·mL -1 protein. Mean residue molar ellipticities [θ] were calculated from the measured ellipticity taking into account the protein concentration, the molecular weight of SP-BN (19,902 Da, DNAstar program) and the number of amino acids per molecule (177). Estimations of the secondary structure content from the CD spectra were performed by using the CDPro suite program and the α-helix and β-sheet contents were calculated using three different methods, CONTIN/LL, SELCON3 and CDSSTR employing their mean value [20]. CD spectra in the presence of chaotropes were obtained after preincubating the samples with the denaturant for 1 h at 25°C. Samples at the corresponding chaotrope concentration, 0.8–8 M GdmCl and 0.5–7.5 M urea, were prepared from stock solutions of 9.75 M GdmCl (Sigma) and 9.87 M urea (Sigma) respectively. Control samples in the absence of protein were used to subtract a baseline from samples with protein. The reversibility of chemical unfolding was analyzed by dialyzing extensively samples with the highest chaotrope concentration towards 5 mM AMT buffer, 150 mM NaCl pH 7 at 4°C and by recording the spectra thereafter.

Temperature studies were carried out with 0.115 mg·mL -1 protein in 5 mM AMT buffer pH 7 with 150 or 500 mM NaCl. The temperature dependence of the CD signal was determined by heating the samples from 25 to 85°C at 30°C·h -1 and collecting the ellipticity at 208 nm every 0.2°C. After the T ramp the sample was cooled back to 25°C and the CD spectrum was recorded again.

Intrinsic and Extrinsic Fluorescence Studies

The intrinsic fluorescence emission spectra of the propeptide were recorded at 25°C in a SLM-Aminco AB2 spectrofluorimeter using a 1-cm quartz cell with excitation (290 nm) and emission slits set at 4 nm and scan speed of 2 nm·s -1 . Samples contained 0.09 mg·mL -1 protein in 5 mM AMT buffer, 150 mM NaCl pH 7. The spectra of the protein in the presence of chaotropes, 0.8–6.6 M GdmCl or 0.5–7.5 M urea were recorded after preincubating the samples for 4 h at 25°C. Spectra were also recorded of samples dialyzed as described above to check the reversibility of protein unfolding.

Extrinsic fluorescence of 11 μM bis-ANS probe (4,4´-bis-1-phenylamine-8-naphftalene sulfonate from Thermo Fisher Scientific) was determined as follows. Samples containing chaotropes were prepared in a final volume of 200 μL by adding 11 μL of the probe from a 200 μM solution in methanol to 29 μg protein in 5 mM AMT buffer, 150 mM NaCl pH 7 with the corresponding amount of GdmCl or urea. After incubation at 37°C for 5 min, the emission spectra (400–625 nm) were recorded in the spectrofluorimeter connected to a water-bath thermostatized at 37°C. The excitation wavelength was 395 nm and samples without protein were used as blank. Scan speed and slit widths were as described above.

Analysis of CD and Fluorescence Emission Data

The transition curves obtained by CD and fluorescence spectroscopy from the propeptide unfolding experiments were analyzed according to the following equations: (1) (2) (3) (4)

In the unfolding process by chaotropes, in Eqs (1) and (2): Yobs is the observed parameter ([θ] 220 or FI350) at each denaturant concentration, YN and YU are the parameter values in native and unfolded conditions respectively, D is the denaturant concentration (mol·L -1 ), ΔG 0 H2O is the change of free energy in the absence of denaturant (kJ·mol -1 ), m is a measure of the dependence of the free energy on the denaturant concentration (kJ·mol -1 ·M -1 ), R is the gas constant (8.314 J·mol -1 ·K -1 ) and T is the absolute temperature (K). The midpoint of the unfolding curve (D1/2) is the denaturant concentration at which half of the protein is unfolded and can be calculated as D1/2 = ΔG 0 H2O/m since ΔG 0 H2O and m are obtained from Eq (1). Alternatively, D1/2 can be determined directly from fitting the same data to Eq (2) being b and a constants. In temperature analysis, D and D1/2 in Eq (2) were substituted by T and the transition temperatures respectively (all of them in°C for easy recognition) and Yobs was [θ] 208 in Eqs (2) and (3). The change of free energy, ΔG 0 was calculated for each temperature (K) according to Eq (3). The enthalpy change, ΔH 0 (kJ·mol -1 ) and the entropy change, ΔS 0 (kJ·mol -1 ·K -1 ) were the intercept and the slope, respectively, in Eq (4)

Analytical Ultracentrifugation

Hydrodynamic studies of 0.15 mg·ml -1 SP-BN in 5 mM Tris-HCl pH 7.0 containing either, 0, 150 or 500 mM NaCl were performed at 20°C and 48,000 rpm in an Optima XL-1 (Beckman-Coulter Inc) analytical ultracentrifuge equipped with UV-visible optics. The partial specific volume of SP-BN was 0.73 mL·g -1 estimated from its amino acid composition with the program SEDNTERP, version 1.09 (retrieved from RASMB server) [21] the solvent density ρ was 1.000 g·mL -1 and the solvent viscosity η was 1.002 cpoise in the absence of salt, 1.005 g·mL -1 and 1.017 cpoise with 150 mM NaCl and 1.019 g·mL -1 and 1.049 cpoise with 500 mM NaCl respectively, estimated with the same program.

Differential Scanning Calorimetry

Differential scanning calorimetry (DSC) experiments were performed on a VP-DSC (MicroCal) differential scanning microcalorimeter with a cell volume of 514.9 μL. The propeptide (0.2–0.6 mg·mL -1 ) in 5 mM Tris-HCl buffer pH 7, 150 or 500 mM NaCl and the corresponding buffer were degassed for 3 min at room temperature in a chamber under vacuum and gentle stirring and then loaded into the sample and reference cells where overpressure was kept to prevent degassing. The measurements were taken every 0.1°C and the scan rate was 60°C·h -1 . DSC scans began at 20°C and were over at the highest temperature possible in the VP-DSC (

120°C). Second scans were obtained by reheating samples after cooling them for 20 min upon the completion of the first scan. The apparent Cp profiles were obtained by subtracting the instrumental baseline (obtained with buffer in both cells) from the experimental thermograms. Then, the thermograms were normalized for protein concentration based on a monomer of 19,902 Da and the pre- and post-transition baselines were subtracted.

The reversible transitions were subjected to thermodynamic deconvolution analysis and the thermogram peak of each successive scan was fitted to the Non Two-State model with two peaks according to Eq (5) of the DSC tutorial guide with the Origin Microcal software, as this model accounted for the best fitting of the data (the lowest χ 2 / DoF (Degree of Freedom)). (5) where Tm1 (melting temperature), ΔHVH1 (van´t Hoff enthalpy change) and ΔHcal1 (calorific enthalpy change) account for the thermal transition of domain 1 whereas Tm2, ΔHVH2 and ΔHcal2 account for the thermal transition of domain 2. Cp is the excess heat capacity (kJ·mol -1 ·K -1 ).

The activation energy for irreversible transitions (EA), considered as one-step processes from native to irreversibly inactivated state of the protein [22], was determined by Eq (6) where Cp ex max is the maximum excess of heat capacity obtained at the Tm.

Samples of 0.6 mg·mL -1 protein in 5 mM Tris-ClH buffer 150 mM NaCl pH 7 containing 2 M urea in the same buffer were subjected to DSC in the temperature range 20–120°C at 60°C·h -1 scan rate. The chaotrope was added to the reference cell at the same concentration than was added to the sample cell.

Analytical Procedures

Along the propeptide production and purification processes, protein concentration was routinely determined with colorimetric methods as described [13] but once purified, the propeptide concentration was calculated from its absorbance at 280 nm using 20,790 M -1 ·cm -1 as molar extintion coefficient at 280 nm [14].

Free thiols of purified SP-BN were titrated by adding 40 μL of 5 mM 5,5´- dithio-bis-(2-nitrobenzoic acid) (DTNB, Sigma) to 38.4 μg of protein in 960 μL of 5 mM Tris-HCl pH 8.2, 150 mM NaCl (assay buffer). Protein was previously dialyzed towards the same buffer at 4°C. The absorbance at 412 nm was recorded after 2 h at 22°C to check the release of 5-thiobis-(2-nitrobenzoic acid) (TNB) [23]. L-Cysteine (Fluka) was the standard and ε412 = 13,800 M -1 cm -1 was used to quantify cysteines. For titration under denaturing conditions the same protocol was applied except that the protein was preincubated with the assay buffer containing 5.7 M GdmCl, which was added 30 min prior to DTNB addition. Free thiols were also analyzed in the fusion MBP-SP-BN (33.5 μg) and in MBP (75.2 μg released from the fusion by factor Xa) after removing the 2-mercaptoethanol contained in the purification buffer by extensive dialysis towards the assay buffer.

How to Stabilize Protein: Stability Screens for Thermal Shift Assays and Nano Differential Scanning Fluorimetry in the Virus-X Project

A protocol is presented to rapidly test the thermal stability of proteins in a variety of conditions through thermal shift assays and nano differential scanning fluorimetry. Buffer systems, salts and additives, together comprising three unique stability screens, are assayed with proteins to identify suitable buffers for functional and structural studies.


The Horizon2020 Virus-X project was established in 2015 to explore the virosphere of selected extreme biotopes and discover novel viral proteins. To evaluate the potential biotechnical value of these proteins, the analysis of protein structures and functions is a central challenge in this program. The stability of protein sample is essential to provide meaningful assay results and increase the crystallizability of the targets. The thermal shift assay (TSA), a fluorescence-based technique, is established as a popular method for optimizing the conditions for protein stability in high-throughput. In TSAs, the employed fluorophores are extrinsic, environmentally-sensitive dyes. An alternative, similar technique is nano differential scanning fluorimetry (nanoDSF), which relies on protein native fluorescence. We present here a novel osmolyte screen, a 96-condition screen of organic additives designed to guide crystallization trials through preliminary TSA experiments. Together with previously-developed pH and salt screens, the set of three screens provides a comprehensive analysis of protein stability in a wide range of buffer systems and additives. The utility of the screens is demonstrated in the TSA and nanoDSF analysis of lysozyme and Protein X, a target protein of the Virus-X project.


Many biotechnologically-useful enzymes originate from viral sources, such as the tobacco etch virus (TEV) protease 1 and human rhinovirus type 3C (HRV 3C) protease 2 . The Horizon2020 Virus-X project (Figure 1) 3 . The aim of this program is (a) to extend the range of the properties of known enzyme families and (b) to characterize novel enzymes of yet unknown function (Enzyme X). Crystallographic structure determination plays a pivotal role in target protein characterization, in particular in those cases where the protein sequences have evolved beyond recognition 4 . Protein stability is a key factor in the crystallization process samples must be conformationally homogenous and structurally sound over a period of time to form high-quality, diffracting crystals. Furthermore, it is essential for activity assays that the proteins exist in their active conformation, which can also be facilitated by a favorable molecular environment.

Despite the development in the technology available to crystallographers, protein crystallization remains a time-consuming and labor-intensive empirical process 5 . Preliminary biophysical experiments to improve the protein stability in solution give clearly a better starting point for protein crystallization and consume usually only a comparatively small amount of protein sample 6 , 7 , 8 , 9 . The large number of target proteins to be studied in this project also necessitates scalable, high-throughput stability assays. One of the most popular methods for pre-crystallization biophysical characterization of the proteins is the thermal shift assay (also known as TSA or differential scanning fluorimetry, DSF) 10 , 11 .

TSAs employ an environmentally-sensitive fluorescent dye to track the thermal denaturation of protein samples. Many commonly-used dyes have variable fluorescence activity depending on the polarity of their environment, often displaying a high fluorescence output in hydrophobic environments but undergoing rapid quenching in polar environments 12 . Proteins generally cause pronounced increases in dye fluorescence as their hydrophobic cores become exposed during denaturation, often followed by a decrease in dye fluorescence at very high temperatures as proteins begin to aggregate (Figure 2).

While a hydrophobicity-sensitive dye is often a good choice for a general-use TSA dye, it can be unsuitable for proteins with large, solvent-exposed hydrophobic regions, which often display detrimentally high background fluorescence. Fluorophores with alternative modes of action exist (see Discussion), but it may instead be desirable to track denaturation through intrinsic protein fluorescence with nanoDSF.

Tryptophan residues that are buried in nonpolar regions of a protein fluoresce with an emission maximum of 330 nm. As a protein sample unfolds and these residues become exposed to a polar solvent, their emission maximum undergoes a bathochromic shift to 350 nm 13 . nanoDSF exploits this shift in emission maximum to probe the unfolding of a protein sample without the need for extrinsic fluorophores 14 .

Melt curves showing single denaturation steps can be analyzed by fitting data to a Boltzmann sigmoidal model. The temperature at the inflection point of the unfolding transition (Tm) is used as a quantitative measure of protein thermal stability and a benchmark to compare the favorability of different conditions.

Melt curves of the same protein in different conditions sometimes possess a degree of heterogeneity that can make a Boltzmann sigmoidal fitting unfeasible. To discern Tm values from data that deviates from the classic curve topology, numerical methods can be used such as those employed in NAMI, an open-source TSA data analysis program 11 . Alternative thermodynamic frameworks can also be used to analyze more complex curves with multiple denaturation steps, such as the ProteoPlex methodology 15 .

The stability screens were designed for use in TSA and nanoDSF experiments to rapidly identify favorable conditions for a target protein (Figure 3, screen compositions are available in the supplementary information). Information gathered with the screens can be used at many stages of the crystallographic pipeline including: sample storage purification, minimizing yield loss through protein unfolding during the purification process assay design, reinforcing protein functionality in activity assays with thermally stabilizing buffers and finally crystallization, guiding rationally-designed crystallization trials.

Choosing a suitable buffer system basis for a protein sample is vital incompatible pH values can lead to the deactivation or denaturation of a protein. However, the presence of co-crystallized buffer molecules resolved in a large number of X-ray crystal structures (Table 1) could also be indicative of a stabilizing effect that is separate to simple pH regulation and instead stems from the chemical features of the buffer molecule.

Formulated using several of Good's buffers 17 , 18 , 19 alongside other commonly biologically-compatible buffer systems, the pH screen is designed to deconvolute the chemical effect of a buffer molecule on protein stability from the actual pH of the resulting solution. By providing three pH values for each buffer system and incorporating pH value redundancy between different systems, the pH screen can identify both favorable pH values and favorable buffer systems for a target protein.

The salt screen contains commonly laboratory salts as well as chaotropes, chelants, heavy metals and reducing agents. The screen can give a general indication of the affinity of a protein sample to the environments with high ionic strengths, but each subgroup of the compounds can also provide information on the potential structure of a protein. For example, a chelant significantly destabilizing a protein could be indicative of important structural metals within the sample. If the sample is also strongly stabilized by a metal cation within the screen, this can provide a promising starting point for further structural experiments.

Osmolytes are soluble compounds that affect the osmotic properties of their environment. In nature, they can be used as "chemical chaperones", enforcing the folding of disordered proteins and stabilizing them, especially in stress conditions 20 , 21 , 22 . These characteristics make them attractive additives in protein crystallography usable as cryoprotectants during the crystal harvesting, mounting and storage processes 23 . Osmolytes' potential use also extends to the purification of proteins. A significant proportion of recombinant proteins expressed in E. coli can be insoluble and difficult to recover in the native state using standard purification methods. Osmolytes can be used to stabilize and salvage proteins from insoluble fractions, increasing purification yields 24 .

The osmolyte screen was designed using established compounds present in Protein Data Bank entries 25 and the Dragon Explorer of Osmoprotection-Associated Pathways (DEOP) 26 database and iteratively optimized using standard proteins. The screen is built around eight subclasses of osmolyte: glycerol, sugars and polyols, non-detergent sulfobetaines (NDSBs), betaines and their analogues, organophosphates, dipeptides, amino acids and their derivatives and a final miscellaneous group. Each osmolyte is present in multiple concentrations based on its solubility and effective concentration ranges for comparison.


1. Preparation of Protein Sample

  1. Formulate the stability screens as 500 µL aliquots in 96-well blocks and seal for storage. Transfer 10 µL of each condition of a stability screen into the corresponding well of a 96-well plate using a multi-channel pipette to save time (Figure 4A).
  2. Prepare 1 mL of an approximately 1 mg mL -1 protein solution in an appropriate buffer system. While the composition of an appropriate buffer varies with each protein sample, a good first buffer to try is 10 mM sodium phosphate with 100 mM NaCl, pH 7.2.
    NOTE: Acceptable protein concentrations vary case-by-case, but concentration ranges of 0.5 - 5 mg mL -1 typically produce analyzable curves. Dilute buffers are recommended for use with the stability screens to avoid masking the effects of each condition. Typical buffer compositions are approximately 10 mM buffer with around 100 mM NaCl.
  3. If performing a TSA experiment, add SYPRO Orange dye to the protein sample to a final concentration of 20x. Mix either by inversion or brief vortexing.
  4. Transfer 10 µL of the protein solution into each well of the 96-well plate prepared in Step 1.1 (Figure 4B).
  5. Seal and centrifuge the 96-well plate for 2 min at 600 x g to ensure the protein sample and screen component are mixed (Figure 4C).
  6. Re-seal the stability screen deep well block and store the screen at 4 °C for up to 4 months. Store the salt screen in darkness, as some components are photosensitive.
  7. If performing a nanoDSF experiment, continue to step 2. If performing a TSA experiment, skip to step 4.

2. Preparing a nanoDSF Experiment

  1. Ensure that the equipment is clean, paying particular attention to any dust near the sample rack. If the system has a backscattering mirror, clean it using ethanol and a lint-free tissue.
  2. Open the sample drawer by pressing the Open Drawer button. (Figure 5A).
  3. Load the capillaries with approximately 10 µL from each well of the 96-well plate by touching one end of the capillary into the solution, then place them into the corresponding capillary holders of the sample rack (Figure 5B). Be careful not to contaminate the middle of the capillaries with fingerprints, etc., as this could interfere with fluorescence readings throughout the experiment.
  4. Immobilize the capillaries with the magnetic sealing strip (Figure 5C).

3. Programming a nanoDSF Experiment

  1. Launch a preliminary scan to detect the position and intensity of each capillary by pressing the Start Discovery Scan button in the Discovery Scan tab. Increase or decrease the incident excitation strength from an initial power of 10% until the peak of every capillary scan is between 4,000-12,000 units (Figure 5D).
  2. To ensure the sample is folded and sufficiently concentrated, an initial melt scan with a steep temperature gradient is recommended. In the Melting Scan tab, program a melt scan by setting the Temperature Slope option to 7.0 °C min -1 , Start Temperature to 25 °C and End Temperature to 95 °C, then launch the nanoDSF experiment by pressing the Start Melting button. If the resulting melt curves do not show a detectable inflection point, consider concentrating the sample further or checking if the protein is folded properly.
  3. Repeat steps 2.1-2.4 to prepare the samples for a full experiment.
  4. In the Melting Scan tab, program a melt scan by setting the Temperature Slope option to 1.0 °C min -1 , Start Temperature to 25 °C and End Temperature to 95 °C, then launch the nanoDSF experiment by pressing the Start Melting button.

4. Performing a TSA Experiment

  1. Open the sample drawer by firmly pressing the indent on the right-hand side of the drawer. Place the 96-well tray in the RT-PCR system with well A1 to the back-left (Figure 4D).
  2. Click the New Experiment button to begin setting up a TSA experiment.
  3. In the Experiment Properties tab, click the Melt Curve option when asked What type of experiment do you want to set up? and the Other option when asked Which reagents do you want to use to detect the target sequence?
  4. In the Plate Setup/ Define Targets and Samples tab, enter a target name then set Reporter as ROX and Quencher as None.
  5. In the Plate Setup/ Assign Targets and Samples tab, assign every well of the 96-well plate to the target name entered in the previous step. In the same tab, set Select the dye to use as the passive reference as None.
  6. In the Run Method tab, delete steps until there is a total of three. Set the first step to 25.0 °C, ramp rate 100%, time 00:05 the second step to 95.0 °C, ramp rate 1%, time 01:00 the third step to 95.0 °C, ramp rate 100%, time 00:05. Choose to collect data using the Collect Data dropdown menu or by pressing the Data Collection icon (Figure 6).
  7. Set the Reaction Volume Per Well to 20 µL.
  8. Press the Start Run button to begin the TSA experiment.
  1. Choose a wavelength to plot a melt curve with. For nanoDSF experiments, the ratio of fluorescence intensities at 330 nm and 350 nm (corresponding to tryptophan in nonpolar and polar environments, respectively) 13 is commonly used. For most TSAs, the dye emission maximum is suitable for melt curve plotting (the emission maximum of SYPRO Orange is 569 nm) 12 .
  2. Calculate the Tm values of each condition by determining the inflection point(s) of each melt curve. Most nanoDSF systems automatically calculate Tm values by numerical differentiation of melt curves after data acquisition. If the software used does not automatically calculate Tm values, free, alternative GUI-driven software such as NAMI 11 can automate data analysis and downstream processing, giving the option to produce a heatmap summarizing Tm values for the entire 96-well experiment (the accompanying reference provides guidance and resources for data processing with NAMI).
  3. Compare Tm values of all conditions surveyed. The stability screens contain two wells in each screen (A1 and A2) that contain only water. Taking the water-only values as a benchmark allows calculation of ΔTm values, addressing systematic errors and allowing easy comparison of stabilizing effects. Higher Tm values indicate thermally stabilizing conditions which are recommended for downstream use. Promising conditions often show a concentration dependence in their stabilization.

Representative Results

Lysozyme was assayed with the stability screens and Protein X, a target protein in the Virus-X project, was assayed with the osmolyte screen. Both proteins generally produced melt curves with clearly-defined denaturation transitions in both TSA and nanoDSF experiments (see accompanying figures for representative curves). In a few cases where the samples that did not produce interpretable curves with a defined denaturation transition were interpreted as denatured and not included in Tm comparisons.

Figure 7 shows sample results from the salt screen, exemplifying the thermally-stabilising properties of ammonium chloride towards lysozyme. Concentration dependencies such as that shown above are often indicative of promising conditions, but it can be useful to compare the results of increasing ion concentration with several different salts to see if thermal stabilisation generally arises from an increase in buffer ionic strength or if the presence of specific ions confers additional stability.

Comparison of Tm values of lysozyme with the pH screen (Figure 8) reveals two pieces of information. Firstly, there is a general trend of increasing stability with decreasing pH values. Secondly, the range of Tm values obtained using different buffer systems with identical pH values can be significant.

Data in Figure 8 suggests that the agreement between TSA and nanoDSF in this experiment is generally good, but nanoDSF shows a tendency to identify slightly higher Tm values and slightly larger Tm shifts than TSA. However, some wells with pH values above 8.5 show large discrepancies between Tm values obtained from TSA and nanoDSF. Differences could potentially be attributed to the denaturation mechanism of the protein at different pH values for example, a hydrophobicity-sensitive dye could give a comparatively low Tm reading if hydrophobic regions of a protein become exposed to solvent significantly faster than the environment of tryptophan residues changes in polarity.

Figure 9 shows a heatmap of Tm values obtained with lysozyme using the osmolyte screen. Conditions with the highest Tm increases compared to the control wells (wells A1 and A2, containing deionized water) are colored dark blue. Especially stabilizing conditions identified in Figure 9 include glycerol, 1 M D-sorbitol, 100 mM hypotaurine and 10 mM Ala-Gly (wells A4-A6, A9, E7 and F8, respectively).

Figure 10 shows a heatmap of Tm values obtained with Protein X. TSA and nanoDSF experiments with the Osmolyte screen reveal that the majority of osmolytes tested give either a minor increase in Tm (within 1 °C) or have a detrimental effect on Protein X's stability. In particular, dipicolinic acid at a concentration of 10 mM (well D1) appears to denature the sample at room temperature. The TSA and nanoDSF results quickly identify dipicolinic acid as an incompatible additive for Protein X which should be avoided when working with the protein. Nevertheless, high concentrations of D-sorbitol and arabinose (wells A9 and B9, both at 1 M) as well as glycerol and TMAO (wells A4-A6 and E1, respectively) were identified as thermally-stabilizing.

For lysozyme, combinations of conditions yielding the highest Tm values from each stability screen were tested to probe for a synergistic combined effect. Figure 11 shows a general increase in Tm values as more components of the buffer system (pH, salt and osmolyte) are added. In the case of MES, ammonium sulphate and D-sorbitol, a Tm increase as large as 10 °C can be observed when all components are present compared to MES alone. Figure 11 shows that a noticeable synergistic effect can occur when individual components of a buffer are optimized and combined with the stability screens.

On a more general note, Figure 11 also illustrates the magnitude of ΔTm values that can be observed in TSA and nanoDSF experiments. The magnitude of ΔTm achievable varies significantly based on the protein system, but any ΔTm value around and above 5 °C is often indicative of a beneficial stabilizing effect.

Figure 1: Bioprospecting. Sample collection from an extreme environment in the Virus-X Project. Please click here to view a larger version of this figure.

Figure 2TSA schematic. Annotated example of a typical melt curve obtained from a TSA experiment. This curve is characteristic of classic "two-state" protein unfolding, where the sample population transitions from folded to denatured without detectable partially-folded intermediates. Please click here to view a larger version of this figure.

Figure 3: Stability screens workflow. Standard workflow of buffer optimization using the stability screens. Please click here to view a larger version of this figure.

Figure 4: Workflow of a standard TSA experiment with the stability screens. From left to right: (A) Pipetting aliquots of the stability screens into a 96-well plate. (B) Pipetting protein sample with fluorescent dye into the plate. (C) Sealing the plate before centrifugation. (D) Placing the plate into an RT-PCR system. Please click here to view a larger version of this figure.

Figure 5Workflow of a standard nanoDSF experiment. (A) Opening the capillary loading rack. (B) Loading the capillaries into the rack. (C) Immobilizing the capillaries with the magnetic sealing strip. (D) Programming an experiment. Please click here to view a larger version of this figure.

Figure 6: User interface for the RT-PCR system. A TSA experiment has been programmed. Please click here to view a larger version of this figure.

Figure 7: Salt screen sample data. (A) Ratio of fluorescence intensities at 350 nm vs 330 nm for a label-free nanoDSF experiment with lysozyme. Samples correspond to the wells C7-C12 of the salt screen (1.5 M - 0.2 M ammonium chloride). Calculated Tm values for each condition are superimposed on the plot. (B) Summary of Tm values calculated from the data presented in Figure 7A. (C) Fluorescence intensities at 590 nm for a TSA experiment using lysozyme with a hydrophobicity-sensitive reporter dye. Like Figure 7A, the samples correspond to the wells C7-C12 of the salt screen. Calculated Tm values are superimposed on the graph. (D) Summary of Tm values calculated from data present in Figure 7C. Please click here to view a larger version of this figure.

Figure 8pH screen sample data. Summary of Tm shifts obtained with lysozyme and the pH screen. Each point represents an independent condition points at the same pH value are of different buffer systems at the same pH. Tm shifts are calculated relative to a control Tm of 68.0 °C for TSA experiments and 71.9 °C for nanoDSF experiments. Please click here to view a larger version of this figure.

Figure 9: Osmolyte screen sample data (lysozyme). Summary of Tm values obtained from a label-free nanoDSF experiment with lysozyme and each well of the osmolyte screen. Tm values (in °C) are compared to the wells A1 and A2 which contain water as a control. A heatmap was generated based on ΔTm values compared (blue denotes a Tm increase and red a Tm decrease). Please click here to view a larger version of this figure.

Figure 10Osmolyte screen sample data (Protein X). (A) Summary of Tm values obtained from a label-free nanoDSF experiment with Protein X and the osmolyte screen. Tm values are compared to wells A1 and A2 which contain water as a control. A heatmap was generated based on ΔTm values compared (blue denotes a Tm increase and red a Tm decrease). (B) nanoDSF curves obtained from well A9 (1 M D-sorbitol, a stabilizing condition) and D1 (10 mM dipicolinic acid, a destabilizing condition). Please click here to view a larger version of this figure.

Figure 11: Buffer optimization effect on Tm. TSA Tm values of lysozyme combined with the conditions of each screen that afforded the largest increase in Tm. 100 mM acetic acid, pH 4.2, and 100 mM MES, pH 5.6, were chosen as the buffer systems, alongside 1.5 M ammonium sulphate as the salt. Osmolyte concentrations were identical to those found in the osmolyte screen: 10 mM Ala-Gly, 1 M D-sorbitol, 50 mM L-lysine and 100 mM hypotaurine. Error bars represent the standard deviation of six replicates. Please click here to view a larger version of this figure.

Molecule PDB Code Frequency of Co-crystallisation
Phosphate PO4 5132
Acetate ACT 4521
2-(N-Morpholino)-Ethanesulfonic Acid (MES) MES 1334
Tris(hydroxymethyl)aminomethane (tris) TRS 1155
Formate FMT 1072

Table 1: Summary of buffer molecule co-crystallization frequency in Protein Data Bank (PDB) entries. Data obtained through PDBsum 16 from a total of 144,868 entries (correct as of 12-5-18).


Critical aspects within the protocol include the centrifugation step and proper sealing of the 96-well plate for TSA experiments (step 1.5). Centrifugation ensures that the protein sample and screen condition come into contact and mix. Additionally, if an unsealed plate is used for a TSA experiment, there is a significant risk of solvent evaporating throughout the experiment, causing an increase in sample concentration and increasing the chance of premature protein aggregation.

TSAs and nanoDSF are amenable to a wide range of protein samples the vast majority of samples can produce interpretable melt curves with a hydrophobicity-based reporter dye or through dye-free nanoDSF. If standard fluorescence sources are not suitable for your protein, the simplest modification to the protocol that could be explored is the choice of fluorophore. Several alternative dyes could be suitable for TSA experiments. Examples include N-[4-(7-diethylamino-4-methyl-3-coumarinyl)phenyl]maleimide (CPM), a compound that fluoresces after reacting with a thiol 27 , and 4-(dicyanovinyl)julolidine (DCVJ), a compound that varies its fluorescence based on the rigidity of its environment, increasing its fluorescence as a protein sample unfolds 28 , 29 (the latter dye often requires high concentrations of sample).

Alternative methods of melt curve analysis are available if Tm is not automatically calculated by the instrument software. If data is homogenous and only one denaturation step is apparent in the melt curves, a truncated dataset can be fitted to a Boltzmann sigmoid with the following equation:

Where F is the fluorescence intensity at temperature T, Fmin and Fmax are the fluorescence intensities before and after the denaturation transition, respectively, Tm is the midpoint temperature of the denaturation transition and C is the slope at Tm. While this method works well for simple two-step denaturation processes, it is unsuitable for complex melt curves with multiple transitions.

One of the major advantages of TSA is its accessibility TSA experiments can be performed in any RT-PCR system with filters at suitable wavelengths for the fluorescence dye employed. This coupled with the low cost of consumables, ease of operation and relatively low amount of protein needed, make TSA a valuable technique for a wide range of project scales, both in industry and academia.

As well as indicating favorable buffer conditions, the screens contain some wells that may give clues to the presence of structural metals within a sample protein. Wells that may be of particular interest in the salt screen are G6 and G7, which contain 5 mM EDTA and 5 mM EGTA, respectively. Significant thermal destabilization in these wells may be indicative of important metal ions in the protein that are sequestered by the chelants. Compounds within the osmolyte screen can also potentially provide clues to the function of a protein. Many of the compounds in the screen belong to classes of molecule that are common substrates of enzymes. For example, the general stabilization afforded by saccharides (present in wells A11-B10) for lysozyme could be attributed to their structural similarity to established substrates of the enzyme, N-acetylglucosamine oligomers 30 .

The TSA and nanoDSF protocols outlined above can also be adapted to study protein-ligand interactions. Ligands that bind specifically to a protein can increase its thermal stability by introducing new interactions within the complex. A dose-dependent positive shift in protein Tm is a promising sign of a successful protein-ligand interaction. The speed, throughput and low cost of screening compound libraries with TSAs has made it a very popular method in early-stage drug discovery.

Optimizing the buffer conditions of target proteins and their ligand complexes can be essential for a project's success, as many literature examples demonstrate 31 , 32 , 33 , 34 . With a typical assay taking under 2 h including setup time, TSAs and nanoDSF coupled with stability screens represent a fast, inexpensive technique for buffer optimizations.


The authors have nothing to disclose.


This project has received funding from the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation programme (grant agreement n° 685778). This work was supported by the Biotechnology and Biological Sciences Research Council (BBSRC, grant numbers BB/M011186/1, BB/J014516/1). DB thanks the BBSRC Doctoral Training Partnership Newcastle-Liverpool-Durham for a studentship and Durham University Department of Biosciences for contributing toward the funding this work. We thank Ian Edwards for his help and the Durham University Department of Chemistry Mass Spectrometry department for their instrumental analysis of Protein X. We are grateful to Arnthor Ævarsson for his work with the Virus-X project, and thanks also to Claire Hatty and NanoTemper GmbH for lending and assisting with the Prometheus NT.48 system for this project. Finally, thank you to Frances Gawthrop and Tozer Seeds for their support as part of the BBSRC iCASE award.

Watch the video: Conformational stability: Protein folding and denaturation. MCAT. Khan Academy (February 2023).