Destabilization of the dimer interface is a common consequence of diverse ALS‐associated mutations in metal free SOD1

Abstract

Neurotoxic misfolding of Cu, Zn‐superoxide dismutase (SOD1) is implicated in causing amyotrophic lateral sclerosis, a devastating and incurable neurodegenerative disease. Disease‐linked mutations in SOD1 have been proposed to promote misfolding and aggregation by decreasing protein stability and increasing the proportion of less folded forms of the protein. Here we report direct measurement of the thermodynamic effects of chemically and structurally diverse mutations on the stability of the dimer interface for metal free (apo) SOD1 using isothermal titration calorimetry and size exclusion chromatography. Remarkably, all mutations studied, even ones distant from the dimer interface, decrease interface stability, and increase the population of monomeric SOD1. We interpret the thermodynamic data to mean that substantial structural perturbations accompany dimer dissociation, resulting in the formation of poorly packed and malleable dissociated monomers. These findings provide key information for understanding the mechanisms and energetics underlying normal maturation of SOD1, as well as toxic SOD1 misfolding pathways associated with disease. Furthermore, accurate prediction of protein–protein association remains very difficult, especially when large structural changes are involved in the process, and our findings provide a quantitative set of data for such cases, to improve modelling of protein association.

Abbreviations

ALS

amyotrophic lateral sclerosis

apo

unmetallated

AUC

analytical ultracentrifugation

ΔC_p,d

specific heat capacity of dimer dissociation

ΔC_p,M↔U

specific heat capacity of monomer unfolding

ΔC${\hspace{0.17em}}_{{p,\mathrm{N}}_2}$_↔2U

total specific heat capacity of dimer unfolding

fALS

familial ALS

ΔG_d

Gibbs free energy of dimer dissociation

ΔH_d

dimer dissociation enthalpy

ITC

isothermal titration calorimetry

K_d

equilibrium constant for dimer dissociation

M

folded monomer

N₂

folded dimer

pWT

pseudo wild‐type

sALS

sporadic ALS

ΔS_d

dimer dissociation entropy

SEC

size exclusion chromatography

SOD1

Cu, Zn‐superoxide dismutase

Introduction

Amyotrophic lateral sclerosis (ALS, or Lou Gehrig's disease) is a common, rapidly progressive motor neuron disease, for which there is no cure, and treatment options are very limited.1 Mutations in homodimeric Cu, Zn‐superoxide dismutase (SOD1) (Fig. 1) account for ∼20% of familial and 3–5% of sporadic ALS cases (fALS and sALS, respectively),1 and an increasing body of evidence also suggests a role for wild‐type SOD1 in disease.2 Clinical symptoms are similar in fALS and sALS;1 thus, studies on SOD1 are critical for understanding the molecular mechanisms of disease. They are also important for advancing knowledge of natural protein maturation and association, as most proteins are homooligomers.3 Despite extensive and intensive ongoing research, accurate prediction of protein–protein association remains very difficult, especially when large conformational changes are involved in the process.4, 5

Figure 1.

Structure of the apo SOD1 homodimer. Each monomer forms a Greek key β‐barrel and contains a single disulfide bond (orange). In the absence of bound metals, loops IV (green) and VII (blue) are disordered and the structure is more flexible than the metal‐bound (holo) protein.48, 50 The sites of ALS‐associated mutations characterized herein are shown in red sticks (PDB code 1HL4).

Many, predominantly missense, mutations distributed throughout SOD1 have been associated with fALS (http://alsod.iop.kcl.ac.uk/home.aspx), and confer a toxic gain‐of‐function to the protein.1 A characteristic feature of ALS is the formation of protein aggregates in motor neurons,6 and SOD1 is a component of aggregates in SOD1‐linked fALS and in some sALS cases,7 as well as in animal models of the disease.8 Thus, a prominent hypothesis is that ALS is caused by protein misfolding and aggregation, analogous to toxic misfolding of other proteins in various neurodegenerative diseases such as Huntington's, Alzheimer's and prion disorders.9 The mechanisms by which SOD1 forms aggregates, however, remain poorly understood.10

Extensive research indicates that an increased population of unmetallated (apo) monomeric forms of SOD1 may promote toxic aggregation.11, 12, 13, 14, 15, 16, 17, 18, 19, 20 Folding and thermodynamic stability measurements of SOD1 can reveal if and how mutations create instability and so favour misfolding and aggregation.10 Here, we report direct quantitative measurements of the thermodynamics of dimer dissociation for chemically and structurally diverse ALS‐associated mutations distributed throughout the SOD1 structure (Fig. 1) by isothermal titration calorimetry (ITC) and size exclusion chromatography (SEC). ITC is a powerful, yet little explored, tool for characterizing the thermodynamics of homooligomeric protein–protein association using measurements of the heat changes that accompany binding.21 ITC measurements can also provide a valuable resource for improving modelling of protein association.4, 5, 21 In our ITC experiments, concentrated dimeric apo SOD1 is diluted into buffer in small volume injections (see Supporting Information, Figs. 2(A,B) and S1). Owing to mass action, dilution results in dimer dissociation and the observed heat decreases with successive injections. By fitting this decay to a dimer dissociation model, a single ITC experiment can define the dissociation equilibrium constant (K _d) and enthalpy (ΔH _d), from which the dissociation Gibbs free energy (ΔG _d), and entropy (ΔS _d) can be determined. Furthermore, from measurements of ΔH _d as a function of temperature, the change in the specific heat capacity for dissociation (ΔC _p,d) can be obtained, which relates to the magnitude of conformational changes that accompany the process.22, 23, 24 Using ITC to analyze the thermodynamics of apo SOD1 dimer dissociation at physiological temperature and pH, as well as parallel measurements of dissociation using SEC, we obtain the striking result that diverse mutations decrease dimer interface stability and increase the formation of structurally disrupted monomers. These findings provide valuable quantitative data on the energetics and mechanism of SOD1 subunit association for defining the molecular basis of natural protein maturation and disease and for improving the prediction of protein–protein association.

Figure 2.

Isothermal titration calorimetry reveals decreased dimer stability for apo SOD1 mutants. (A) Raw ITC data for the I113T mutant at 37°C. Each peak corresponds to the heat measured for a small volume injection of protein solution into the ITC sample cell. The endothermic heats of dissociation decrease for successive injections due to the increase in protein concentration in the cell and the accompanying shift of the equilibrium towards native dimer. (B) Integrated heats at 37°C for each injection (diamonds) are shown for pWT (black), E100G (blue), I113T (green), A4V (orange), and V148G (red), in order of increasing values of K _d. The dashed lines are the fits of the data to a dimer dissociation model [Eq. (1), Supporting Information] with fitted values shown in Figures 2(C), 3 and summarized in Table I. (C) For all mutants, the K _d is increased relative to the pWT as measured by ITC at 37°C. (D) The gray bars represent the increase in the K _d at 37°C relative to 25°C [ΔK _d = K _d (37)– K _d (25)]. SEC and ITC experiments at ∼23 and 25°C, respectively, were performed for H43R, I113T, A4V, and V148G. Extrapolating ΔG _d (25) values determined by ITC to ΔG _d (23) using Eq. (4) (Supporting Information) and a ΔC _p,d of 1.7 kcal (mol dimer)⁻¹ °C⁻¹ results in very little change in ΔG _d [0.03–0.09 kcal (mol dimer)⁻¹] and K _d (0.02–0.6 μM), which are smaller than the experimental uncertainties. Thus, the K _d (25) values were calculated as the average of the ITC and SEC values. For pWT, H46R, and G93A, the heat of dissociation was too small at 25°C to be measurable by ITC, and so the SEC values were used. Error bars reflect the standard deviation of multiple measurements of K _d (37) and K _d (25).

Results and Discussion

Apo SOD1 dissociation is endothermic and strongly enhanced with increasing temperature

ITC experiments were conducted for pWT (pseudo wild‐type) apo SOD1 and fALS‐associated mutants yielding K _d and associated thermodynamic parameters (Figs. 2, 3, and Table 1). We performed measurements at physiological temperature, 37°C, which has not been reported previously for apo SOD1. In addition, we conducted ITC measurements at 25°C and compared these with measurements by SEC (Fig. 4). Values of K _d obtained by the two methods are in excellent agreement [Fig. 3(B) and Table 1], and are consistent with results obtained previously at lower than physiological temperature by SEC, analytical ultracentrifugation and chemical denaturation.11, 14, 17, 25

Figure 3.

Thermodynamics of apo SOD1 dimer dissociation. (A) Kirchoff plot of ΔH _d versus temperature. The ΔC _p,d is determined from the temperature dependence of ΔH _d and is similar for diverse mutants. For V148G (red circles), A4V (orange squares), I113T (green diamonds), and H43R (purple triangles), the ΔC _p,d is 1.1 ± 0.1, 0.8 ± 0.1, 0.7 ± 0.2, and 0.8 ± 0.2 kcal (mol monomer)⁻¹ °C⁻¹, respectively. Thus, the average ΔC _p,d is ∼0.9 kcal (mol monomer)⁻¹ °C⁻¹ [∼1.7 ± 0.2 (kcal (mol dimer)⁻¹°C ⁻¹)]. (B) ΔG _d determined by ITC at 37 and 25°C (left bar, gray; middle, dark blue, respectively), and by SEC at ∼23°C (right, light blue). The expected difference in ΔG _d at 25 and 23°C calculated using Eq. (4) (Supporting Information) is less than 0.1 kcal (mol dimer)⁻¹. (C) ΔH _d (red) and TΔS_d (blue) determined by fitting heats of dissociation at 37°C. Error bars reflect the standard deviation of multiple (3–10) ITC experiments. The determined values are summarized in Table I.

Table 1.

Dimer Interface Stability of Apo SOD1 Variants Measured by ITC and SEC

SOD1 variant	ΔH d [kcal (mol dimer)−1)] ITC (37°C)a	K d (µM) ITC (37°C)a	ΔG d [kcal (mol dimer)−1] ITC (37°C)a, b	ΔΔG d [kcal (mol dimer)−1] ITC (37°C)c	ΔG d [kcal (mol dimer)−1] ITC (25°C)a, b	ΔG d [kcal (mol dimer)−1] SEC (23°C)b, d
pWT	30.8 ± 8.8	0.067 ± 0.050	10.3 ± 0.5	na	nd	11.0 ± 0.4
V148I	11.4 ± 2.2	0.5 ± 0.2	8.9 ± 0.2	1.3	nd	nd
G93S	17.6 ± 4.6	1.3 ± 0.5	8.4 ± 0.3	1.8	nd	nd
H46R	16.2 ± 4.4	1.4 ± 1.1	8.4 ± 0.4	1.9	nd	10.3 ± 0.3
E100G	16.0 ± 4.8	2.6 ± 1.7	8.0 ± 0.4	2.3	nd	nd
G37R	7.8 ± 1.8	5.0 ± 1.9	7.6 ± 0.2	2.7	nd	nd
H43R	23.0 ± 1.4	5.7 ± 0.4	7.5 ± 0.0	2.7	9.7 ± 0.1	10.2 ± 0.1
G93A	14.0 ± 2.0	8.4 ± 3.5	7.2 ± 0.3	3.0	nd	10.4 ± 0.3
I113T	30.2 ± 2.5	10.2 ± 3.0	7.1 ± 0.2	3.1	8.4 ± 1.1	9.5 ± 0.6
A4T	39.2 ± 3.8	10.6 ± 3.3	7.1 ± 0.2	3.1	nd	nd
A4S	9.0 ± 2.6	10.9 ± 0.2	7.0 ± 0.0	3.1	nd	nd
G93R	45.6 ± 1.8	20.0 ± 4.1	6.7 ± 0.1	3.5	nd	nd
A4V	37.2 ± 3.8	34.2 ± 13.3	6.4 ± 0.3	3.8	7.9 ± 0.7	7.2 ± 0.2
V148G	50.6 ± 1.4	76.8 ± 34.0	5.9 ± 0.3	4.3	7.3 ± 0.1	7.0 ± 0.1

na, not applicable; nd, not determined.

^aValues for each mutant are derived from fitting integrated raw ITC data to a dimer dissociation model [Eq. (1), Supporting Information and Fig. 2(B)] and are an average and standard deviation of 3–10 experiments

^bΔG _d values were obtained from K _d values using Eq. (3) (Supporting Information)

^cΔΔG _d = ΔG _d,pWT – ΔG _d,mutant; positive values indicate a decrease in interface stability for the mutant relative to pWT

^dData obtained by SEC. For pWT, G93A and H46R, heats of dissociation at 25°C were too small to be measurable by ITC, thus K _d at this temperature was measurable only by SEC. Values are the average of 3–6 experiments and error bars represent the 95% confidence intervals.

Figure 4.

Size exclusion chromatography reveals decreased dimer stability for SOD1 mutants. (A) SEC elution profile for pWT (black), I113T (green), and V148G (red) at 0.9 and 0.2 μM total dimer, shown as solid and dotted lines, respectively. (B) Percent native dimer (N₂) as a function of total dimer concentration, [N₂]_o. Data points are calculated from experimental elution volumes and fitted parameters using Eq. (7) (Supporting Information). Solid lines correspond to Eq. (6) (Supporting Information) and fitted values of K _d (Table I). (C) SEC data were measured as a function of total dimer concentration, [N ₂]_o. The linear transformation of the data [Eq. (9), Supporting Information] gives the fitted K _d values (Table I) at Y = 1 as indicated by the arrows. Symbols with error bars represent the average and 95% confidence intervals for 3–6 SEC measurements.

The dimer interface is markedly less stable at 37°C than at lower temperature [Figs. 2(D), 3(B), and Table 1]. The dissociation is endothermic [Fig. 3(A,C)], and measurements as a function of temperature reveal that it has a large positive ΔC _p,d [Fig. 3(A)]. Thus, the dissociation enthalpy is strongly temperature‐dependent, and dissociation is increasingly favorable and measurable by ITC at higher temperature due to the increased heat of dissociation.

Diverse fALS‐associated mutations weaken the apo SOD1 dimer interface

The ITC and SEC measurements clearly demonstrate a common effect of weakening of the apo SOD1 dimer interface by diverse ALS‐associated mutations located throughout the protein structure (Fig. 1). The ITC data for all the apo SOD1 variants are well fit by a 2‐state dimer dissociation model (N₂ ↔ 2M) [see Supporting Information, Fig. 2(B)]. The ΔH _d and K _d values from the fits and the corresponding ΔG _d values are summarized in Figures 2(C), 3(B,C), and Table 1. At 37°C, the pWT apo SOD1 dimer interface has a relatively high affinity, with a K _d of 67 nM and ΔG _d of 10.3 kcal (mol dimer)⁻¹. All mutations increase the K _d at 25 and 37°C but to different extents [Fig. 2(C,D)], corresponding to decreases in interface stability of 1.3–4.3 kcal (mol dimer)⁻¹ at 37°C [Fig. 3(B) and Table 1]. The mutations are more destabilizing at 37°C than at lower temperature, resulting in a significant population of monomer at physiological temperature and protein concentration (∼40 µM;26 ∼2% for pWT to ∼50% for V148G).

Notably, the effects of mutations on the integrity of the dimer interface cannot be predicted readily based on their structural contexts. In general we find mutations in or near the dimer interface, such as A4V, I113T, and V148G, tend to be the most destabilizing, but distant mutations, notably G93R, can also markedly weaken the interface [Figs. 1 and 2(C)]. Increases in K _d may be caused by short‐ or long‐ range perturbation of the interface region or by perturbations to regions distant from the interface that undergo structural changes upon dissociation. These effects may be related to the high connectivity of the interface to the rest of the protein structure and changes in coupled motions observed by computational modelling.27 The generality of enhanced dimer dissociation is supported by the modelling of many mutants and more limited experimental measurements of stability under nonphysiological conditions.13, 17, 25, 28 Chemical denaturation measurements for several mutants have shown that weakening of the dimer interface in apo SOD1 is not observed in the holo protein.16 Together these results indicate that the effects of fALS‐associated mutations may commonly weaken the dimer interface of apo SOD1.

Similar long range effects of mutations on interfaces have been reported for some other oligomeric proteins.29, 30, 31 In particular, inherited mutations in transthyretin (TTR) are proposed to cause familial amyloid polyneuropathy (FAP) by inducing conformational changes that promote dissociation of the native tetramer, leading to formation of partially folded monomers that can self‐assemble into amyloid fibrils.31 For SOD1, the finding that all the mutations studied destabilize the interface is especially significant in light of antibody binding studies that identified exposure of the SOD1 dimer interface in aggregates from ALS patients,32, 33 implicating dimer dissociation as a common occurrence in disease. The propagating effects of many mutations leading to increased population of monomer support the proposal that monomer formation is a key contributor to protein misfolding pathology in ALS.

Thermodynamics reveal apo SOD1 dimer dissociation is accompanied by extensive disruption of structure

The thermodynamic data also provide insights into the structural changes accompanying apo SOD1 dissociation. The measured enthalpies [Fig. 3(C) and Table 1] are much (up to 20‐fold) larger than expected values calculated using empirical models based on surface area changes between the crystal structure of the SOD1 dimer and its constituent monomer.23, 34 In addition, the value of ΔC _p,d [1.7 ± 0.2 kcal (mol dimer)⁻¹°C ⁻¹, Fig. 3(A)] is relatively large; the large values of ΔC _p,d and ΔH_d indicate a substantial increase in exposure of hydrophobic residues upon dissociation.35 For protein folding, ΔC _p can be calculated fairly accurately based on changes in solvent accessible surface area.22, 23, 34 Applying this approach, and assuming no structural change in the monomer upon dissociation, the estimated ΔC _p,d for SOD1 is ∼0.5 kcal (mol dimer)⁻¹°C⁻¹, much lower than the experimental value. Notably, large ΔH _d and ΔC _p,d values have been reported for some other proteins, and were explained by significant protein conformational changes and/or changes in solvent binding upon subunit dissociation.21, 22, 23, 36 The total ΔC _p for unfolding of the apo SOD1 dimer to unfolded monomers, Δ $C_{{p,\mathrm{N}}_2}$ _↔2U, has been measured previously as 3.3 ± 0.8 kcal (mol dimer)⁻¹°C⁻¹;37 combining this value with ΔC _p,d, the change in heat capacity for monomer unfolding, ΔC _{p , M↔U} (={Δ $C_{{p,\mathrm{N}}_2}$ _↔2U − ΔC _p,d}/2), can be calculated as 0.8 ± 0.4 kcal (mol monomer)⁻¹ °C⁻¹. These values of ΔC _p,d and ΔC _{p , M↔U} are consistent with measurements of the denaturant‐dependence of dimer dissociation and monomer unfolding for apo SOD1 which, similar to changes in heat capacity, are roughly correlated with changes in solvent exposed surface area.17 Thus, the relatively large ΔC _p,d and small ΔC _{p , M↔U} reveal significant disruption of the compact structure of the apo SOD1 dimer upon dissociation to form structurally dynamic monomers (discussed further below).

Additional mechanistic insights into the process of consolidation of the dimer interface during SOD1 maturation may be obtained from further consideration of the dissociation thermodynamic properties combined with the results of structural and folding studies. For all the SOD1 variants, dimer dissociation is enthalpically unfavorable and entropically driven at 37°C [Fig. 3(C)], suggesting increased protein entropy upon dissociation overcomes the loss of solvent entropy due to increased exposure of hydrophobic residues.38, 39 The solution NMR structure of an engineered monomeric apo SOD1 variant containing multiple mutations in the interface that abolish dimerization shows substantial disruption of structure compared to the dimer.40 Also, chemical denaturation m values indicate that the pWT apo SOD1 monomer involves lower than typical burial of solvent exposed surface area upon folding, and the monomer becomes considerably more compact in the presence of stabilizing Na₂SO_4. 17 Moreover, most of the substantial stability of dimeric apo SOD1 originates from the formation of the dimer interface, which is only slightly less stable than in the holo form of the protein,16, 17 while the dissociated apo SOD1 monomer has only marginal stability.17, 25 Thus, the direct measurements of dissociation thermodynamics determined here by ITC are consistent with these previous results obtained under nonphysiological conditions and provide strong evidence for extensive structural disruption upon dissociation of the apo SOD1 dimer. The thermodynamic data also provide information on the mechanisms underlying the natural maturation of SOD1: due to the loose packing of the apo monomer, dimerization confers strong stabilization as well as protection against aggregation by structuring regions of the protein that are highly dynamic in the monomer.

Molecular origin of high sensitivity of the apo SOD1 dimer interface to mutation

The particular sensitivity of the SOD1 dimer interface to disruption by mutations as demonstrated here can be rationalized by additional consideration of the structural and dynamic properties of the interface. The SOD1 interface is not evolutionarily conserved,41 is relatively small (∼10% of monomer surface area) and has high hydrophobicity (∼80% nonpolar residues).42 The small size and high hydrophobicity suggest a low extent and selectivity of monomer–monomer interaction so that association may be easily perturbed by mutation. Indeed, significant changes in the interface were observed in structural studies of A4V and I113T.12 In addition, owing to the small size of the interface, there may be little variation possible for stable orientations of associating monomers.43 Based on computational modelling, the dimer interface and metal binding sites in SOD1 are highly connected,44, 45 and coupled motion between SOD1 monomers is lost in A4V, G37R and H46R dimers.27, 46 These computational studies point to long‐range structural effects caused by mutations that can disrupt proper interface contacts.13, 45, 47 Furthermore, the network of stabilizing interactions that form when metals are bound to SOD1 constrains the flexibility of loops IV and VII (Fig. 1); but, in the absence of metal these loops gain conformational freedom,48, 49, 50 and have been proposed to create energetic frustration in the apo form of the protein.51 These regions of enhanced dynamics, notably in loop IV which forms part of the dimer interface, may be easily perturbed by mutations and compromise the interface stability. Thus, the common weakening of the dimer interface of apo SOD1 with temperature and mutation as measured here rationalizes other experimental and a range of computational results demonstrating the susceptibility of the interface to disruption by modest changes in the covalent structure of the protein. Such disruption may occur not only due to fALS‐associated mutations but also upon covalent modifications of wild‐type SOD1 observed in sALS.52, 53

Conclusions

In summary, we report here ITC and SEC measurements that reveal common destabilization of apo SOD1 caused by diverse mutations whose effects manifest at the dimer interface. To date, relatively few homodimeric proteins have been characterized by ITC, and accurate prediction of the energetics and mechanisms of protein binding remains very challenging.4, 5, 21, 23, 36 Such predictions are particularly difficult when association also involves significant protein structural changes.4, 5, 54 The results reported here show how quantitative analysis of protein interface stability by ITC can provide valuable data both for fundamental understanding and modelling of protein–protein interactions as well as for understanding disease.

Many studies have attempted to correlate various properties of mutant SOD1 with ALS characteristics. A correlation between total protein stability (i.e., combined monomer and dimer interface stability) and disease duration has been reported; however, the data are very scattered.55, 56, 57 Thus, mutational effects on total stability are only part of the story. Our results show that dimerization and monomer stabilities are additional key characteristics to consider for unravelling the origins of toxic effects of SOD1 in ALS.

Supporting information

Supporting Information

Supporting Information