A misfolded dimer of Cu/Zn‐superoxide dismutase leading to pathological oligomerization in amyotrophic lateral sclerosis

Abstract

Misfolding of mutant Cu/Zn‐superoxide dismutase (SOD1) is a pathological hallmark in a familial form of amyotrophic lateral sclerosis. Pathogenic mutations have been proposed to monomerize SOD1 normally adopting a homodimeric configuration and then trigger abnormal oligomerization of SOD1 proteins. Despite this, a misfolded conformation of SOD1 leading to the oligomerization at physiological conditions still remains ambiguous. Here, we show that, around the body temperature (∼37°C), mutant SOD1 maintains a dimeric configuration but lacks most of its secondary structures. Also, such an abnormal SOD1 dimer with significant structural disorder was prone to irreversibly forming the oligomers crosslinked via disulfide bonds. The disulfide‐crosslinked oligomers of SOD1 were detected in the spinal cords of the diseased mice expressing mutant SOD1. We hence propose an alternative pathway of mutant SOD1 misfolding that is responsible for oligomerization in the pathologies of the disease.

Introduction

Mutations in Cu/Zn‐superoxide dismutase (SOD1) gene have been shown to cause a familial form of amyotrophic lateral sclerosis (fALS),1 in which misfolding of mutant SOD1 proteins is observed as a major pathological hallmark.2 A wild‐type SOD1 protein binds copper and zinc ions and also forms an intramolecular disulfide bond to perform an enzymatic reaction removing the toxicity of superoxide radicals (Fig. 1).3, 4 An enzymatically active form of wild‐type SOD1 is a tight homodimer (K _d ∼ 0.1 nM)5 and exhibits quite high thermostability (T _m > 90°C).6 In contrast, pathogenic mutations are known to compromise the thermostability of SOD1,7, 8 reduce the affinity toward copper/zinc ions,9 and/or promote the dimer dissociation,10 thereby triggering the misfolding of SOD1.

Figure 1.

A crystal structure of human Cu/Zn‐superoxide dismutase (SOD1). SOD1 binds copper (cyan) and zinc (pink) ions and forms an intramolecular disulfide bond (S‐S, yellow). Together with the ligands for metal ions, Cys 6 and 111 (yellow) and Trp 32 (red) are also shown with stick models.

The folding process of SOD1 has been extensively characterized in vitro by the analysis of its thermodynamics and kinetics. For example, the thermal unfolding process of SOD1 was examined with differential scanning calorimetry (DSC) (e.g.,11), and also, the refolding and unfolding kinetics as well as the equilibrium denaturation curves of SOD1 have been fully evaluated by using chemical denaturants (e.g.,12, 13, 14). In all of those previous studies, the three‐state model has described the folding process of SOD1, where monomerization of a SOD1 homodimer is assumed to precede unfolding. In such a folding framework, pathogenic mutations have been suggested to shift the equilibria toward the monomeric intermediate, which could hence be regarded as a “misfolded” form leading to the oligomerization. Indeed, formation of abnormal SOD1 oligomers has been observed in the in vitro conditions where the monomerization is favored; for example, in the solution at acidic pH,15 with chemical denaturants,16 with oxidative damages,17 and at raised temperature.18 Molecular dynamics simulations on SOD1 have also predicted destruction of the dimeric configuration by increasing temperature and similarly by fALS mutations.19, 20, 21 Furthermore, the spinal cord extracts of the transgenic mice expressing human SOD1 with a pathogenic mutation, G85R, were separated by gel filtration chromatography, which revealed the presence of monomeric SOD1 species in vivo.22, 23 Therefore, mutation‐induced monomerization of SOD1 has been assumed as an important step for the formation of oligomers and aggregates observed in the disease.

The structural and dynamical properties of SOD1 potentially leading to the misfolding have also been investigated by nuclear magnetic resonance (NMR) (e.g.,24, 25, 26, 27 as well as X‐ray crystallography, e.g.,28). The NMR analysis has revealed the significantly increased fluctuation at the loop regions of SOD1 upon dissociation of the metal ions in vivo 26 as well as in vitro.25 Such increased fluctuations would allow SOD1 to thermally access to conformers that are prone to forming non‐native associations. The crystal structures of metal‐deficient SOD1 have also supported abnormal protein–protein interactions leading to higher‐order arrays of SOD1 molecules.28 Nonetheless, the significant disorder in SOD1 at 25°C and above hampers detection and assignment of the specific resonances for the structural analysis in NMR and also prevents the crystallization.24 Furthermore, in the NMR analysis and the crystal structures, no clear evidence has been provided to support the mutation‐induced monomerization of SOD1 for subsequent oligomerization. Accordingly, the conformations of misfolded SOD1 still remain obscure in near‐physiological conditions.

In that sense, the small‐angle X‐ray scattering (SAXS) is a powerful tool for the analysis of protein conformations, albeit with the lower resolution compared to those in NMR and crystal structures. Even if a protein of interest exhibits structural disorder and oligomerization propensity, the SAXS analysis can provide valuable information on its molecular shape, size, and also relative molecular weight. Indeed, SAXS analysis on mutant SOD1 has been performed albeit at relatively low temperatures (4–16°C) and revealed several non‐native conformations even in oligomeric configurations.29, 30, 31 In this study, therefore, we have examined the thermal unfolding of human SOD1 by SAXS combined with spectroscopic/biochemical experiments at various temperatures and attempted to clarify the misfolded conformation of SOD1 that leads to the pathological oligomerization of mutant SOD1 in the disease.

Results

Thermal denaturation of SOD1 characterized by DSC

To characterize the thermal denaturation of SOD1, we re‐investigated the melting temperature, T _m, of metal‐deficient SOD1 with the intramolecular disulfide bond (apo‐SOD1^S‐S) by utilizing DSC. Also, we examined SOD1 with C6S/C111S mutations (pSOD1) to avoid aberrant disulfide crosslinks during DSC measurements.32 As shown in Figure 2(A,B), a melting temperature, T _m, of apo‐pSOD1^S‐S decreased from 51.1°C to 43.3°C by introduction of a pathogenic mutation (G37R), which is completely consistent with many previous studies (e.g.,8). The observed endothermic transition was not, however, well described by a single Gaussian curve [Fig. 2(C)], when the pre‐ and post‐transition baseline slopes were set as shown in Figure 2(A,B). Indeed, some of previously reported thermograms of apo‐SOD1^S‐S also exhibited an asymmetric and non‐single Gaussian shape.8, 33 As described in Introduction, the unfolding of SOD1 with chemical denaturants has been proposed to proceed through the three‐state model. We thus supposed that the asymmetric thermograms of apo‐pSOD1^S‐S may indicate the existence of a distinct “intermediate” (I) state during the thermal unfolding of apo‐pSOD1^S‐S from its folded (F) to unfolded (U) states. Notably, apo‐pSOD1^S‐S with G37R mutation at the body temperature (∼37°C) is in the middle of the premelting endothermic process [Fig. 2(C)]. For understanding the misfolding of mutant SOD1 in the disease, therefore, it is important to characterize structural changes of apo‐pSOD1^S‐S possibly occurring in its pre‐melting endothermic process.

Figure 2.

DSC thermograms of apo‐pSOD1^S‐S (A, B) Thermograms of (A) apo‐pSOD1^S‐S and (B) G37R apo‐pSOD1^S‐S in the NNE buffer were shown. The protein concentration was 1.0 g L⁻¹ (ca., 60 μM). The heat capacity values, C _p, both in the folded and unfolded states were dependent upon temperature, which was set by the lines shown in the figure. By using those lines, the baseline (a broken curve) was estimated. (C) Thermograms of apo‐pSOD1^S‐S (open circles) and G37R apo‐pSOD1^S‐S (filled circles) after correction with the baselines shown in (A) and (B). We attempted to fit the corrected thermograms with a single Gaussian function (dotted curves), but the data were not well fit by the Gaussian function at the temperature below the T _m (see text).

A locally disordered conformation in the folding intermediate of apo‐SOD1^S‐S

To monitor the structural changes in apo‐pSOD1^S‐S during its thermal denaturation, we took advantage of a single Trp (Trp32) in human SOD1 located in a β‐barrel core structure (Fig. 1). Trp is a useful spectroscopic probe for detecting changes in its local environment, and in general, the wavelength of a Trp fluorescence emission becomes red‐shifted upon protein unfolding.34 Indeed, the emission wavelength of the fluorescence was red‐shifted upon thermal denaturation of apo‐SOD1^S‐S and G37R apo‐pSOD1^S‐S with a single transition at 50.6 and 39.4°C, respectively [Fig. 3(A)]. These values are close to the T _m values obtained with DSC [Fig. 2(A,B)], but unlike the DSC thermograms, the pre‐melting F‐to‐I process was not evident in the temperature dependence of the fluorescence emission wavelength [Fig. 3(A)].

Figure 3.

Effects of temperature on the local structure of apo‐SOD1^S‐S. (A) A ratio of the Trp fluorescence intensities at 365 and 345 nm (F ₃₆₅/F ₃₄₅) was plotted against temperature (left axis): 5 μM apo‐pSOD1^S‐S (open circles) and 5 μM G37R apo‐pSOD1^S‐S (filled circles). The measurements were repeated three times to estimate error bars (standard deviation). The mean residue molar ellipticities at 280 nm of apo‐pSOD1^S‐S (red open circles) and G37R apo‐pSOD1^S‐S (red filled circles) shown in (B) were also plotted against temperature (right axis). (B) The near‐UV CD spectra of (right) 400 μM apo‐pSOD1^S‐S and (left) 400 μM G37R apo‐pSOD1^S‐S in 10 mM Na‐Pi, 100 mM NaCl at pH 7.4 were shown at temperatures from 10 to 60°C in a step of 10°C. (C–E) Modification of Cys 6 with mPEG‐MAL was performed at indicated temperatures using (C) C111S apo‐SOD1^S‐S and (D) G37R/C111S apo‐SOD1^S‐S in the NNE buffer. After modified with mPEG‐MAL, the samples were analyzed with SDS‐PAGE using a 12.5% polyacrylamide gel and stained with Coomassie Brilliant Blue. (E) The band intensities of modified and unmodified SOD1 in (C) and (D) were measured, and the fraction (%) of the modified SOD1 was plotted against temperature: C111S apo‐SOD1^S‐S (open circles) and G37R/C111S apo‐SOD1^S‐S (filled circles) at the protein concentration of either 20 μM (black) or 300 μM (red).

An aromatic amino acid is also known to exhibit distinct CD signals in the near‐UV region that are sensitive to protein tertiary structures.35 As shown in Figure 3(B), negative near‐UV CD signals were observed in apo‐pSOD1^S‐S proteins but became lost upon increasing temperature. This is consistent with the destruction of the tertiary structure upon thermal denaturation and also correlates well with the temperature dependence of the emission wavelength of Trp fluorescence [Fig. 3(A)]. Again, no obvious pre‐melting F‐to‐I process was confirmed in the temperature dependence of near‐UV CD signals. We thus supposed that environments around Trp32 were not significantly affected during the transition process from F to I states in the thermal denaturation of apo‐SOD1^S‐S.

To get further information on the local structural changes in apo‐SOD1^S‐S during its thermal denaturation, we also noted Cys residues in SOD1 and examined their accessibility to a thiol‐specific modifier. Among four Cys residues of SOD1, Cys6 and Cys111 are available for thiol‐specific modifications, while Cys57 and Cys146 are not due to the formation of the disulfide bond (Fig. 1). We thus introduced C6S and C111S mutations in apo‐SOD1^S‐S to examine the reactivity of Cys111 and Cys6 with a thiol‐specific reagent, maleimide‐functionalized polyethylene glycol (mPEG‐MAL), respectively.

The modification at Cys111 with mPEG‐MAL was examined in C6S apo‐SOD1^S‐S with and without G37R mutation and was found to efficiently proceed even at 20°C (Supporting Information Fig. S1). This is consistent with the exposure of Cys111 even in the folded structure of SOD1 (Fig. 1). In contrast, by using C111S apo‐SOD1^S‐S with and without G37R mutation, almost no modification of Cys6 with mPEG‐MAL was detected at relatively low temperatures (∼20°C) [Fig. 3(C–E)]. This is probably because Cys6 is buried in the structural interior of a folded SOD1 protein (Fig. 1). At elevated temperatures, increasing amounts of C111S apo‐SOD1^S‐S became modified with mPEG‐MAL, and the modification almost completed at 46°C (C111S) and 38°C (G37R/C111S) [Fig. 3(C–E)]. These results suggest the exposure of Cys6 during the thermal denaturation of SOD1, but notably, the modification efficiently proceeded at the temperature range well below the T _m characterized by DSC [Figs. 2(A) and 3(B)]. Because Cys6 is located near the dimer interface (Fig. 1), the protein concentration might be an important factor in the efficiency of modification at Cys6. Contrary to such expectation, however, the modification at Cys6 was not significantly affected by the protein concentration (20–300 μM) [Fig. 3(E)]. Rather, the structural changes that associate with the exposure of Cys6 are supposed to occur in the temperature range corresponding to the F‐to‐I transition process [Figs. 2(C) and 3(E)].

Significant loss of secondary structural contents in the I state of SOD1

We further examined secondary structural changes in apo‐SOD1^S‐S during its thermal denaturation by measuring CD spectra in the far‐UV region at various temperatures.36 As shown in Figure 4(A), apo‐pSOD1^S‐S proteins at 10°C exhibited far‐UV CD spectra with a negative peak at 210 nm and a distinct positive shoulder around 230 nm. By increasing the temperature, the negative peak shifted from 210 to 200 nm, and the distinct positive shoulder became vague [Fig. 4(A)]. These CD spectral changes suggest the decreased contents of β‐strands and the formation of random coils.37 In both apo‐pSOD1^S‐S and G37R apo‐pSOD1^S‐S, furthermore, the CD spectra at different temperatures were found to intersect around 210 and 220 nm [Fig. 4(A)]. Assuming a simple two‐state denaturation (F <‐> U), the molar ellipticities at the wavelength where the spectra intersect should be identical between those two states and remain constant throughout the temperatures (i.e., isodichroic points). As shown in Figure 4(A), however, no isodichroic points were confirmed [Fig. 4(A)]. The spectral changes of G37R apo‐pSOD1^S‐S around 210 nm appear to be isodichroic, which was actually not the case [inset in the right panel of Fig. 4(A)]. These results thus support the involvement of the intermediate state(s) other than F and U states in the thermal denaturation of apo‐pSOD1^S‐S and its G37R mutant form.

Figure 4.

Effects of temperature on the secondary structural content of apo‐SOD1^S‐S. (A) Far‐UV CD spectra of (left) 20 μM apo‐pSOD1^S‐S and (right) 20 μM G37R apo‐pSOD1^S‐S in 10 mM Na‐Pi, 100 mM NaCl, pH 7.4, at temperatures from 10 to 60°C in a step of 2°C. Direction of spectral changes by increasing temperature was represented by arrows. The CD spectral changes of G37R apo‐pSOD1^S‐S around 210 nm were enlarged and shown in the inset of the right panel. (B) A ratio of CD at 210 and 200 nm, θ ₂₁₀/θ ₂₀₀, was calculated from the spectra shown in (A) and plotted against temperature: apo‐pSOD1^S‐S (open circles) and G37R apo‐pSOD1^S‐S (filled circles). Measurements were repeated three times to estimate errors (standard deviation).

To evaluate the secondary structural changes of apo‐pSOD1^S‐S during thermal denaturation, we introduced a ratio of the CD signals at 200 and 210 nm (θ ₂₁₀/θ ₂₀₀), which reflects the relative content of random coils and β‐strands.36 As shown in Figure 4(B), the temperature‐dependent decrease of the ratio, θ ₂₁₀/θ ₂₀₀, in apo‐pSOD1^S‐S implies the transition from β‐strands to random coils upon the thermal denaturation. Notably, furthermore, the decrease in the secondary structural content was found to almost complete at 48 and 37°C in apo‐pSOD1^S‐S without and with G37R mutation, respectively. These temperatures were again below the T _m values measured by DSC [Fig. 2(A,B)] and quite similar to those observed in the modification of Cys6 [Fig. 3(E)]. We hence suppose that apo‐SOD1^S‐S loses its secondary structures in conjunction with the exposure of Cys6 during the pre‐melting endothermic process from F to I state.

SOD1 maintains a compact and globular conformation in the I state

Based upon the secondary structural contents and the exposure of Cys6, the I state of apo‐SOD1^S‐S cannot be distinguished from the U state [Figs. 3(E) and 4(B)]. The I state is, however, supposed to be distinct from the U state in that the transition from the I to U state is endothermic with the environmental changes around Trp32 [Figs. 2(C) and 3(A)]. To clarify the conformational features of apo‐SOD1^S‐S in the I state, we hence examined SOD1 proteins with SAXS at various temperatures.

The observed scattering curves of apo‐pSOD1^S‐S proteins were summarized in Supporting Information Figure S2. The radius of gyration (R _g) is one of the conformational parameters of the proteins that can be estimated by the Guinier analysis of the SAXS curves (Supporting Information Figs. S3 and S4). At relatively low temperatures, the R _g value was found to be almost constant at 20.9 Å in both apo‐pSOD1^S‐S proteins [Table 1, Fig. 5(A)]. This was comparable with the one of the crystal structure of a SOD1 dimer (PDB ID 2C9V, 20.9 Å), which was calculated using a program, CRYSOL.38 Further elevation of the temperature (>50°C and >40°C without and with G37R mutation) led to the increase in the R _g values, which was consistent with the thermal denaturation of apo‐pSOD1^S‐S [Table 1, Fig. 5(A)]. Significantly large R _g (32.4 Å) was observed in G37R apo‐pSOD1^S‐S at 60^°C but was dropped to 28.0 Å upon dilution of the protein concentration (from 5.0 to 2.5 g/L) [Table 1, Fig. 5(A)]. Almost no concentration dependence of R _g was confirmed at the lower temperatures [Fig. 5(A)]; therefore, the U state of G37R apo‐pSOD1^S‐S would tend to form a multimeric form. Accordingly, the temperature‐dependence of the R _g values can be described by the thermal denaturation of apo‐pSOD1^S‐S; however, it is quite notable that the R _g value was unaffected (20.9 Å) in the range where the secondary structural contents were dropped [35–50°C and 25–40°C without and with G37R mutation; Figs. 4(B) and 5(A)]. In other words, in the transition from F to I state, apo‐SOD1^S‐S is considered to lose significant amounts of secondary structures but maintain a compact conformation with R _g similar to that of the F state.

Table 1.

SAXS Structural Parameters

apo‐pSOD1S‐S (5.01 g L−1)				G37R apo‐pSOD1S‐S (5.28 g L−1)
Temperature	R g a	I(0)a	M r b	Temperature	R g a	I(0)a	M r b
(°C)	(Å)	(a.u.)	(kDa)	(°C)	(Å)	(a.u.)	(kDa)
10	20.9 ± 0.1	1.63 ± 0.01	34.3	10	20.9 ± 0.1	1.65 ± 0.01	34.7
20	20.6 ± 0.1	1.55 ± 0.01	32.6	20	20.9 ± 0.1	1.58 ± 0.01	33.2
30	20.8 ± 0.1	1.52 ± 0.01	31.9	30	20.9 ± 0.1	1.55 ± 0.01	32.6
40	20.9 ± 0.1	1.50 ± 0.01	31.5	38	21.9 ± 0.2	1.59 ± 0.01	33.4
48	20.3 ± 0.1	1.43 ± 0.01	30.0	40	21.2 ± 0.2	1.42 ± 0.01	29.8
50	20.8 ± 0.1	1.32 ± 0.01	27.7	42	23.3 ± 0.2	1.49 ± 0.01	31.3
52	20.6 ± 0.1	1.20 ± 0.01	25.2	44	24.0 ± 0.2	1.28 ± 0.01	26.9
54	21.8 ± 0.1 (20.8 ± 0.4)c	1.03 ± 0.01 (0.90 ± 0.01)c	21.6 (18.9)c	46	25.9 ± 0.3 (24.0 ± 0.6)d	1.14 ± 0.01 (0.93 ± 0.01)d	24.0 (19.5)d
56	22.9 ± 0.1	0.84 ± 0.01	17.7	48	26.5 ± 0.4	1.00 ± 0.01	21.0
58	25.6 ± 0.1	0.89 ± 0.01	18.7	50	29.2 ± 0.5	1.01 ± 0.01	21.2
60	25.1 ± 0.1 (26.9 ± 0.4)c	0.79 ± 0.01 (0.90 ± 0.02)c	16.4 (18.9)c	60	32.4 ± 0.6 (28.0 ± 0.8)d	1.14 ± 0.02 (0.96 ± 0.02)d	24.0 (20.2)d

^aGuinier analysis using the Q range from 0.02217 to Q _max < 1.3/R _g.

^b M _r, molecular mass calculated by using I(0) value for BSA as the standard.

^cAt concentration of 2.66 g L⁻¹.

^dAt concentration of 2.52 g L⁻¹.

Figure 5.

The I state of apo‐SOD1^S‐S maintains a compact and globular conformation. (A, B) (A) The radius of gyration, R _g, and (B) molecular mass, M _r, were calculated by the Guinier analysis on the scattering curves (Supporting Information Figs. S3 and S4) and plotted against temperature: 5.01 g L⁻¹ (ca., 300 μM) apo‐pSOD1^S‐S (black open circles) and 5.28 g L⁻¹ (ca., 300 μM) G37R apo‐pSOD1^S‐S (red open circles) in the NNE buffer. The data obtained by using 2.66 g L⁻¹ (ca., 160 μM) apo‐pSOD1^S‐S and 2.52 g L⁻¹ (ca., 160 μM) G37R apo‐pSOD1^S‐S were shown as black and red filled triangles, respectively. (C, D) The Kratky plots of the SAXS curves obtained using (C) apo‐pSOD1^S‐S and (D) G37R apo‐pSOD1^S‐S in the temperature from 10 to 60°C. The plots at 50°C in apo‐pSOD1^S‐S and at 40°C in G37R apo‐pSOD1^S‐S were colored red (inset). The plot of the peak intensity in the Kratky analysis (at 0.084 Å⁻¹ of Q) against temperature.

A relative molecular mass, M _r, of the protein provides valuable information on the protein conformation in solution and can also be estimated by the Guinier analysis of the SAXS curves (Supporting Information Figs. S3 and S4). At relatively low temperatures, the M _r values of apo‐pSOD1^S‐S were almost constant at 32 kDa, which corresponds to the molecular mass of a SOD1 dimer [Table 1, Fig. 5(B)]. By increasing the temperature, the estimated M _r values of apo‐pSOD1^S‐S decreased to the molecular mass of a SOD1 monomer (16 kDa), implying the protein monomerization upon the thermal denaturation. The increase in the M _r of G37R apo‐pSOD1^S‐S from 50 to 60^°C was probably due to the formation of multimeric species, because the M _r was decreased by diluting the protein concentration from 5.0 to 2.5 g L⁻¹ [Table 1, Fig. 5(B)]. Taken together, apo‐pSOD1^S‐S was found to remain dimeric in the transition from F to I states.

A compact conformation of apo‐SOD1^S‐S proteins was also confirmed by the Kratky analysis of the SAXS curves. As shown in Figures 5(C,D), a distinct peak was observed at the angular momentum (Q, 0.084 Å⁻¹) in relatively low temperatures, indicating significant globularity of apo‐pSOD1^S‐S. The thermal denaturation of apo‐pSOD1^S‐S was then found to associate with the disappearance of the distinct peak in the Kratky plots [Fig. 5(C,D)]. The remaining peak observed even at 60°C [Fig. 5(C,D)] was probably due to the conformational restriction by the disulfide bond between Cys57 and Cys146.

It is important to note that the monomer–dimer equilibrium depends upon the total concentration of apo‐pSOD1^S‐S. Actually, we performed the SAXS experiments by using significantly higher protein concentrations (5.0 g L⁻¹; ca., 300 μM) than those in CD spectroscopic experiments (20 μM), which might lead to the distinct dependencies of R _g, M _r, and secondary structural contents on temperatures. We hence repeated SAXS experiments at 37°C by using apo‐pSOD1^S‐S in the diluted concentration (0.5 g L⁻¹; ca., 30 μM) and confirmed that the scattering curves of 5.0 and 0.5 g L⁻¹ apo‐pSOD1^S‐S at 37°C were completely overlapped (Supporting Information Fig. S5). Based upon those results, therefore, we suggest that the following two steps occur in the thermal denaturation of apo‐SOD1^S‐S: namely, in the transition from F to I states, the protein remains dimeric with a compact conformation but loses the secondary structural contents. Further increase in the temperature leads to the endothermic transition from I to U states, where the protein eventually becomes monomerized and unfolded.

SOD1 in the I state is susceptible to the formation of abnormal oligomers

At physiological temperatures, a disease‐causing mutation (G37R) was found to make apo‐SOD1^S‐S increasingly accessible to the I state. Mutant SOD1 has been found to form abnormal oligomers in the SOD1‐related fALS cases; therefore, we expected that apo‐SOD1^S‐S in the I state is prone to oligomerization. To test this, apo‐SOD1^S‐S (with Cys6 and Cys111 intact) proteins were incubated overnight at various temperatures and examined for the oligomerization with SDS‐PAGE. As shown in Figure 6(A), oligomers were detected as smears in the high‐molecular‐weight region (>29.0 kDa), and their formation was dependent upon the incubation temperature. More precisely, the oligomerization became evident above the temperature where apo‐SOD1^S‐S populate the I state (45 and 37°C without and with G37R mutation, respectively). These results thus support the oligomerization propensities of apo‐SOD1^S‐S in the I state.

Figure 6.

In vitro and in vivo formation of disulfide‐crosslinked SOD1 oligomers. (A,B) Twenty micromolar solutions of apo‐SOD1^S‐S and G37R apo‐SOD1^S‐S in the NNE buffer were incubated overnight at the indicated temperature. Followed by the treatment with IA, the samples (A) without and (B) with DTT were separated by SDS‐PAGE using a 12.5% polyacrylamide gel. (C) A soluble fraction of lumbar spinal cords of mice were examined for the disulfide‐crosslinked oligomers by SDS‐PAGE (lower) with and (upper) without β‐ME followed by Western blotting using anti‐SOD1 antibody. Non‐transgenic mice (NonTG) and mice expressing human SOD1 (WT, G93A, and G37R) were analyzed at the ages indicated: NonTG and WT mice were asymptomatic, 30‐day‐old G93A and 100‐day‐old G37R mice were pre‐symptomatic, 100‐day‐old G93A mice were symptomatic, and 160‐day‐old G93A and 370‐day‐old G37R mice were terminally ill. (D, E) Effects of redox environment on the oligomerization in vitro were examined at various temperatures in the presence of (D) 5 mM GSH and 0.05 mM GSSG and (E) 5 mM GSH and 1 mM GSSG. The experiments with non‐reducing SDS‐PAGE were performed as described in (A, B).

Furthermore, by addition of a reducing reagent, dithiothreitol (DTT), to the samples before electrophoresis, all of those smears were collapsed to a single band corresponding to the disulfide‐reduced SOD1 monomer [Fig. 6(B)]. These results indicate that the oligomerization proceeds through the formation of the intermolecular disulfide crosslinks. Indeed, the disulfide‐crosslinked SOD1 oligomers, which were sensitive to the treatment with a reducing reagent, were observed as smears in the lumber spinal cords of diseased mice expressing human SOD1 with G93A mutation (100, 160 days of age) and also with G37R mutation (370 days of age) [Fig. 6(C)]. No smears were instead observed in pre‐symptomatic G93A mice (30 days of age), pre‐symptomatic G37R mice (100 days of age), and asymptomatic mice expressing wild‐type human SOD1 as well as nontransgenic mice [Fig. 6(C)]. The pattern of high‐molecular‐weight smears/bands in the samples of diseased mice appeared to be distinct from those of purified protein samples in vitro [Fig. 6(A,C)], implying the involvement of the other proteins in the disulfide‐crosslinked complexes with SOD1. In the in vivo conditions, therefore, the dimeric I state of apo‐SOD1^S‐S might be also reactive to other Cys‐containing proteins to form oligomers sensitive to reductants.

Consistent with the crosslinking via disulfide bonds in the oligomers, no smears were observed when G37R apo‐pSOD1^S‐S, which had no free Cys residues (i.e., C6S/C111S), was incubated at 37^°C or higher (data not shown). Nonetheless, a large majority of intracellular SOD1s is localized in the reducing environment of the cytoplasm,39 where reduction instead of crosslinking of disulfide bonds will be favored. Actually, in the presence of 5 mM GSH/0.05 mM GSSG, which mimics the reducing environment of the cytoplasm, the oligomerization in vitro of apo‐SOD1^S‐S at 37^°C was almost completely suppressed, while smears were observed at higher temperatures (55^°C) [Fig. 6(D)]. SOD1 is also known to localize in the intermembrane space of the mitochondria,40 and a redox balance toward more oxidizing has been reported in the mitochondria of motorneuronal cells, where the ratio of the reduced and oxidized glutathione (GSH/GSSG) was about 5:1.41 We thus repeated our oligomerization experiments in the presence of 5 mM GSH/1 mM GSSG and confirmed the formation of the disulfide‐crosslinked oligomers in G37R apo‐SOD1^S‐S but not apo‐SOD1^S‐S [Fig. 6(E)]. These observations are consistent with the previous finding that the disulfide‐crosslinked oligomers were detected in the mitochondria of spinal motor neurons of fALS‐model mice.42 Based upon these results, we propose that pathogenic mutations increase the population of the I state of apo‐SOD1^S‐S at physiological temperatures and thereby trigger the disulfide‐crosslinked oligomerization as a pathological change in SOD1‐related fALS cases.

Discussion

A folding mechanism of SOD1 has been examined in several experimental conditions in vitro that can perturb the equilibria among protein folding states. In most of those previous studies, the folding of SOD1 has been assumed to proceed in the three‐state model, where the monomerization precedes the unfolding. Actually, our results also support a three‐state model; however, unlike the previous studies, the conformationally distinct intermediate in our model assumes a dimeric configuration.

For in vitro characterization of protein quaternary structures, a protein concentration is an important determinant. In most of the previous studies proposing a monomeric intermediate, the SOD1 concentrations have been examined in approximate range from 5 to 200 μM. Given that the dimer dissociation constant, K _d, of apo‐SOD1^S‐S at 37°C has been estimated as 0.067 and 5.0 μM without and with G37R mutation, respectively,10 monomerization will be relevant at the diluted concentrations of SOD1; for example, in a 5 μM solution of G37R apo‐SOD1^S‐S, a half of the proteins will exist as a monomer. Actually, at relatively high protein concentrations (>100 μM), the thermal denaturation of homodimeric apo‐SOD1^S‐S was sufficiently described with a two‐state model where the monomerization was not required.11 In this study, we confirmed that G37R apo‐SOD1^S‐S was dimeric in the range from 30 to 300 μM at 37^°C [Fig. 5(B), Supporting Information Fig. S5]. This is consistent with the back‐of‐the‐envelope calculation using the above K _d value (5.0 μM) (10); 75 and 95% of 30 μM and 300 μM G37R apo‐SOD1^S‐S, respectively, are expected to be in the dimer. Because the SAXS curves of G37R apo‐SOD1^S‐S at 37°C were almost completely matched between 30 and 300 μM (Supporting Information Fig. S5), more than 75% of 30 μM G37R apo‐SOD1^S‐S would actually populate a dimer. Despite this, our results here showed that such a G37R apo‐SOD1^S‐S dimer at 37°C has a distinct conformation from that observed at lower temperatures (10–20°C) [Figs. 3(E) and 4(B)]. In other words, in the course of the thermal unfolding, apo‐SOD1^S‐S is considered to first become structurally disordered but maintain significant globularity in the dimeric configuration (Fig. 7), and then, further elevation of temperature will lead to the unfolded state of apo‐SOD1^S‐S. We thus propose the alternative three‐state model for the thermal denaturation of apo‐SOD1^S‐S, in which a disordered but globular and compact dimer serves as a folding intermediate (the dimeric I state).

Figure 7.

A proposed mechanism of SOD1 oligomerization. Thermal denaturation of apo‐SOD1^S‐S from the folded (F) to the unfolded (U) state proceeds through the dimeric I state, which is characterized by globular conformations with increased structural disorder. At the body temperature (∼37°C), wild‐type SOD1 mostly exists as the F state, while the population of the I state is significantly increased by pathogenic mutations in SOD1. Disulfide‐shuffled oligomerization occurs in the I state but not the F state.

In our experimental conditions, we could not identify the monomeric intermediate previously predicted in the chemical/thermal denaturation of SOD1. The calorimetric data on 32 μM apo‐pSOD1^S‐S predicted a monomeric intermediate in the temperature range from 55 to 65°C or from 40 to 50°C in the presence of G37R mutation.11 In our current study, the temperature dependencies in the population of the F state and the U state are considered to be represented by those of the secondary structural content [Fig. 4(B)] and the peak intensity in the Kratky plot [Fig. 5(C,D)], respectively; we can thereby simulate the fractions of F, I, and U states at each temperature. As shown in Figure 7, our dimeric I state of apo‐SOD1^S‐S was considered to be dominant in the region from 40 to 50°C or from 30 to 40°C in the presence of G37R mutation, which are lower than that predicted for the monomeric intermediate. The monomeric intermediate might hence form from our dimeric I state upon increase of the temperature, but further attempts will be definitely required to confirm the predicted monomeric intermediates. Compared to the monomeric intermediate, nonetheless, we would like to emphasize that our dimeric I state of mutant SOD1 is more easily accessible in the physiological range of temperatures (Fig. 7).

As revealed by CD spectroscopy as well as chemical modification, the dimeric I state of apo‐SOD1^S‐S is characterized by its structural disorders (Figs. 3 and 4). Actually, many previous studies already revealed the temperature‐dependent structural disorder of apo‐SOD1^S‐S. For example, by monitoring the amide hydrogen/deuterium exchange in SOD1 with mass spectrometry, the distinct regions of SOD1 were found to unfold at different temperatures, resulting in the formation of a partially unfolded β‐barrel structure as an intermediate of the thermal unfolding.43, 44 The dynamical properties of apo‐SOD1^S‐S have also been examined at several temperatures (15, 25, and 37^°C) by NMR, showing that higher temperatures accelerate the exchange among a range of distinct conformations in apo‐SOD1^S‐S.24 Apo‐SOD1^S‐S has been shown to successfully crystallize at 25^°C but not at 37^°C, further supporting the structural disorder at higher temperatures.24 In spite of such accumulated information on the structures of thermally affected SOD1 species, the conformational analysis has not been fully examined on the folding intermediate during the thermal denaturation of SOD1. In that sense, our SAXS analysis here has for the first time revealed that the folding intermediate has a compact and globular conformation with the dimeric configuration.

As described above, we have identified the dimeric I state in the thermal denaturation of apo‐SOD1^S‐S, but significance of the dimeric configuration in the SOD1 misfolding needs to be further discussed. Actually, our proposed mechanism appears not to be consistent with the previous in vivo observation that G85R SOD1 was isolated predominantly as a monomer from the spinal cord extracts of the transgenic mice.22, 23 Nonetheless, it should be noted that G85R SOD1 exist as the disulfide‐reduced state in the transgenic mice22, 23 and also that reduction of the disulfide bond in SOD1 facilitates the monomerization.45 Actually, the other mutant SOD1s (D90A and G93A) expressed in the transgenic mice were equipped with the intramolecular disulfide bond and were predominantly dimeric in their non‐native fractions of the spinal cord extracts.22 Furthermore, it is notable that the dimeric I state but not the F state of apo‐SOD1^S‐S is converted to disulfide‐crosslinked oligomers (Fig. 6). Previously, we proposed that the oligomerization of apo‐SOD1^S‐S occurs through the shuffling of a disulfide bond among SOD1 proteins.46 As schematically shown in Figure 7, the conformational destabilization of SOD1 can allow its free thiolate groups at Cys6 and Cys111 to nucleophilically attack the canonical disulfide bond between Cys57 and Cys146, resulting in the formation of the disulfide‐crosslinked oligomers. Actually, in addition to the naturally solvent‐exposed Cys111 (Supporting Information Fig. S1),47 Cys6 was found to become exposed at higher temperatures [Fig. 3(C)]. While those free Cys residues (Cys6/111) could in principle attack the disulfide bond within a molecule (i.e., the intramolecular shuffling) and also in another molecule (i.e., the intermolecular shuffling) (Fig. 7), the dimeric configuration of apo‐SOD1^S‐S in the I state might favor the inter‐molecular shuffling of the disulfide bond. This hypothesis appears to be consistent with the concentration‐dependent oligomerization of G37R apo‐SOD1^S‐S at 37^°C: the oligomerization was evident at >20 μM but suppressed at 2 μM (Supporting Information Fig. S6). Furthermore, the oligomerization was found to require relatively oxidizing conditions as in mitochondria and was instead suppressed in the reducing environment of the cytoplasm [Fig. 6(D,E)]. We hence speculate that the oligomerization is triggered by the accumulation of apo‐SOD1^S‐S in the dimeric I state to a certain level of its concentration in oxidizing environment such as mitochondria.

Overall, this study revealed the dimeric I state in the thermal denaturation of SOD1, which was further characterized by the distinct conformation from those of the misfolded monomers proposed so far. A pathogenic mutation allowed SOD1 to access the dimeric I state at physiological temperatures and thereby facilitated the formation of oligomers cross‐linked via disulfide bonds (Fig. 7). Given that the disulfide‐crosslinked SOD1 oligomers were observed in the lumber spinal cords of the diseased fALS model mice [Fig. 6(C)],42, 48 genetic and/or pharmaceutical modulation of the dimeric I state could control the disease course.

Materials and Methods

Preparation of purified recombinant SOD1 proteins, electrophoresis, analysis on the solvent accessibility of Cys residues, oligomerization analysis, detection of oligomers in transgenic mice, DSC, and spectroscopic analysis are described in the Supplemental Experimental Procedures.

Small‐angle X‐ray scattering

SAXS experiments were performed using NANO‐Viewer (Rigaku) equipped with a high brightness X‐ray generator, RA‐Micro7 (Rigaku); the scattering images were recorded by a hybrid pixel array detector, PILATUS 200K (DECTRIS). Prior to the SAXS measurement, the protein samples were purified with the gel filtration column (G2000SW, TOSOH) equilibrated with the NNE buffer and collected as mono‐dispersed fraction. The measurements were then performed at a temperature range from 10 to 60°C using apo‐SOD1^S‐S (ca., 5.0 g L⁻¹) in the NNE buffer. The solution temperature was monitored using a thermocouple placed on a Peltier temperature‐controlling cuvette holder, and controlled in an error of ±0.1°C. A successive series of scattering images (3 min × 10 frames) was recorded at a detector distance of 411 mm with the X‐ray wavelength, λ, of 1.5418 Å. Only images free from radiation damages (10 frames in all experiments) were circular‐averaged and normalized by the exposure time (3 × 10 min) and the protein concentration (w/v) to obtain a scattering intensity, I(Q), at the angular momentum, Q, which is defined by 4πsin(θ)/λ; 2θ is the scattering angle. Experimental SAXS data were manipulated using a program, PRIMUS.49 The forward‐scattering intensity, I(0), and the radius of gyration, R _g, were calculated from the observed scattering curves by the Guinier analysis using the following equation.

$$Ln I\left(Q\right)=LnI\left(0\right)-\frac{{R_g}^2}{3}{Q}^2$$ (1)

A relative molecular mass of apo‐SOD1^S‐S was estimated based upon I(0). I(0) is proportional to the molecular mass of a protein, and bovine serum albumin (BSA, Sigma #A7638) was used as a standard for molecular mass determination.50 In this study, I(0) and molecular mass of BSA were 3.16 ± 0.03 and 66,400 Da, respectively.

Supporting information

Supporting Information