Initiation and elongation in fibrillation of ALS-linkedsuperoxide dismutaseMadhuri Chattopadhyaya, Armando Durazoa, Se Hui Sohna, Cynthia D. Strongb, Edith B. Grallaa, Julian P. Whiteleggea,c,and Joan Selverstone Valentinea,1

Departments of aChemistry and Biochemistry and cPsychiatry and Biobehavioral Sciences, University of California, Los Angeles, CA 90095; and bDepartmentof Chemistry, Cornell College, Mount Vernon, IA 52314

This Feature Article is part of a series identified by the Editorial Board as reporting findings of exceptional significance.

Edited by Alan Fersht, University of Cambridge, Cambridge, United Kingdom, and approved October 8, 2008 (received for review July 24, 2008)

Familial amyotrophic lateral sclerosis (fALS) caused by mutations incopper–zinc superoxide dismutase (SOD1) is characterized by thepresence of SOD1-rich inclusions in spinal cords. Similar inclusionsobserved in fALS transgenic mice have a fibrillar appearancesuggestive of amyloid structure. Metal-free apo-SOD1 is a rela-tively stable protein and has been shown to form amyloid fibers invitro only when it has been subjected to severely destabilizingconditions, such as low pH or reduction of its disulfide bonds. Here,by contrast, we show that a small amount of disulfide-reducedapo-SOD1 can rapidly initiate fibrillation of this exceptionallystable and highly structured protein under mild, physiologicallyaccessible conditions, thus providing an unusual demonstration ofa specific, physiologically relevant form of a protein acting as aninitiating agent for the fibrillation of another form of the sameprotein. We also show that, once initiated, elongation can proceedvia recruitment of either apo- or partially metallated disulfide-intact SOD1 and that the presence of copper, but not zinc, ionsinhibits fibrillation. Our findings provide a rare glimpse into thespecific changes in a protein that can lead to nucleation and intothe ability of amyloid nuclei to recruit diverse forms of the sameprotein into fibrils.

amyloid � amyotrophic lateral sclerosis � neurodegeneration �protein aggregation

The antioxidant metalloprotein copper-zinc superoxide dis-mutase (SOD1) is a 153-residue, �-rich, homodimeric pro-

tein that is abundantly present in the cytoplasm. Each subunit ofthe mature form contains a copper ion, a zinc ion, and a disulfidebond (1). In vitro studies have shown that the presence of themetal cofactors, copper and zinc, protect the disulfide bond fromreduction, suggesting a possible explanation for the persistenceof the disulfide bond in the reducing environment of thecytoplasm (2). More than 100 mutations in SOD1 have beenlinked to the familial form of amyotrophic lateral sclerosis(fALS), a fatal neurodegenerative disease caused by selectivedeath of motor neurons. Neuronal death is attributed to a toxicgain of function by mutant SOD1, but the exact mechanism oftoxicity is unknown. Mutations in SOD1 that have been identi-fied in fALS patients occur in 74 positions that are well dispersedover the length of this 153-residue polypeptide (3).

Proteinaceous aggregates have been found in the spinal cordsof ALS patients, and immunomicroscopy has confirmed that theprotein inclusions present in the spinal cords of SOD1-fALSpatients are rich in SOD1 (4, 5). Transgenic mice expressinghuman SOD1 mutants that cause fALS share many symptomswith their human counterparts, including progressive motorneuron degeneration and the presence of detergent-resistant,SOD1-rich aggregates in their spinal cords (6, 7). These aggre-gates have recently been shown to consist primarily of full-length,unmodified SOD1 (6). The visible proteinaceous inclusions havea fibrillar appearance and bind thioflavin S, suggesting anamyloid-like structure (7–9). Taken together, these results sug-

gest that the ability to misfold into an amyloid form underphysiologically accessible conditions represents a fundamentalproperty of this protein and that this property is related to thetoxicity of the ALS-SOD1 mutant proteins.

Bound copper and zinc, the intrasubunit disulfide bond, andthe dimeric nature of the protein all contribute to the unusuallyhigh stability of mature SOD1, and the absence of one or moreof these posttranslational modifications may destabilize the struc-ture of SOD1 sufficiently to allow misfolding into an amyloid form.Previous studies have demonstrated that apo (metal free) SOD1can be induced to form insoluble, fibrillar structures, but onlyunder destabilizing conditions such as low pH (10), elevatedtemperature, the presence of trif luoroethanol (11), or removalof the intrasubunit disulfide bond by mutation (12).

In the current study, we report that small amounts of disulfide-reduced, metal-free, monomeric SOD1 can rapidly initiate fibrilformation in disulfide-intact forms of SOD1 under mild, phys-iologically relevant conditions. Once initiated, SOD1 amyloidfibrils elongate by recruiting apo or partially metallated, dimeric,disulfide-intact SOD1. We also show that the presence of a smallamount of disulfide-reduced mutant SOD1 can initiate fibrilla-tion in WT SOD1, suggesting that WT SOD1 can participate inthe etiology of ALS. Our studies show that even when fibrillationis initiated by the completely immature, disulfide-reduced, apoform of SOD1, more mature and structurally stable forms ofSOD1 can participate in the elongation process. These findingshave important implications in understanding the role of SOD1maturation in fALS etiology and demonstrate the ability of WTSOD1 to contribute to ALS pathogenesis. These results alsoconstitute a clear demonstration of an amyloid-forming processin which initiation of fibrillation can be separated mechanisti-cally from elongation.

ResultsApo SOD1, normally dimeric, can be dissociated into monomersby treatment with a chaotrope such as guanidinium hydrochlo-ride (Gdm�HCl) or by reduction of its intrasubunit disulfide bond(13–15), and monomerization has been implicated in SOD1aggregation pathways (16). To determine the effect of partialmonomerization, apo human WT SOD1 was incubated with

Author contributions: M.C., E.B.G., J.P.W., and J.S.V. designed research; M.C. and A.D.performed research; M.C., A.D., S.H.S., C.D.S., and J.P.W. contributed new reagents/analytictools; M.C., E.B.G., J.P.W., and J.S.V. analyzed data; and M.C., E.B.G., J.P.W., and J.S.V. wrotethe paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Freely available online through the PNAS open access option.

See Commentary on page 18649.

1To whom correspondence should be addressed. E-mail: jsv@chem.ucla.edu.

This article contains supporting information online at www.pnas.org/cgi/content/full/0807058105/DCSupplemental.

© 2008 by The National Academy of Sciences of the USA

www.pnas.org�cgi�doi�10.1073�pnas.0807058105 PNAS � December 2, 2008 � vol. 105 � no. 48 � 18663–18668

BIO

PHYS

ICS

FEA

TURE

ART

ICLE

SEE

COM

MEN

TARY

Dow

nloa

ded

by g

uest

on

Apr

il 10

, 202

1

agitation at 37°C, pH 7.4, in the presence of either 1 M Gdm�HClor 50 mM DTT, both conditions known to induce partial dimerdissociation into monomers. An exponential rise in thioflavin T(TfT) fluorescence was observed after a lag time of �38 (� 4.2)h in the case of Gdm�HCl and 2.1 (� 0.3) h in the case of DTT.(Fig. 1A). Turbidity developed under both conditions, andelectron microscopy confirmed the presence of amyloid fibrils(Fig. 1B). Other reducing agents, specifically tris(2-carboxyeth-yl)phosphine (TCEP) and glutathione, were also effective inpromoting rapid SOD1 fibrillation (Fig. 1C). The presence ofdioxygen in these reactions did not alter the kinetics, becausesamples prepared under anaerobic conditions showed similar lagphases [supporting information (SI) Fig. S1]. The ability ofglutathione to promote fibrillation under anaerobic conditionssuggests that reducing conditions in vivo such as those found inthe cytoplasm are capable of promoting fibrillation.

Gdm�HCl treatment leaves the intrasubunit disulfide bond ofSOD1 intact, but 50 mM DTT treatment, in our experience,generates a mixture of disulfide-intact SOD1 (SOD1S-S) anddisulfide-reduced SOD1 (SOD12SH). To determine the mini-mum concentration of DTT required for fibrillation, apo-SOD1was incubated with 50 �M to 25 mM DTT (Fig. 1D). Surpris-ingly, as low as 250 �M DTT was capable of promoting theformation of amyloid fibrils. At this concentration of reducingagent, we expect only a small fraction of the protein to bereduced, suggesting that a small amount of disulfide-reducedSOD1 is sufficient to initiate fibrillation.

SOD1S-S is a dimer, whereas SOD12SH is a monomer and, thus,size-exclusion chromatography (SEC) can be used to separatethem. SEC on an aliquot collected during lag phase from afibrillation reaction containing 5 mM DTT (Fig. 2A) showed the

presence of dimeric SOD1S-S eluting at 19.8 min (peak a) andmonomeric SOD12SH eluting at 21.2 min (peak b). Treatment ofSOD1 from these 2 peaks with 4-acetamido-4�-maleimidyl-stilbene-2,2�-disulfonic acid (AMS), a thiol modifying agent,followed by SDS/PAGE demonstrated that 2 extra AMS unitsper subunit had been added to the SOD1 in peak b, confirmingthe assignments of peak a as SOD1S-S and peak b as SOD12SH

(Fig. S2).To test the hypothesis that the presence of only a small amount

of SOD12SH is sufficient to initiate fibrillation of SOD1S-S, westudied the effect of replacing the reducing agent with SEC-purified SOD12SH or SOD1S-S (peaks a and b, respectively, in Fig.2A). Because agitation alone has been shown to promotemisfolding and amyloid formation in many proteins, because ofrepeated transient exposures to the air–water interface, we alsoused SOD1S-S that had been isolated from the same reaction asthe SOD12SH, and thus had been similarly agitated, as a control.We found that a very small amount of SOD12SH (5% of theoverall reaction) was sufficient to induce fibrillation, whereasSEC-purified SOD1S-S had no effect (Fig. 2B). Amyloid fibrilformation is proposed to occur by the formation of 1 or moreamyloid nuclei during the lag phase, which then recruit solubleprotein during the elongation phase to generate mature fibrils.Our results suggest that SOD12SH itself can initiate amyloidformation, even in the absence of reducing agents, by formingnuclei, which then recruit SOD1S-S to form mature fibrils. Thismechanism is likely to be relevant to fALS pathogenesis in vivobecause disulfide-reduced SOD1 is present in the spinal cords oftransgenic ALS mice that eventually develop SOD1-rich aggre-gates in their terminal stages (17).

In addition to Cys-57 and Cys-146, which form the disulfidebond, each subunit of SOD1 contains 2 reduced cysteine

Fig. 1. Fibrillation of SOD1 can occur in mild conditions. Apo- SOD1 (50 �M)in 10 mM potassium phosphate buffer, pH 7.4 with additions as indicated wasincubated with constant agitation at 37 °C in 96-well plates. TfT fluorescencewas monitored as an indicator of fibril formation. (A) Fibrillation of SOD1 isstimulated by 1 M Gdm�HCl (red) or 50 mM DTT (blue). No addition is indicatedby black circles. (B) Electron micrographs of SOD1 fibrils generated in thepresence of 1 M Gdm�HCl (Left) and 5 mM DTT (Right). (Scale: 1 cm � 200 nm).(C) Multiple reducing agents catalyze fibril formation. Apo-SOD1 alone (blackcircles), with 10 mM DTT (red squares), 10 mM TCEP (blue triangles), or 10 mMglutathione (green triangles). (D) DTT concentration dependence of fibrilformation for apo-SOD1. DTT concentrations are: 50 �M (black circles), 250�M (red squares), 1 mM (blue triangles), 5 mM (purple inverted triangles), and25 mM (green diamonds).

Fig. 2. Disulfide-reduced SOD1 with free thiol groups initiates fibrillation inthe absence of reducing agent. (A) Separation of SOD1SH-SH (peak b) andSOD1S-S (peak a) generated during the lag phase of a fibrillation reaction in 5mM DTT performed in the manual format by SEC. (B) These peaks were usedas isolated or treated with NEM to block free thiols and added to fibrillationreactions without reducing agent. Apo-SOD1 (47.5 �M) in phosphate bufferwas incubated with 2.5 �M SOD1 in the following forms: SOD12SH (peak b), E;SOD1S-S (peak 2), ‚; SOD1–4NEM (peak b, NEM treated), �; SOD1S-S-2NEM(peak a, NEM treated), �.

18664 � www.pnas.org�cgi�doi�10.1073�pnas.0807058105 Chattopadhyay et al.

Dow

nloa

ded

by g

uest

on

Apr

il 10

, 202

1

residues, Cys-6 and Cys-111. Previous cell culture and in vitroaggregation studies suggested that these 2 nondisulfide cys-teine residues might play an important role in the formation ofdisulfide-cross-linked SOD1 oligomers and aggregates (18–20). To understand and distinguish the roles of Cys-6 andCys-111 from those played by the disulfide-forming residues,Cys-57 and Cys-146, we investigated the effect of mutations inthese positions on the ability to initiate fibrillation in ourpresent system (Fig. 3A). Preparations of disulfide-reducedWT SOD1 and C6A,C111S SOD1 (AS-SOD1) were generatedby overnight treatment of the apoproteins with 10 mM TCEP,followed by SEC-HPLC purification. The same proteins withcysteine thiolates blocked by reaction with N-ethylmaleimide(NEM) were also tested, as were single-cysteine mutants C57SSOD1 and C146S SOD1. In the absence of reducing agents,only disulfide-reduced WT SOD1 and AS-SOD1 were capableof nucleating the conversion of soluble SOD1 into amyloidfibrils. The absolute requirement that both C57 and C146 bepresent and unalkylated suggests that these residues partici-pate in the formation of amyloid nuclei by intermoleculardisulfide cross-linking.

In addition to initiation, intermolecular disulfide cross-linkingmight be involved in fibril elongation. For example, apo-SOD1was found to form soluble, high molecular weight, disulfide-linked structures that bind thioflavin T (TfT) when incubated forextended periods at 37 °C and pH 7.4, with no agitation (18). Toprobe for the presence of intermolecular disulfide cross-linkingunder our conditions, SOD1 fibrils grown in 5 mM DTT weredenatured in Gdm�HCl with or without TCEP, and the products

were analyzed by HPLC-MS (Fig. 3B). Similar amounts ofmonomeric SOD1 were observed in both cases (peak at 30 min),suggesting that most of the SOD in the fibrils was not cross-linked by disulfide bonds. No dimers, trimers or higher-orderoligomers were detected by HPLC. Gel electrophoresis on fibrilsdenatured in SDS with or without the addition of TCEPconfirmed the presence of similar amounts of monomeric SODunder both conditions. Low amounts of dimers and higher-orderoligomers were also detected; their presence may be caused byincomplete denaturation of fibrillar SOD1 in 0.5–1% SDS (Fig.3C). As before, the presence of TCEP had no effect on the yieldof monomeric or oligomeric SOD1. HPLC-MS of AS-SOD1fibrils formed in the presence of 5 mM DTT and denatured inGdm�HCl, in the absence of TCEP, yielded similar amounts ofmonomeric SOD in both cases, showing that the presence of freethiolates at Cys-6 and Cys-111 had no effect on the outcome ofthe reaction (Fig. S3). These results suggest that the nuclei,whose formation likely involved intermolecular disulfide cross-linking, developed into mature fibrils by recruiting disulfide-intact SOD1 through noncovalent interactions.

In vitro studies have shown that the intrasubunit disulfidebond in FALS-SOD1 mutant proteins is more susceptible toreduction than that in WT SOD1 and thus is more likely to bepresent under the reducing conditions prevalent in vivo (2). Wetherefore tested the ability of reduced ALS-SOD1 mutantproteins to induce fibrillation of WT SOD1. Apo-SOD1S-S wasincubated with small amounts of apo-G93A2SH, apo-G37R2SH,and apo-D101N2SH (Fig. 4A) in the absence of reducing agents.In each case, the reduced mutant protein, present at 5% of theoverall protein concentration, initiated fibrillation with a lagtime comparable with that of WT SOD12SH. The presence ofmutant SOD1 in the fibrils was verified by Fourier transformmass spectrometry (FTMS) on the denatured, TCEP-reducedfibrils (Fig. 4B). Small amounts of reduced ALS-SOD1 mutantproteins were also found to initiate fibrillation in 1:1 mixtures ofWT and mutant-SOD1 proteins (Fig. S4).

In the native, holo enzyme, each subunit of the SOD1 dimerbinds 1 copper and 1 zinc ion, and their presence has aprofound impact on the stability of the protein. Fully metal-lated WT SOD1 melts at a very high 92 °C. The absence ofbound copper lowers the melting point to 78 °C, and thesuccessive removal of zinc ions lowers the melting point furtherto 62 °C (1 zinc per dimer) and 52 °C (apo-SOD1) (1). Weexamined the effect of adding small amounts of SOD12SH tothe different metal-bound forms of SOD1. As shown in Fig.4C, SOD1 with 1 zinc ion per dimer fibrillated with kineticsvery similar to those for apo-SOD1, whereas isolated SOD1containing zinc ions in both subunits showed a much slowerrate of elongation. The absence of an effect caused by thebinding of 1 zinc ion per apo-SOD1 dimer is surprising,because 1 zinc per dimer confers considerable structuralstabilization in other assays (21). In contrast, the presence of2 copper ions per dimer, either alone or in conjunction withbound zinc, completely inhibited the formation of amyloidfibrils, showing that copper-bound, disulfide-intact forms ofSOD1 are not easily recruited by SOD1 amyloid nuclei.

DiscussionThe ability of a specific form of a protein to act as an initiatingagent for the fibrillation of another form has been demonstratedinfrequently, presumably because of experimental difficultiesassociated with pinpointing initiation events. The few knowninstances include defining the minimum peptide sequence re-quired to initiate fibrillation in Ure2p (22) and A� (23) anddetermining the oligomeric status of nuclei and the conforma-tion changes required for their formation in polyglutaminepeptides (24) and islet amyloid polypeptide (25). Here, we havebeen able to demonstrate the requirements for fibril initiation

Fig. 3. Both Cys-57 and Cys-146, but not Cys-6 or Cys-111, are necessary toinitiate amyloid formation. (A) 2.5 �M SOD12SH (E), AS2SH (�), SOD1–4NEM(‚), AS-2NEM (�), C57S ({), and C146S (�) were added to 47.5 �M apo-SOD1in the absence of reducing agents. (B) The amyloid nuclei elongate via non-covalent interactions to form mature fibrils. SOD1 fibrils grown in the pres-ence of 5 mM DTT were denatured in 6.5 M Gdm�HCl with or without 20 mMTCEP and analyzed by HPLC-MS. (Left) Total ion count of Gdm�HCl-denatured(solid line) or Gdm�HCl-and-TCEP-denatured (dashed line) SOD1 fibrils. (Right)Reconstructed mass spectrum of peak at 30 min for LCMS of Gdm�HCl dena-turation (solid line) and Gdm�HCl-TCEP denaturation (dashed line) reactions,showing monomeric SOD at 15,845 Da. (C) PAGE of amyloid fibrils grown in 5mM DTT and denatured in SDS with or without 20 mM TCEP show similaramounts of monomers, dimers, and higher oligomers, suggesting fibril elon-gation does not involve intermolecular disulfide cross-linking.

Chattopadhyay et al. PNAS � December 2, 2008 � vol. 105 � no. 48 � 18665

BIO

PHYS

ICS

FEA

TURE

ART

ICLE

SEE

COM

MEN

TARY

Dow

nloa

ded

by g

uest

on

Apr

il 10

, 202

1

and elongation in SOD1 individually. Our studies show that aspecific intermediate in SOD1 maturation pathway, disulfide-reduced, monomeric apo-SOD1, can initiate the rapid conver-sion of more mature forms of soluble SOD1 into amyloid fibrils.These results may be especially relevant to SOD1 fibrillationmechanisms in vivo because SOD1 monomers, including disul-fide-reduced SOD1, have been detected in spinal cord extractsof transgenic mice expressing fALS SOD1 mutants well beforethey reach terminal stages of the disease (26, 27).

We have also shown that nucleation by disulfide-reduced(monomeric) SOD1 requires the presence of cysteine thiolates atboth positions 57 and 146, suggesting that the nucleus is formedby intermolecular disulfide cross-linking using these cysteines.Similar species can form in vivo despite the reducing environ-ment of the cytoplasm as shown by the presence of disulfide-cross-linked SOD1 oligomers in spinal cord extracts from ALSSOD1 transgenic mice. Cys-146 was one of the cysteine residuesinvolved in their formation (28). Once SOD1 nuclei form,elongation proceeds via recruitment of apo- or partially-metallated disulfide-intact SOD1 through noncovalent interac-tions. Although the binding of zinc offers SOD1 limited protec-tion from being assembled into fibrils, copper alone or inconjunction with zinc prevents SOD1 from being incorporatedinto amyloid fibrils. Because SOD1 aggregates isolated from the

spinal cords of transgenic ALS mice are copper-deficient (17),copper binding may be an essential mechanism for protectionfrom fibrillation in vivo.

It is important to note, however, that our studies do not showthat disulfide reduction is absolutely required for initiation. Inthe presence of Gdm�HCl, disulfide-intact, apo-SOD1 can beinduced to form amyloid fibrils, albeit with a significantly longerlag phase, possibly via a mechanism involving disulfide-intactmonomers.

We found that small amounts of disulfide-reduced mutantscan initiate fibrillation of WT SOD1. This result supports theidea that WT SOD1 participates in disease progression and mayhelp explain the mechanism. Transgenic mice coexpressing WTand mutant SOD1 often show an earlier onset and acceleratedprogression of ALS-like neurodegenerative symptoms (9, 29,30). Because the SOS bond in ALS mutant proteins is moreeasily reduced than that in WT SOD1 (2), the reduced forms aremore likely to be present in higher quantities in vivo, allowing forearlier nucleus formation, which could be followed by efficientincorporation of the more abundant WT SOD1. This processmay be particularly important for unstable mutants such asL126Z or A4V that have low steady-state concentrations in vivo(31), and might require WT protein to sustain fibril growth, butit could also occur with more stable mutants in which WT andmutant proteins are present in approximately equal amounts.

Fig. 4. Disulfide-reduced mutants can initiate fibrillation of SOD1, while copper, but not zinc, can inhibit fibril elongation. (A) Small amounts ofdisulfide-reduced fALS-SOD1 mutants can initiate fibril formation of apo-SOD1; 2.5 �M SOD12SH (red squares), G93A2SH (orange circles), G37R2SH (aqua triangles),and D101N2SH (black triangles) were incubated with 47.5 �M apo-SOD1 in the absence of reducing agents. Fibril formation was confirmed in all cases by electronmicroscopy (data not shown). (B) High-resolution mass spectrometry in an LTQ-FTMS instrument shows that WT SOD1 fibrils grown by the addition of G93A2SH

and denatured in the presence of 6.5 M Gdm�HCl and 20 mM TCEP contain both WT SOD1 and G93A. Shown is the isotopically-resolved reconstructed intact-ionmass spectrum from the z � 10 peak. (C) Cu and Zn binding to apo-SOD1 can attenuate and, in some cases, abolish fibril formation initiated by SOD12SH; 2.5 �MSOD12SH was added to 47.5 �M apo-SOD1 (red squares), E2,EZn-SOD1 (black circles), as-isolated SOD1 (light blue triangles), Cu2E2-SOD1 (dark blue triangles), andCu2Zn2-SOD1 (gray diamonds). As-isolated SOD1 refers to SOD1 purified from yeast. It contains 0.6 eq. copper and 2.5 eq. zinc ions per dimer.

18666 � www.pnas.org�cgi�doi�10.1073�pnas.0807058105 Chattopadhyay et al.

Dow

nloa

ded

by g

uest

on

Apr

il 10

, 202

1

Supporting this idea, we found that in 1:1 mixtures of WT andmutant proteins the fibrils incorporated both proteins.

In conclusion, our studies have allowed us to separate theprocesses of initiation and elongation in SOD1 fibril formationand have demonstrated that reduction of the intrasubunit disul-fide bond dramatically accelerates the initiation process. Ourobservation of the initiation of the fibrillation of WT SOD1 bythe addition of small amounts of disulfide-reduced ALS SOD1mutant proteins provides a feasible model for initial stages of thedisease process in SOD1-linked fALS.

Materials and MethodsSOD1 Expression and Purification. WT and fALS SOD1 mutants were ex-pressed in Saccharomyces cerevisiae and purified according to publishedprocedures (14). C6A and C111S SOD1 (AS-SOD1) was expressed in Esche-richia coli and purified by using a similar procedure. Purified SOD1 wasdemetallated by dialysis in a Slide-a-lyzer (Pierce; 10, 000 Da) against 50 mMEDTA, 50 mM NaCl, pH 3.8 as described (14) except that, in the last step, 20mM potassium phosphate, pH 7.0 was used as the dialysis buffer. Apo-SOD1was flash-frozen in liquid nitrogen and stored at �20 °C before use. Metalcontent of apo-SOD1 and metallated SOD1 was determined by inductivelycoupled plasma (ICP)-MS. Typically, apo-SOD1 contained �0.07 equivalentsof Cu and Zn per dimer.

Metal Titrations. Metal titrations were carried out according to the proceduresoutlined by Goto et al. (32). Briefly, apo-SOD1 was dialyzed against 100 mMsodium acetate, pH 5.5 and concentrated to �300 �M with a YM-10 microcon-centrator (Millipore/Fisher). One equivalent of copper sulfate or zinc sulfatewas added from an aqueous 10 mM stock solution in 2–4 equal aliquots at 1-hintervals with slow stirring at 4 °C. When necessary, the second equivalent ofcopper or zinc was added in a similar fashion and stirred overnight at 4 °C. Toremove unbound metal, the reaction was transferred to a YM-10 microcon-centrator, concentrated to 200 �L, and overlaid with an equal volume of 20mM potassium phosphate, pH 7.4. The solution was concentrated, and thewash step was repeated 2–3 additional times. Cu2Zn2SOD1 was prepared byadding the necessary equivalents of zinc, followed by copper, to 300 �Mas-isolated SOD1 in 100 mM sodium acetate, pH 5.5 whose copper and zinccontent had been determined by ICP-MS. Zinc binding was allowed to proceedovernight at 4 °C, and then copper was added followed by overnight incuba-tion. The sample was washed in the manner outlined above. Metal incorpo-ration was verified by ICP-MS and UV-visible spectroscopy (32).

Amyloid Formation in Vitro. All solutions were filtered through a 0.22-�msyringe filter before use. Amyloid conversion was carried out by using 2formats: automated and manual. Reactions contained 50 �M apo-SOD1 in 10mM potassium phosphate, pH 7.4 with or without with various concentrationsof reducing agents (DTT, TCEP, or glutathione) or 1 M Gdm�HCl, as indicated.In the automated format, 200 �L of solution was pipetted into a chamber ina 96-well plate followed by a Teflon ball (1/8-in diameter) and 2 �L of TfT froma 4 mM aqueous stock. The plate was agitated at 300 rpm (3-mm rotationdiameter) in a Fluoroskan plate-reader (Thermo) at 37 °C. Fluorescence mea-surements were recorded every 15 min by using �ex � 444 nm, �em � 485 nm,with an integration time of 200 �s. All reactions were performed in replicatesof 3 or more. Reactions carried out anaerobically were assembled in a glovebox by using reagent solutions that were purged with argon and left openinside the anaerobic chamber for 1 h before use.

In the manual format, reactions of the same composition as above wereperformed at a final volume of 1 mL in 2-mL glass vials containing a Teflon ball(1/8-in diameter) and agitated on an orbital shaker at 250 rpm (19-mmrotation diameter). The vials were sealed with plastic caps containing a PTFEseptum and purged with nitrogen. At intervals, aliquots were withdrawn byusing a Hamilton gas-tight syringe through the septum. TfT fluorescence wasmeasured by diluting 10 �L of the reaction mix into 150 �L of 40 �M TfT stocksolution made in 5 mM potassium phosphate, pH 7, and fluorescence emissionat 480 nm was recorded on a spectrophotometer using �ex � 444 nm (excita-tion and emission slit widths � 2 mm). Lag times were calculated by fitting TfTfluorescence data to the formula:

F � F0 � �A � ct� /�1 � expk� tm � t� ,

where lag phase was calculated as tm � 2/k (33).

SEC-HPLC. Disulfide-oxidized and -reduced apo-SOD1, which are dimeric andmonomeric, respectively, were separated on a 7.5-mm � 30-cm TSK G2000 SWcolumn (Toyosoda) equipped with a guard column on an Agilent 1200 HPLCusing a mobile phase containing 50 mM NaCl, 50 mM potassium phosphate(pH 6.7), 5 mM DTT. Apoprotein (0.5–1 mg) reacted overnight with 10 mMTCEP was loaded at a concentration of 3–5 mg/ml. SOD1S-S and SOD12SH

collected from HPLC were concentrated by using YM-10 microconcentrators toa final concentration of 3–5 mM, so that the concentration of accompanyingDTT never exceeded 50 �M when HPLC-purified SOD1 was added to apo-SOD1during initiation reactions. This procedure was done to ensure that initiation,where it occurred, was caused by the presence of disulfide-reduced apo-SOD1and not DTT. As shown in Fig. 1D, apo-SOD1 fibrillation did not occur in DTTconcentrations of 50 �M or less.

NEM and AMS Labeling. Disulfide-intact or disulfide-reduced apo-SOD1 at 3–5mg/ml was treated with 5 mM NEM at 37 °C for 1 h. Excess NEM was removedby 3 rounds of concentration and redilution in a YM-10 microconcentrator, in20 mM potassium phosphate, pH 7.4. The presence of NEM bound to cysteineswas checked by ESI-MS. AMS labeling was carried out before gel electrophore-sis similar to published procedures (34).

Amyloid Denaturation Monitored by Gel Electrophoresis. Fibrils from 1 well ofthe 96-well plate were transferred to a 1.5-mL microcentrifuge tube andcentrifuged at 16,000 � g. The supernatant was decanted, and the pelletedfibrils were resuspended in deionized water (Nanopure) and centrifugedagain to remove all soluble SOD1. The resulting pellet was resuspended in100 �L of 20 mM potassium phosphate, pH 7.4. Ten microliters of the fibrilsuspension was incubated with 0.5 or 1% SDS with or without 20 mM TCEPin a total volume of 20 �L for 1 h at 50 °C. Twenty microliters of nonde-naturing loading dye (without SDS or �-mercaptopethanol) was added tothe reactions, and 20 �L of the resulting solution was loaded on a 5%stacking, 15% resolving SDS polyacrylamide gel and electrophoresed at 150V until the samples stacked and then at 200 V. The gels were Coomassie-stained to visualize SOD1.

Liquid Chromatography/MS. SOD1 fibrils from a well from reactions in the96-well plate were collected by centrifugation at 16,000 � g for 15 min,resuspended in 200 �L of distilled water, and centrifuged again to remove allreducing agent, chaotrope, and soluble SOD1. The fibrils were denatured byincubation in 6.5 M Gdm�HCl with or without 20 mM TCEP at 50°C for 1 h. Thereaction was then diluted 1:1 with 5% acetonitrile containing 1% formic acidand loaded on a reverse-phase PL-RPS column connected to an API Sciex massspectrometer. SOD1 was eluted off the column by using a water/acetonitrilegradient.

FTMS. High-resolution mass spectrometry was performed on an LTQ-FT instru-ment from Thermo Fisher, operated in nanospray mode. The sample wasdenatured according to the above procedure and desalted by using a C18

Zip-Tip (Millipore) according to the manufacturer’s protocol, with final elu-tion in 50% acetonitrile containing 0.1% formic acid. The sample (5 �L) wasloaded into a pulled-glass nanospray emitter (Proxeon), and a stable spray wasestablished at 1,800 V relative to the ion transfer tube. The instrument wasoperated in FTMS mode at 100,000 resolution at m/z � 400. After recordingthe full mass spectrum, an ion isolation experiment was performed on the �10species centered at 1,587.0 � 3 Da. The zero-charge spectrum was obtained byusing Xtract software (Thermo Fisher). A collision-activated dissociation (CAD)experiment was performed in the ion trap (energy � 12), and the fragmentswere analyzed by FTMS. The complex MS/MS spectrum was matched to SOD1sequences by using Prosight PC software (Thermo Fisher), confirming thepresence of both WT SOD1 and G93A in the CAD spectrum.

Electron Microscopy. Fibril samples from a single well in the 96-well plate werespun down at 16,000 � g, washed with water, and resuspended in 100 �L ofdeionized water. Five microliters of the fibril suspension was deposited on aformavar-coated copper grid (Ted Pella, Inc.). The sample was allowed toadsorb for 5 min, blotted, washed for 30 s with distilled water, blotted andthen stained with freshly filtered 2% uranyl acetate (5 �L) for 3 min andblotted again. The grids were air-dried for 30 min before insertion into a JEOL1200 electron microscope operated at 80 keV. Fibrils were typically visualizedat a magnification of 75,000�.

ACKNOWLEDGMENTS. We thank Sean Lehman and Emmanuel Koli for pro-tein purification of AS-WT SOD1 and Dr. Amir Liba for metal analysis of SOD1by ICP-MS. This work was supported by National Institute of NeurologicalDisorders and Stroke Grant NS-049134.

Chattopadhyay et al. PNAS � December 2, 2008 � vol. 105 � no. 48 � 18667

BIO

PHYS

ICS

FEA

TURE

ART

ICLE

SEE

COM

MEN

TARY

Dow

nloa

ded

by g

uest

on

Apr

il 10

, 202

1

1. Valentine JS, Doucette PA, Potter SZ (2005) Copper-zinc superoxide dismutase andamyotrophic lateral sclerosis. Annu Rev Biochem 74:563–593.

2. Tiwari A, Hayward LJ (2003) Familial amyotrophic lateral sclerosis mutants of copper/zinc superoxide dismutase are susceptible to disulfide reduction. J Biol Chem278:5984–5992.

3. Bruijn LI, Miller TM, Cleveland DW (2004) Unraveling the mechanisms involved inmotor neuron degeneration in ALS. Annu Rev Neurosci 27:723–749.

4. Kato S, et al. (1997) Pathological characterization of astrocytic hyaline inclusions infamilial amyotrophic lateral sclerosis. Am J Pathol 151:611–620.

5. Shibata N, et al. (1994) Cu/Zn superoxide dismutase-like immunoreactivity in Lewy body-like inclusions of sporadic amyotrophic lateral sclerosis. Neurosci Lett 179:149–152.

6. Shaw BF, et al. (2008) Detergent-insoluble aggregates associated with amyotrophiclateral sclerosis in transgenic mice contain primarily full-length, unmodified superox-ide dismutase-1. J Biol Chem 283:8340–8350.

7. Wang J (2003) Copper-binding-site-null SOD 1 causes ALS in transgenic mice: Aggre-gates of non-native SOD 1 delineate a common feature. Hum Mol Genet 12:2753–2764.

8. Basso M, et al. (2006) Insoluble mutant SOD1 is partly oligoubiquitinated in amyotro-phic lateral sclerosis mice. J Biol Chem 281:33325–33335.

9. Jaarsma D, Teuling E, Haasdijk ED, De Zeeuw CI, Hoogenraad CC (2008) Neuron-specificexpression of mutant superoxide dismutase is sufficient to induce amyotrophic lateralsclerosis in transgenic mice. J Neurosci 28:2075–2088.

10. DiDonato M, et al. (2003) ALS mutants of human superoxide dismutase form fibrousaggregates via framework destabilization. J Mol Biol 332:601–615.

11. Stathopulos PB, et al. (2003) Cu/Zn superoxide dismutase mutants associated withamyotrophic lateral sclerosis show enhanced formation of aggregates in vitro. ProcNatl Acad Sci USA 100:7021–7026.

12. Furukawa Y, Kaneko K, Yamanaka K, O’Halloran TV, Nukina N (2008) Complete loss ofpost-translational modifications triggers fibrillar aggregation of SOD1 in familial formof ALS. J Biol Chem 283:24167–24176.

13. Arnesano F, et al. (2004) The unusually stable quaternary structure of human Cu,Zn-superoxide dismutase 1 is controlled by both metal occupancy and disulfide status.J Biol Chem 279:47998–48003.

14. Doucette PA, et al. (2004) Dissociation of human copper-zinc superoxide dismutasedimers using chaotrope and reductant: Insights into the molecular basis for dimerstability. J Biol Chem 279:54558–54566.

15. Lindberg MJ, Normark J, Holmgren A, Oliveberg M (2004) Folding of human superox-ide dismutase: Disulfide reduction prevents dimerization and produces marginallystable monomers. Proc Natl Acad Sci USA 101:15893–15898.

16. Rakhit R, et al. (2004) Monomeric Cu/Zn superoxide dismutase is a common misfoldingintermediate in the oxidation models of sporadic and familial ALS. J Biol Chem279:15499–15504.

17. Jonsson PA, et al. (2006) Disulphide-reduced superoxide dismutase-1 in CNS of trans-genic amyotrophic lateral sclerosis models. Brain 129:451–464.

18. Banci L, et al. (2007) Metal-free superoxide dismutase forms soluble oligomers underphysiological conditions: A possible general mechanism for familial ALS. Proc NatlAcad Sci USA 104:11263–11267.

19. Niwa J, et al. (2007) Disulfide bond mediates aggregation, toxicity, and ubiquity-lation of familial amyotrophic lateral sclerosis-linked mutant SOD1. J Biol Chem282:28087–28095.

20. Karch CM, Borchelt DR (2008) A limited role for disulfide cross-linking in the aggre-gation of mutant SOD1 linked to familial amyotrophic lateral sclerosis. J Biol Chem283:13528–13537.

21. Potter SZ, et al. (2007) Binding of a single zinc ion to one subunit of copper-zincsuperoxide dismutase apoprotein substantially influences the structure and stability ofthe entire homodimeric protein. J Am Chem Soc 129:4575–4583.

22. Taylor KL, Cheng N, Williams RW, Steven AC, Wickner RB (1999) Prion domain initiationof amyloid formation in vitro from native Ure2p. Science 283:1339–1343.

23. Lazo ND, Grant MA, Condron MC, Rigby AC, Teplow DB (2005) On the nucleation ofamyloid �-protein monomer folding. Protein Sci 14:1581–1596.

24. Wetzel R (2008) Kinetics and thermodynamics of amyloid fibrils assembly. Acc ChemRes 39:671–679.

25. Eakin CM, Berman AJ, Miranker AD (2006) A native to amyloidogenic transitionregulated by a backbone trigger. Nat Struct Mol Biol 13:202–208.

26. Rakhit R, et al. (2007) An immunological epitope selective for pathological monomer-misfolded SOD1 in ALS. Nat Med 13:754–759.

27. Zetterstrom P, et al. (2007) Soluble misfolded subfractions of mutant superoxidedismutase-1s are enriched in spinal cords throughout life in murine ALS models. ProcNatl Acad Sci USA 104:14157–14162.

28. Furukawa Y, Fu R, Deng HX, Siddique T, O’Halloran TV (2006) Disulfide cross-linkedprotein represents a significant fraction of ALS-associated Cu, Zn-superoxide dis-mutase aggregates in spinal cords of model mice. Proc Natl Acad Sci USA 103:7148–7153.

29. Deng HX, et al. (2006) Conversion to the amyotrophic lateral sclerosis phenotype isassociated with intermolecular linked insoluble aggregates of SOD1 in mitochondria.Proc Natl Acad Sci USA 103:7142–7147.

30. Jaarsma D, et al. (2000) Human Cu/Zn superoxide dismutase (SOD1) overexpression inmice causes mitochondrial vacuolization, axonal degeneration, and premature mo-toneuron death and accelerates motoneuron disease in mice expressing a familialamyotrophic lateral sclerosis mutant SOD1. Neurobiol Dis 7:623–643.

31. Hoffman EK, Wilcox HM, Scott RW, Siman R (1996) Proteasome inhibition enhances thestability of mouse Cu/Zn superoxide dismutase with mutations linked to familialamyotrophic lateral sclerosis. J Neurol Sci 139:15–20.

32. Goto JJ, et al. (2000) Loss of in vitro metal ion binding specificity in mutant copper-zincsuperoxide dismutases associated with familial amyotrophic lateral sclerosis. J BiolChem 275:1007–1014.

33. Cohlberg JA, Li J, Uversky VN, Fink AL (2002) Heparin and other glycosaminoglycansstimulate the formation of amyloid fibrils from-synuclein in vitro. Biochemistry41:1502–1511.

34. Furukawa Y, Torres AS, O’Halloran TV (2004) Oxygen-induced maturation of SOD1: Akey role for disulfide formation by the copper chaperone CCS. EMBO J 23:2872–2881.

18668 � www.pnas.org�cgi�doi�10.1073�pnas.0807058105 Chattopadhyay et al.

Dow

nloa

ded

by g

uest

on

Apr

il 10

, 202

1