Hydrogen–deuterium exchange in vivo to measure turnover of an ALS-associated mutant SOD1 protein in spinal cord of mice


  • George W. Farr,

    1. Howard Hughes Medical Institute, Yale University School of Medicine, Boyer Center, New Haven, Connecticut 06510
    2. Department of Genetics, Yale University School of Medicine, Boyer Center, New Haven, Connecticut 06510
    Search for more papers by this author
  • Zheng Ying,

    1. Howard Hughes Medical Institute, Yale University School of Medicine, Boyer Center, New Haven, Connecticut 06510
    2. Department of Genetics, Yale University School of Medicine, Boyer Center, New Haven, Connecticut 06510
    Search for more papers by this author
  • Wayne A. Fenton,

    1. Department of Genetics, Yale University School of Medicine, Boyer Center, New Haven, Connecticut 06510
    Search for more papers by this author
  • Arthur L. Horwich

    Corresponding author
    1. Howard Hughes Medical Institute, Yale University School of Medicine, Boyer Center, New Haven, Connecticut 06510
    2. Department of Genetics, Yale University School of Medicine, Boyer Center, New Haven, Connecticut 06510
    • Department of Genetics/HHMI, Yale School of Medicine, Boyer Center, 295 Congress Ave., New Haven, CT 06510
    Search for more papers by this author


Mutations of cytosolic Cu/Zn superoxide dismutase 1 (SOD1) in humans and overexpression of mutant human SOD1 genes in transgenic mice are associated with the motor neuron degenerative condition known as amyotrophic lateral sclerosis (ALS; Lou Gehrig's disease). Gain-of-function toxicity from the mutant protein expressed in motor neurons, associated with its misfolding and aggregation, leads to dysfunction and cell death, associated with paralyzing disease. Here, using hydrogen–deuterium exchange in intact mice in vivo, we have addressed whether an ALS-associated mutant protein, G85R SOD1–YFP, is subject to the same rate of turnover in spinal cord both early in the course of the disease and later. We find that the mutant protein turns over about 10-fold faster than a similarly expressed wild-type fusion and that there is no significant change in the rate of turnover as animals age and disease progresses.


Amyotrophic lateral sclerosis (ALS; Lou Gehrig's disease) is a progressive neurodegenerative disease affecting motor neurons.1 About 2% of cases are associated with mutations in the abundant cytosolic free radical scavenging enzyme, Cu/Zn superoxide dismutase 1 (SOD1).2, 3 Studies suggest that these mutations lead to a dominant gain of toxic function associated with misfolding and aggregation,3–6 resembling the behavior of several other proteins associated with neurodegenerative disorders. The ability to model the effects of SOD1 mutations in transgenic mice has enabled the investigation of misfolding and aggregation in vivo during the progression of the disease.7–9

We have focused on the ALS-associated SOD1 mutation, G85R, which results in misfolding of SOD1 to such an extent that this mutant subunit does not reach the native state. In particular, it does not acquire the normal intrasubunit disulfide bond (C57–C146), does not become detectably metallated with Cu(II) or Zn(II), and fails to form the normal active homodimer.10 The abnormal protein has been observed to rapidly turn over in transfected cultured cells.11 In transgenic mice, its steady-state level is less than that of equivalently expressed wild-type human SOD1, consistent with more rapid turnover.10 Nevertheless, transgenic animals bearing G85R human SOD1 or a fusion of G85R with yellow fluorescent protein (YFP) develop ALS.9, 10, 12 In the latter animals, we have observed that disease is associated with progressive development of soluble oligomers, detected by gel filtration of spinal cord extracts, and of insoluble aggregates in motor neurons, observed by fluorescence microscopy. These data raised questions, first, as to whether turnover of the fusion protein in vivo was as rapid as observed in the cultured cell system, and, second, as to whether there might be a progressive decrease in the rate of turnover of the G85R SOD1–YFP fusion protein during the course of disease that could account for the apparent progressive increase in oligomers and aggregates of the mutant protein. Here, we have addressed this possibility by carrying out a hydrogen–deuterium exchange experiment in vivo in both mutant G85R SOD1–YFP and wild-type SOD1–YFP transgenic mice, comparing the rate of turnover of the respective SOD1–YFP proteins in spinal cord at early and later times of life.

Results and Discussion

We first tested whether G85R SOD1–YFP is more rapidly turned over than the wild-type SOD1–YFP fusion in stable doxycycline-regulated NSC-34 (neuroblastoma-derived) transformants that we produced. The mutant protein was indeed turned over at least several-fold more rapidly not only largely via the proteasome but also to some extent by the autophagy pathway (see Supporting Information Figs. 1 and 2).

Figure 1.

Protocol for measuring turnover of SOD1–YFP proteins in spinal cord of transgenic mice. The scheme presents the major steps in the labeling and analysis protocol. After 10 weeks of treatment with 8% D2O in their water supply, mice were returned to 100% H2O. After 6 days to allow time for body water and amino acid pools to return to ∼ 100% hydrogen,14 one mouse was sacrificed at each time point indicated in Fig. 3, its spinal cord quickly removed, a soluble extract prepared, and the labeled SOD–YFP recovered by immunoaffinity capture with an anti-YFP antibody resin. After extensive washing with 8 M urea at neutral pH to remove any associated proteins, the SOD–YFP was eluted with 10 M urea at pH 2, neutralized, and digested with trypsin as previously described.14 The resulting peptides were loaded onto a two-part C18 column, and the column was interfaced directly with a LTQ-XL Orbitrap mass spectrometer and eluted with an acetonitrile gradient as described in Materials and Methods. The chromatogram and resulting peptide spectra were analyzed with Xcalibur and ProMass software to identify individual peptides and calculate their masses (see Fig. 2).

Figure 2.

Spectra of the largest tryptic peptide from SOD1–YFP. A: A portion of the spectrum of the largest tryptic peptide of SOD1–YFP recovered from a wild-type transgenic mouse before treatment with D2O is shown. About 15 scans across a chromatographic peak were summed to produce this spectrum. This is the isotope envelope of the 4+ charge state; corresponding envelopes for the 3+ and 5+ charge states were also prominent in the spectrum. ProMass software reduced this spectrum to the uncharged peptide spectrum shown in the inset, with an average mass of 4535.4 Da. This mass is in good agreement with the mass (4535.9 Da) calculated from the sequence of this peptide. MS/MS sequencing of this peptide confirmed that it had the expected amino acid sequence (not shown). B: The same region of the spectrum of a SOD1–YFP peptide recovered from a wild-type transgenic mouse after treatment with 8% D2O for 10 weeks (0 days of washout). As in A), the 4+ charge state is shown, and 3+ and 5+ envelopes are also present in the spectrum. Note that the isotope envelope is much broader than that in A) and that it is shifted towards higher m/Z values. Similarly, the deconvoluted spectrum in the inset is also broader and has a centroided average mass of 4540.5 Da, about 5 Da greater than that in A). The broader isotope envelope and mass shifts result from deuterium incorporation into this peptide, with an average content of about five deuteriums per peptide. Because only the 93 nonreadily exchangeable hydrogens on the 24 nonessential amino acids in this peptide would be expected to be labeled, this value represents a labeling efficiency of about 67% (5/0.08 × 93), assuming tissue deuteration equal to that in the water supply.

To monitor turnover of SOD1–YFP fusion proteins in mouse spinal cord, where we had observed earlier that the most significant RNA and protein expression is in motor neurons,9 we first deuterated the mice by supplying 8% D2O in the drinking water for 10 weeks, a time period sufficient to achieve significant, although not necessarily equilibrium, deuteration of many proteins, including SOD1–YFP (Refs.13,14; Fig. 1). The deuterium was then chased by returning the water supply to 100% H2O. An initial washout period of 6 days allowed replacement of D2O with H2O in spinal cord extracellular and intracellular fluid and nonessential amino acid pools.13, 14 Subsequently, we monitored the level of deuteration in SOD1–YFP fusion proteins immunoaffinity-captured from the soluble fraction of spinal cord lysates prepared at various times of chase up to ∼ 3 weeks (Fig. 1). In such an approach, if the soluble fusion protein is very stable, its level of deuteration should change slowly during the chase. In contrast, if the protein is relatively unstable, its deuteration would be predicted to change more rapidly as deuterated molecules are degraded and replaced by newly translated, fully protonated protein during the chase.

For technical reasons, it did not prove possible to directly determine the mass of the intact affinity-captured SOD1–YFP fusion proteins. Therefore, tryptic proteolysis was carried out on the captured proteins followed by HPLC/MS (Fig. 1). The largest tryptic peptide, extending from aa 330–370, was well-separated in HPLC and easily detected by MS and was informative for level of deuteration. For example, Figure 2 compares the isotope envelopes of the 4+ charge state of the fully protonated peptide (top panel) and the deuterated peptide obtained from a wild-type animal after 10 weeks of exposure to 8% D2O (lower panel). These profiles, along with those of the 3+ and 5+ charge states, could be reduced to uncharged peptide masses (see inset panels), giving a centroided mass difference of ∼ 5 Da (see Fig. 2 legend for more details).

Wild-type SOD1–YFP and G85R SOD1–YFP animals were sacrificed at various times after commencement of washout, beginning at 4 months of age (D2O having been commenced at 1.5 months), and the masses of this peptide were measured. The data are plotted in Figure 3. Strikingly, even after 18 days, wtSOD1–YFP had lost only ∼ 1 Da, reflecting the extreme stability of the wild-type protein, with a t1/2 ∼ 22 days. By contrast, the G85R SOD1–YFP protein underwent a far more rapid exchange, with deuteration decaying with single exponential kinetics and a half-life of ∼ 2.6 days. By 18 days of washout, it had exchanged to the level of the fully protonated state, indicating that the deuterated protein had turned over and been replaced by newly made, protonated molecules. We then asked whether the mutant protein would be similarly rapidly turned over at a later stage in the course of the disease, at a time when aggregation becomes significant,9 commencing a chase at 7 months of age (D2O started at 4.5 months). Again, the rate of turnover of soluble G85R SOD1–YFP was rapid. [Note that this experiment does not assess turnover of any insoluble, aggregated material because it is not recovered by the immunoaffinity capture step.] We thus conclude that the rate of turnover of the mutant protein is not affected with progression of disease.

Figure 3.

Turnover of SOD1–YFP proteins evaluated by exchange of deuterons to protons in reporting peptide. Individual centroided masses of the largest SOD1–YFP tryptic peptide were estimated as illustrated in Figure 2 for peptides recovered from individual mouse spinal cords after D2O washout for the indicated times. Results from a wild-type cohort that were 4 months of age at the start of the washout and from two G85R mutant cohorts that were 4 and 7 months at the start of the washout are shown. The lines are single exponential decay curves fit to the data points in Kaleidagraph, with the estimated half-times for turnover indicated. Because of the limited extent of washout in the wild-type cohort, the value shown is probably a lower limit to the turnover half-time of the wild-type protein. The turnover of the mutant protein is clearly much faster than that of the wild-type, and, more importantly, the rate of turnover does not decline significantly as the animals age and their disease progresses.

In summary, the behavior in vivo of the G85R SOD1–YFP protein responsible for ALS in these transgenic animals resembles that observed in culture, namely, there is rapid turnover of the mutant protein. Importantly, such rapid turnover persists until relatively late in the course of the disease. [Similar experiments in a limited number of animals at 9 months of age, the average life-span of this strain, showed the same rapid turnover (not shown).] Despite such rapid turnover, however, a fraction of the protein accretes into other forms, for example, soluble oligomers and insoluble aggregates, and either of these species or others not so far detected (e.g., misfolded monomers) might be responsible for progressive neuronal damage. Thus, at least before the very terminal stage of disease (the last 3–4 days, where we observe wholesale aggregation throughout the spinal cord including sciatic axons), it appears that a reduction in the rate of proteasomal or autophagosomal turnover is not responsible for onset and progression of disease. Rather, it seems that the high rate of turnover is simply not sufficient to prevent the accumulation of toxic species. Whether this is a function of certain conformations being inaccessible to turnover or simply due to a stochastic effect remains to be seen.


ALS, amyotrophic lateral sclerosis; LC–MS, liquid chromatography–mass spectrometry; SOD1, superoxide dismutase1; YFP, yellow fluorescent protein.

Materials and Methods


The transgenic mouse strains, wild-type SOD1–YFP strain 592 and mutant G85R SOD1–YFP strain 641, have been described.9 All animal experiments were conducted according to protocols approved by the Yale University Institutional Animal Care and Use Committee.

Hydrogen–deuterium exchange

Mice were injected intraperitoneally with a loading dose of 200 μL of 98% D2O, and their water bottles were then filled with 8%D2O/92% H2O. After 10 weeks, a “chase” was performed by switching to 100% H2O in the water bottle. At various times thereafter, animals were euthanized, and spinal cords were harvested. The cords were immediately homogenized in 50 mM Tris (pH 7.4) with protease inhibitor cocktail (Roche), and soluble SOD–YFP was affinity-captured using anti-YFP antibody matrix as previously described,9 with modifications as follows. After loading the matrix and washing it batchwise with PBS, four 0.2 mL washes with 50 mM Tris (pH 7.4), 8 M urea were carried out to remove SOD–YFP-associated proteins. The essentially pure SOD–YFP was then eluted in acid-urea and further processed for LC–MS, as previously described. One-eighth of the sample was loaded onto a 250 μm i.d. × 1.5 cm C18 (5 μm, Phenomenex) column, and the column was washed with buffer A (0.1% formic acid, 5% acetonitrile). This column was then attached to a 75 μm i.d. × 12 cm C18 (3 μm, Phenomenex) column and the combination mounted in-line to an LTQ Orbitrap XL mass spectrometer (Thermo Scientific) with voltage supplied directly to the column to establish nanoelectrospray ionization. Peptides were eluted with a linear gradient to 80% buffer B (0.1% formic acid, 80% acetonitrile) in buffer A. Analysis of peptide masses was performed using XCalibur and ProMass Deconvolution software (Thermo Scientific).

Cell culture studies, monitoring stability of induced SOD1–YFPs

Stable NSC-34 cell lines with Tet-regulated wt SOD1–YFP or G85R SOD1–YFP expression were produced using the Clontech system. For immunoblot analysis, cells were lysed in 50 mM Tris–HCl (pH 7.5), 150 mM NaCl, 0.5% sodium deoxycholate, 1% NP-40, and protease inhibitor cocktail (Roche completeMini). The proteins were fractionated in 10% SDS-PAGE, transferred to PVDF membrane (Millipore), and probed with mouse monoclonal antibodies against GAPDH (MAB374: Chemicon) or SOD1 (G-11; Santa Cruz Biotechnology), followed by a rabbit antimouse HRP-conjugated secondary antibody (Jackson Immuno Research Labs). Cells were induced with 0.5 μg/mL doxycycline (Dox) for 12 h to express the fusion proteins. In one protocol, they were washed to remove Dox, returned to the growth medium, and harvested at various times thereafter for immunoblot analysis (S.F.1). In the other protocol, they were washed and then incubated in growth medium with 100 μg/mL cycloheximide to arrest further translation in the presence of inhibitors of the proteasome, 2.5 μM MG-132, or of autophagy, 10 mM 3-methyladenine, and the cells incubated for various times before harvest for analysis.


The authors thank Howard Hughes Medical Institute for generous support of this work.