|Home | About | Journals | Submit | Contact Us | Français|
The N17 domain of the huntingtin protein is post-translationally modified and is the master regulator of huntingtin intracellular localization. In Huntington’s disease (HD), mutant huntingtin is hypo-phosphorylated at serines 13 and 16 within N17, and increasing N17 phosphorylation has been shown to be protective in HD mouse models. Thus, N17 phosphorylation is defined as a sub-target of huntingtin for potential therapeutic intervention. We have previously shown that cellular stress can affect huntingtin nuclear entry and phosphorylation. Here, we demonstrate that huntingtin localization can be specifically affected by reactive oxygen species (ROS) stress. We have located the sensor of this stress to the N17 domain, specifically to a highly conserved methionine at position 8. In vitro, we show by circular dichroism spectroscopy structural studies that the alpha-helical structure of N17 changes in response to redox conditions and show that the consequence of this change is enhanced N17 phosphorylation and nuclear targeting of endogenous huntingtin. Using N17 substitution point mutants, we demonstrate that N17 sulphoxidation enhances N17 dissociation from the endoplasmic reticulum (ER) membrane. This enhanced solubility makes N17 a better substrate for phosphorylation and subsequent nuclear retention. This ability of huntingtin to sense ROS levels at the ER, with phosphorylation and nuclear localization as a response, suggests that ROS stress due to aging could be a critical molecular trigger of huntingtin functions and dysfunctions in HD and may explain the age-onset nature of the disorder.
Huntington’s disease (HD) is caused by a CAG expansion in the first exon of the Htt gene, resulting in a polyglutamine expansion in the huntingtin protein (1). Huntingtin is a large, 3144 amino acid protein containing repetitive HEAT-repeat sequences that contribute to the overall solenoid structure of the protein (2). Intramolecular interactions between huntingtin sub-domains, seen in cell models (3,4) and with purified huntingtin (5), have also been shown to influence huntingtin structure. Huntingtin is a scaffold protein with no known catalytic activity, but does contain several known localization signals. At the carboxyl terminus of the protein, there is a chromosome region maintenance protein 1 (CRM1) dependent nuclear export signal (NES) (6); at positions 174–206 there is an importin beta1/beta2 dependent proline-tyrosine nuclear and ciliary localization signal (PY-NLS) (7); and the N17 domain, the first 17 amino acids at the amino terminus of the protein, is a key regulator of huntingtin localization via endoplasmic reticulum (ER) lipid-binding and an additional CRM1 NES (8–11).
Huntingtin localization is dependent on N17 post-translational modifications which include phosphorylation (12–13), acetylation (12) and SUMOylation (14,15). Phosphorylation of N17 serine 13 (S13) and serine 16 (S16) is of clinical interest as S13 and S16 are hypo-phosphorylated in HD (16). In HD mouse models, phospho-mimetic mutants of S13 and S16 (17) as well as phosphorylation induced with GM1 ganglioside treatment, were seen as protective (18). Thus, N17 phosphorylation at S13 and S16 has been established as a potential therapeutic sub-target for HD. N17 is an amphipathic alpha helix (8,10) and functions both as a NES (9) and ER membrane anchor (8) to target huntingtin to the cytoplasm under normal conditions. Stress dependent phosphorylation of S13 and S16 results in nuclear retention of huntingtin by preventing N17 binding to CRM1 and promotes huntingtin localization to nuclear puncta (9,16). At the single molecule level, N17 phosphorylation triggers a conformational change that relies on the flexibility of normal polyglutamine tracts (<36 repeats), allowing interaction with distal domains in huntingtin (3–5). What is not fully understood is what could trigger this phosphorylation in response to cell stress. We hypothesized that regulation of N17 phosphorylation occurs through modification of the N17 domain itself.
Here, we define a novel function of huntingtin as a reactive oxygen species (ROS) sensor through direct oxidation of N17 methionine 8 (M8). We show that ROS stress promotes huntingtin phosphorylation and localization to nuclear puncta. Using circular dichroism (CD) spectroscopy on synthetic N17 peptides, we show that N17 alpha-helicity is sensitive to M8 redox state and demonstrate the capacity of M8 oxido-mimetic mutation to affect N17 nuclear targeting and phosphorylation using recombinant mutant N17 fragments. These results have led to a modification of earlier mechanistic models of huntingtin localization (8,11,19), in which we now establish that M8 sulphoxidation is what triggers huntingtin ER release, allowing the soluble protein to be modified at S13 and S16. Taken together with the documented age-related increases in oxidative stress, this data also suggests that ROS stress could be the trigger of huntingtin nuclear entry and accumulation seen in HD (20) and could explain the age-onset nature of HD.
We previously reported that N17 phosphorylation and nuclear retention could be triggered by various cell stresses including the unfolded protein response at the ER (8). We also previously reported that N17 membrane association could be affected by temperature (8) and heat shock (19). However, these stresses trigger global protein misfolding and were therefore used as broad tools that likely activate a plethora of response pathways. In this study, we sought to determine a specific stress that could affect huntingtin phosphorylation and nuclear localization. In particular, we were interested in age-associated stresses, given the typical late onset of HD. We were also aware of protective effects in animal HD models of anti-oxidants such as XJB-5-131 (21) and curcumin (22). Thus, we decided to explore the role of ROS stress on phosphorylation of N17.
We first assessed the effect of oxidative stress on phosphorylation of endogenous full-length human huntingtin. We used a dose-response immunofluorescence assay in human retinal epithelial (RPE1) cells immortalized by human telomerase (hTERT) catalytic subunit. Intracellular ROS was induced using 3-nitroproprionic acid (3-NP), a mitochondrial complex II inhibitor that elevates intracellular ROS levels (23). Huntingtin phosphorylation was assessed through immunofluorescence with a previously validated, affinity purified, phospho-specific polyclonal antibody to huntingtin S13 and S16 phospho epitope (α-pN17) (16) directly conjugated to a fluorescent dye. Dosing cells with 3-NP resulted in a significant increase in both levels of phospho-N17 (Fig. 1A and B) and the number of nuclear phospho-puncta (Fig. 1A and C). Thus, endogenous human huntingtin phosphorylation and localization is responsive to ROS.
We hypothesized that the regulation of N17 phosphorylation occurs through modification of N17 itself, so we next looked for residues in N17 which could be post-translationally modified through oxidation. Huntingtin N17 sequence alignment across vertebrate species reveals conservation of M8 in mammals and evolution of a second methionine at position 4 in some reptile and bird species (Fig. 2A). Interestingly, both methionine residues are lost in the Sea Lancelet (Branchiostoma floridae) and Owl Limpet (Lottia gigantean). Methionine sulphoxidation is a post-translational modification where methionine residues can be reversibly oxidized to methionine sulphoxide through addition of a free oxygen onto the sulphur group (24). M8 oxidation has been seen to affect huntingtin small fragment aggregate properties (25) and nuclear magnetic resonance (NMR) has revealed the M8 side chain to be near the ER membrane-aqueous interface (9). This suggested that M8 was in a critical position in the N17 helix. We analyzed synthetic N17 peptide structure by CD spectroscopy, under either reducing conditions with dithiothreitol (DTT) or oxidizing conditions with hydrogen peroxide (H2O2) (Fig. 2B and C). Unlike cell studies with 3-NP, in vitro we used H2O2 because no mitochondria are present as a source of ROS. Reducing and oxidizing conditions impose contrasting shifts in N17CD spectra, with oxidation promoting a more helical structure. This indicated to us that there was a profound structural change in N17 as a result of oxidation.
Next, we tested whether this oxidation effect was due to modification of M8. Recombinant fluorescent N17 constructs, M8Q-yellow fluorescent protein (YFP) and M8L-YFP, were generated to mimic the oxidized and reduced states of N17 M8, respectively. Transfection and quantification of nuclear fluorescence in STHdhQ7/Q7 cells showed that oxido-mimetic M8Q-YFP significantly increased nuclear targeting compared to M8L-YFP and N17Wt-YFP (Fig. 3A and B). We previously defined the hydrophobic face of N17 as a CRM1-dependent NES (11) and showed interaction with CRM1 was blocked by phosphorylation at S13 and S16, promoting nuclear retention. Increased nuclear targeting of oxido-mimetic M8Q suggested that it was hyper-phosphorylated in comparison to N17Wt and that oxidation of N17 M8 promotes S13S16 phosphorylation.
We attempted CD spectroscopy on synthetic N17 peptides with the start methionine (M1) removed, however no structure was observed (Supplementary Material). This suggests that the start methionine is important to N17 structure. N17 contains two possible redox sensitive residues, M1 and M8, so we next sought to identify which was responsible for inducing the change seen in N17 alpha helicity under oxidizing conditions. CD spectroscopy was performed as described in Figure 2 on synthetic N17 peptides containing leucine point mutations at positions 1 (M1L) or 8 (M8L). In this way, only one redox target was available in each peptide. Both M1L and M8L mutants responded to redox conditions, with weaker effect than wild-type sequence (Fig. 3C–E), indicating that both methionine residues were likely contributing to redox sensing. However, mutation of M8 alone was sufficient to alter N17 localization (Fig. 3A and B).
Although it was possible that the observed nuclear accumulation of M8Q was due to hyper-phosphorylation, it was also possible that M8 oxidation directly impaired nuclear export independently of S13S16 phosphorylation. To address this, we assessed nuclear export of recombinant oxido-resistant and oxido-mimetic N17 constructs using leptomycin B to impair export from the nucleus and quantify the resulting nuclear accumulation. In the context of phospho-resistant S13 and S16 alanine mutants (S13AS16A), N17 with either M8A or M8Q was still responsive to leptomycin B, and hence contained a functional NES (Fig. 4A and B). These data establish that M8 oxidation does not directly affect the NES activity of N17. Interestingly, a striking difference was also noted between the localizations of M8Q (Fig. 3A, Panel c) and M8QS13AS16A (Fig. 4A, Panel e). The absence of M8QS13AS16A nuclear accumulation further indicates the involvement of serine phosphorylation in M8Q nuclear localization and supports the hypothesis that N17 oxidation indirectly inhibits nuclear export by promoting N17 phosphorylation (11).
Using a quantitative pull-down assay, we directly showed that M8 oxidation resulted in enhanced N17 phosphorylation. We performed western blot analysis of N17Wt-YFP, M8Q-YFP and M8L-YFP recombinant protein following YFP-immunoprecipitation from STHdhQ7/Q7 cells. N17 moieties were assessed for their level of phosphorylation using western blot analysis with α-pN17, normalizing to α-YFP as an internal loading control (Fig. 5A). Oxido-mimetic M8Q-YFP was shown to be phosphorylated approximately 2.5 times more than N17Wt-YFP (Fig. 5B). Thus, a mimic of oxidation of N17 M8 is sufficient to promote phosphorylation.
M8 oxidation alters the structure and promotes the phosphorylation of N17. We hypothesized that methionine oxidation disrupts N17 anchoring to the ER membrane, enhancing the solubility and kinase accessibility of the protein. To test that solubility modulated N17 phosphorylation, we compared the phosphorylation of recombinant E5AE12A and M8P N17 mutants, which have previously been shown to have vesicular and un-targeted localization (8). We first validated the targeting of these mutants using a Pearson’s correlation with mCherry or ER Tracker™ dye (Fig. 6A and B). YFP was used as a control for both maximum correlation (with mCherry) and minimum correlation (with ER tracker). E5AE12A-YFP showed poor correlation to mCherry (below 0.4) and high correlation with ER tracker (above 0.9), indicating poor solubility. In comparison, M8P-YFP showed poor correlation to the ER tracker signal but a very high correlation to mCherry. Therefore, M8P-YFP is fully soluble, whereas E5AE12A-YFP is not.
We then compared these two mutants in a similar assay as in Figure 5 and could note a significantly higher phosphorylation of M8P-YFP but significantly lower modification of E5AE12A-YFP (Fig. 6C and D). The elevated phosphorylation of soluble N17, and the reduced phosphorylation of membrane bound N17, indicates that the degree of membrane association affects N17 phosphorylation. Taken together with the previous evidence, the increase in N17 phosphorylation from both oxidation and solubility suggests that M8 oxidation acts to promote N17 phosphorylation by increasing solubility and accessibility to kinases, or perhaps removing N17 from the proximity of phosphatases at the ER.
Here, we show that huntingtin has a normal biological function as a ROS sensor, specifically through direct oxidation of N17 M8. In support of this novel, role of huntingtin and the importance of M8 in N17 biology are observations from our sequence alignment of N17 (Fig. 2). Many bird and reptile species, which are historically considered to have high basal metabolic rates and long lifespans, respectively, have evolved a second methionine at position 4 in addition to the one found at position 8. In contrast, largely immobile species, such as the owl limpet and lancelet, have no methionines in N17 outside of M1. The correlation between N17 methionine number and species metabolism and lifespan supports the concept that methionines present in N17 are involved in sensing oxidative stress. It may be that species with high-ROS loads have evolved additional methionine sensors in N17 for increased regulation or sensitivity to oxidative stress. Huntingtin M8 oxidation has also been seen to affect huntingtin small fragment overexpression aggregate properties (25), suggesting M8 oxidation also affects the properties of mutant huntingtin in the disease context. In Parkinson’s disease, alpha-synuclein is oxidized at methionines (26). Like huntingtin, alpha-synuclein contains a small alpha-helical lipid interaction domain (27), suggesting a similar method of regulation by oxidation may be occurring. As adult stem cell neurogenesis is triggered by ROS signalling (28), deficits in neurogenesis within the striatal interneuron population of adult HD brains (29) is also consistent with the role of huntingtin as an oxidative stress sensor and suggests dysfunction of this role may be occurring in HD.
We suggest a model in which huntingtin trafficking between the ER membrane and nuclear puncta is controlled by increased ROS levels in the cytoplasm and ER (Fig. 7). We hypothesize that huntingtin, anchored to membranes by N17, can sense elevated ROS levels through oxidation of methionines in N17, primarily M8. This oxidation induces an increase in N17 alpha helicity and, as suggested by the solubility assays, promotes release of huntingtin from membranes. Previous solid-state NMR data has shown that M8 is located very near the ER membrane-aqueous interface (9), and is presumably accessible by reactive oxygen atoms. Increased huntingtin solubility enhances kinase accessibility, resulting in increased levels of phosphorylated huntingtin. Dissociation of huntingtin from the ER membrane also allows nuclear entry mediated by a downstream PY-NLS (7). Phosphorylation of N17, which we have previously shown to regulate nuclear export by inhibiting CRM1 interaction (11), allows huntingtin nuclear accumulation and localization to phospho-puncta (16).
This model is consistent with the nuclear accumulation of mutant huntingtin seen in HD (20). We propose that M8 oxidation is still able to enhance the solubility and nuclear translocation of mutant huntingtin, however export from the nucleus is impaired. Because mutant huntingtin is hypo-phosphorylated (16), the observed nuclear accumulation is likely due to the altered conformation of the polyglutamine expansion preventing CRM1 interaction, not enhanced phosphorylation as in the wild-type pathway. It is possible that steric hindrance from the polyglutamine expansion also prevents kinase accessibility which results in the observed hypo-phosphorylation of mutant huntingtin. This explanation is consistent with a previous study which showed that polyglutamine expanded huntingtin, rendered fully soluble through M8P mutation, increased cellular toxicity and nuclear accumulation (8). This further indicates that the soluble form of mutant huntingtin contributes to nuclear accumulation and toxicity. It also implies that the protective effects of anti-oxidants in HD models (21,22) function by lowering the oxidative environment of the cell. This reduces the oxidation, and therefore solubility and toxicity, of mutant huntingtin. We suggest that soluble, non-phosphorylated, nuclear mutant huntingtin is the toxic species in HD which accumulates in the nucleus due to its altered conformation preventing CRM1 binding, which is consistent with the other past studies by us (30) and by others (31).
Given the nuclear localization of huntingtin following oxidative stress it would next be interesting to explore the purpose of huntingtin ROS-sensing capabilities and the role huntingtin could be playing in the nucleus in response to oxidative stress. As oxidative stress causes DNA damage, one promising hypothesis is that huntingtin is involved in DNA damage repair. In support of this concept, a recent genome wide association study, that looked to identify genetic modifiers of HD age-onset, almost exclusively uncovered pathways involved in DNA damage, mitochondrial function and peroxisome function (32). This suggests that ROS and DNA damage are critical determinants of HD age of onset. A role of huntingtin in DNA damage repair is consistent with the transcriptional regulation of huntingtin by p53, given the known role of p53 in DNA damage response (33). It could also explain why Ataxia Telangiectasia Mutated heterozygous knockout crosses with HD model mice show improved HD disease phenotypes (34). Several groups have also reported that somatic expansion of the inherited mutant huntingtin CAG tract occurs through the removal of oxidized base lesions by DNA repair pathways (35–38). Future studies will address the exact role of huntingtin in the nucleus at DNA in response to ROS stress.
The role of huntingtin as an oxidative stress sensor presents an interesting connection between age-related increases in oxidative stress and HD age of onset. We speculate that the recognition and response of huntingtin to oxidative stress is a secondary function which becomes active later in life with increasing ROS load. In the case of mutant huntingtin, the triggering of this secondary function by ROS may initiate HD by prompting mutant huntingtin nuclear accumulation and resulting pathology. Interestingly, as mice have a relatively low-ROS load due to their short lifespan, this may also explain why clinically relevant allele lengths do not generate an obvious phenotype in mouse models (39). The synthetic polyglutamine repeats, in excess of 100, used in mice to elicit disease phenotype may in fact mimic the somatically expanded huntingtin seen in patients at age-onset, essentially skipping the premanifest stage of the disease. This suggests that modelling HD in mice with mutant huntingtin and additionally using ROS stress, such as the historic 3-NP disease model of striatal degeneration (23), may provide a more accurate model of human disease. The role of huntingtin as an oxidative stress sensor may also contribute to tissue specificity of HD. Brain and cardiac tissues are regions of high ROS in the body and cardiac failure is historically a major cause of death in HD (40). In these regions of high ROS, it may be that huntingtin function as an oxidative sensor is more prominent, and therefore sensitive to mutant huntingtin dysfunction toxicity.
In conclusion, we show here that huntingtin functions as an oxidative stress sensor and provide a specific molecular mechanism of ROS sensing through oxidation of M8 in N17. We suggest that ROS stress due to aging could be a critical molecular trigger of disease by prompting huntingtin nuclear entry, and may be a contributing factor in explaining the age-onset and tissue specificity of HD.
T-antigen immortalized mouse STHdhQ7/Q7 (a kind gift from M.E. MacDonald) were grown as previously described. hTERT-immortalized retinal pigment epithelial cells were cultured at 37°C in Dulbecco’s Modified Eagle’s Medium (DMEM) (Life Technologies) containing 10% fetal bovine serum (FBS) (Life Technologies), 0.26% NaHCO3 (Bioshop), and 0.01mg/mL hygromycin B (Invitrogen).
Cells were washed with phosphate buffered saline (PBS) then fixed and permeabilized with ice-cold methanol at−20°C for 12min. Cells were washed three times with PBS in 5min intervals at room temperature then blocked with 2% FBS in PBS three times in 12min intervals at room temperature. Primary phospho-N17 antibody (New England Peptide) directly conjugated to Alexa 488 was diluted 1:15 in antibody solution (2% FBS, 0.02% (v/v) Tween-20 in PBS) and incubated overnight at 4°C. Finally, washes in PBS were performed as previously described, and nuclei were counterstained with Hoechst dye for 15min at room temperature prior to imaging in PBS.
To directly conjugate the primary phospho-N17 antibody to Alexa Fluor 488, the phospho-N17 antibody was incubated with 1 µl Alexa Fluor 488 carboxylic acid succinimidyl ester (Life Technologies) per 10 µL antibody and 5% NaHCO3 overnight at 4°C on a rotator. The solution was then run through a Sephadex bead (Amersham Pharmacia Biotech AB) column and collected to the dye front.
Cells were treated with 3-NP acid (Sigma), prepared in in PBS at pH 8.0 and diluted in DMEM, for 1h at 37°C. Image intensity was calculated using MATLAB in 40 images over four trials. Images were thresholded to the cell using the Otsu method and average intensity was calculated per image. Nuclear puncta were also quantified using an open source speckle counting pipeline in Cell Profiler (www.cellprofiler.org). The number of puncta per cell was counted in>500 cells over four trials.
CD was performed on an Aviv CD spectrometer model 215. Peptides were obtained from Genscript or New England Peptide and used at 0.25mg/mL solutions in phosphate buffer (10 mm NaH2PO4, 0.14 M NaCl, 1 mm ethylenediaminetetraacetic acid, pH 7.4). Solutions were scanned between 260 nm and 195 nm and maintained at 25°C. Triplicate scans were averaged and secondary structure determination was done using the CONTILL algorithm against a database of 56 proteins, 43 soluble and 13 membrane proteins.
Double-stranded synthetic DNA oligonucleotides were generated (McMaster MOBIX facility) encoding the first 17 amino acids (Wt, M8Q, M8P and M8QS13AS16A mutants) of huntingtin containing a Kozac consensus translation start site at the start methionine and 5’ AgeI and 3’ SalI sites. Annealed oligonucleotides were cloned between AgeI/SalI sites of peYFPN1 (BD Biosciences/Clontech) containing a peYFPC1 multiple cloning site to create Htt N17-eYFPN1 plasmids. Subsequent M8L, M8AS13AS16A and E5AE12A N17 mutant peYFPN1 plasmids were created using the Q5 site-directed mutagenesis kit from New England Biolabs as directed. All DNA manipulation enzymes were purchased from New England Biolabs. All plasmid constructs were verified by nucleotide sequencing (McMaster MOBIX facility).
STHdhQ7/Q7 cells were transfected using Turbofect (Fermentas) and imaged approximately 24h after transfection. Cells were fixed with 4% (w/v) paraformaldehyde in PBS at 33°C for 20 min and nuclei were counterstained with Hoechst dye for 15min at room temperature prior to imaging in PBS. In cells treated with leptomycin B (Sigma), 10 ng/mL was applied for 30 min. Percent nuclear fluorescence was calculated in approximately 100 cells per construct over 3 trials using Image J. Nuclear and whole cell areas were defined using Hoechst staining and YFP fluorescence, respectively. YFP fluorescence intensity was calculated for nuclear and whole cell areas and percent nuclear fluorescence was calculated using the equation: percentage nuclear fluorescence = (nuclear intensity/whole cell intensity) × 100.
Microscopy was performed on a Nikon Ti Eclipse epifluorescence inverted microscope using a Plan Apochromatic 60x oil immersion objective and Orca-Flash 4.0 digital camera (Hamamatsu). Widefield microscopy was performed using a light-emitting diode Lumencor Spectra Light Engine, dichroic filters from Semrock and filter wheel from Sutter Instruments. Confocal imaging was performed using 405 nm and 555 nm laser lines and Nikon C2 confocal head. All image acquisition was done using Nikon NIS-Elements Advanced Research version 4.1.1 64-bit acquisition software.
STHdhQ7/Q7 cells were transfected with YFP fusion proteins as described above. 24h post transfection, cells were lysed in NP40 lysis buffer (150 mm NaCl, 50 mm Tris pH 8.0, 1% NP40) containing 1X protein and phosSTOP inhibitor cocktails (Roche). Supernatants were incubated with Living Colors GFP polyclonal antibody (Clontech), or without antibody as a control, for 15 min on ice followed by overnight with rotation at 4°C. Protein A agarose beads (Sigma) in NP40 lysis buffer were added for 1h with rotation at 4°C and were subsequently washed with PBS. Purified proteins, and unaltered supernatant diluted 1:10, were resolved on a 12% sodium dodecyl sulfate-polyacrylamide gel and electroblotted to a poly vinyl difluoride membrane. Membranes were blocked with 5% non-fat dry milk in Tris-buffered Saline Tween 20 for 1h followed by overnight incubation with the phospho-N17 antibody (1:2500). Rabbit anti-mouse horse radish peroxidase-conjugated secondary antibody (1:50000) (Abcam) was incubated for 45min. Bands were visualized using Immobilon Western Chemiluminescent Substrate (Millipore) and Microchemi (DNR Bio-Imaging Systems).
Transfection and fixation of STHdhQ7/Q7 were performed as described above. Co-transfection of YFP fusion proteins with mCherryC1 (BD Biosciences), or counterstaining with ER Tracker™ Red (Molecular Probes), was performed prior to fixation to visualize the cells soluble phase or ER, respectively. Pearson correlation quantification between red and green channels was done per cell using the Coloc2 ImageJ plugin on defined region of interest. Every cell was represented as its own N for each condition. Fifty cells were quantified for each condition over three trials and graphed on a box-whisker plot. The box represents 25–75% of data and the line indicates the data median. The whiskers represent 5–95% of data and outliers were removed if they fell outside 2 standard deviations from the mean.
All statistical analysis was done using the Real Statistics Excel (Microsoft) add-on. If data passed Shapiro–Wilk normality assumptions groups were compared using a Student’s t-test. If data was not normal, groups were compared using the Mann–Whitney method.
Supplementary Material is available at HMG online.
The authors would like to thank Raquel and Richard Epand for assistance and advice with CD spectroscopy.
Conflict of Interest statement. None declared.
This work was supported with operating grants to R.T. from the Huntington Society of Canada, the Canadian Institutes of Health Research (CIHR MOP-119391), the Krembil Family foundation and CHDI Inc.