Studying C9orf72 dipeptide repeat polypeptide aggregation using an analytical ultracentrifuge equipped with fluorescence detection

Abstract

Sedimentation velocity, using an analytical ultracentrifuge equipped with fluorescence detection and electrophoresis methods are used to study aggregation of proteins in transgenic animal model systems. Our previous work validated the power of this approach in an analysis of mutant huntingtin aggregation. We demonstrate that this method can be applied to another neurodegenerative diseases studying the aggregation of three dipeptide repeats (DPRs) produced by aberrant translation of mutant c9orf72 containing large G₄C₂ hexanucleotide repeats. These repeat expansions are the most common cause of familial amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD). We analyzed the aggregation patterns of (Gly-Pro)₄₇, (Gly-Ala)₅₀, and (Gly-Arg)₅₀ fused to fluorescent proteins in samples prepared from D. melanogaster, and (Gly-Ala)₅₀ in C. elegans, using AU-FDS and SDD-AGE. Results suggest that (GP)₄₇ is largely monomeric. In contrast, (GA)₅₀ forms both intermediate and large-scale aggregates. (GR)₅₀ is partially monomeric with some aggregation noted in SDD-AGE analysis. The aggregation of this DPR is likely to represent co-aggregated states with DNA and/or RNA. The power of these methods is the ability to gather data on aggregation patterns and characteristics in animal model systems, which may then be used to interpret the mitigation of aggregation through genetic or molecular therapeutic interventions.

Graphical Abstract

Introduction

Protein aggregation is associated with several neurodegenerative diseases, including Parkinson’s disease, Alzheimer’s disease, Huntington’s disease, amyotrophic lateral sclerosis (ALS), and frontotemporal dementia (FTD).[1–3] Such aggregation may be associated directly with toxicity, and, if so, it is critical to identify the toxic species in the aggregation pathway, whether they might be oligomers, larger fibrils, or amyloid (or amyloid-like) plaques. While in vitro biochemical approaches can be applied to the study of protein aggregation, they don’t assess toxicity in the context of the cellular milieu. Assessing protein aggregation in situ or in vivo extends our understanding of how a cell responds to different levels of protein aggregation. Various approaches have been used to address this, including microscopy, electrophoretic methods, and more recently, analytical ultracentrifugation modified with fluorescence detection capability (AU-FDS). The latter method can be used to detect fluorescently labeled protein aggregates in complex mixtures for relevant biochemical and biomedical problems.[4–10]

In general, AU measures sedimentation behavior either at equilibrium (SE) or as a measure of transport velocity (SV), which can then be analyzed by appropriate mass transport equations. New methods to collect and analyze data over a wide range of polymer sizes using the sedimentation velocity (SV) approach have been developed, which makes it appropriate for measuring protein size (1 S to 1,000 S) and polydispersity within the same experiment for the range seen for protein aggregation associated with neurodegenerative disease.[11,12]

The development of AU-FDS [4,6] has allowed for the analysis of fluorescently labeled macromolecules. The selectivity of detecting fluorescent signals provides the opportunity to measure protein aggregates in crude extracts prepared from complex cellular and organismal model systems. We have pioneered the development of such AU-FDS methods to study aggregates produced in transgenic animal model systems,[7,9,10] and have applied methods to extend the size range observed up to 250,000 S.[7,9]

We developed this methodology through studies of polyglutamine and huntingtin aggregation in both fly and nematode transgenic models [7,9], and have published the detailed methods for much of the experimental approach applied in this work.[10,13] Semi-Denaturing Detergent Agarose Gel Electrophoresis, (SDD-AGE) [14,13], a method that has been developed to characterize the aggregate size distribution of the targeted protein population, is well suited for the study of protein aggregation associated with neurodegenerative diseases, and it is an effective quick screen for establishing levels of protein aggregation, thus complementing the information that can be gathered using AU-FDS.

We extend the application of AU-FDS to the study of fluorescently labeled polypeptides derived from aberrant expression of the C9orf72 gene principally in D. melanogaster. Mutations in the C9orf72 gene have been identified as a common familial cause of amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD).[15,16] The mutations include the insertion of variable lengths of a hexanucleotide repeating sequence with the nucleotide repeat unit being GGGGCC. Studies have demonstrated that pathogenesis can arise from both RNA toxicity and protein toxicity through various gain of function mechanisms. Protein gain of function arises in part from the translation of a set of separate dipeptide repeat sequences (DPRs) from mRNA produced from both sense and antisense DNA reading frames through a non-ATG-primed process referred to as Repeat-Associated Non-ATG translation, or RAN translation. Five unique DPRs are produced: polyGP, polyGA, polyGR, polyAP, and polyPR, and a subset of these polypeptide chains have been implicated in both protein aggregation and pathogenesis.[17–22]

Molecular interactions of some of these DPRs with cellular machinery have been identified, resulting in toxicity, with strong evidence in particular for disabling the functions of the proteasome through our own work and the work of others, [23,24] disrupting nucleocytoplasmic transport, [25–28] and by impacting cellular substructures that contain RNA and/or DNA.[31–34] The latter dysfunction is thought to be primarily owing to charge interactions of polyGR and polyPR DPRs with RNA and/or DNA, thus disrupting normal cellular interactions of such polynucleotides with arginine-rich regulatory and structural proteins.

The role of protein aggregation in causing these disruptions has not been as well studied, although polyGA sequences have been shown to form aggregates through the classic amyloid-like assembly pathway, involving fibril formation and subsequent inclusion body deposition.[29] The question of which species in the aggregation pathway is most toxic is still poorly understood, yet understanding the prevalence of such aggregation intermediates remains a critical precursor to the development of targeted therapeutics.

Materials and Methods

Drosophila melanogaster fly lines and maintenance.

The Drosophila lines used throughout the study were maintained at room temperature. Drosophila lines used to study DPR protein expression were graciously donated from the Taylor and Pandey laboratories.[27,30] Transgene expression was driven by a DaGal-4 promoter line, with the Daughterless (Da) promoter allowing for pan-tissue expression in third instar larvae.

Crude Lysate Preparation for D. melanogaster Samples.

Lysates were made from all DPR fly lines crossed with a Da-Gal4 fly line to be analyzed by biochemical and biophysical methods.[10,13] Between six and ten third instar larvae were collected for each sample and frozen at −80°C. Larvae were then washed with cold, fully supplemented 2X lysis buffer (100 mM HEPES, pH 7.3, 100 mM KCl, 2 mM EGTA, 2 mM DTT, 100 mM PMSF, protease inhibitor cocktail (Sigma-Aldrich)). Larvae were suspended in 300 μL fully supplemented lysis buffer and were flash frozen three times and allowed to thaw. Samples were transferred to a glass tissue homogenizer, and homogenized for 50 twists, placed on ice for 5 minutes, and homogenized with 10 additional twists. Homogenized samples were transferred to a new microfuge tube. After 45 minutes of gravity sedimentation, the lysate ‘supernatant’ was stored at −80°C in aliquots for SDD-AGE and AUC. Coomassie Plus Protein Assay Reagent (Thermo Scientific) was used to determine lysate protein concentration and established the amount of sample loaded in further experiments.

Caenorhabditis elegans Maintenance.

C. elegans with a genotype of snb-1/GA50::GFP+mtl-2/mCherry were maintained according to standard methods, at 18°C-20°C on nematode growth media (NGM) seeded with OP50 E. Coli.[31] Synchronized populations for crude extracts were obtained from gravid adults after treatment with 20% (v/v) alkaline hypochlorite solution (3.0 mL bleach, 3.75 mL 1 M NaOH, 8.25 mL ddH₂O) for 5 minutes and allowing eggs to hatch in M9 buffer overnight. On the next day the synchronized populations were transferred to fresh NGM plates. Synchronized day one adult C. elegans were harvested and washed with M9 Buffer (22 mM sodium phosphate, 22 mM potassium phosphate, 85 mM NaCl, 1 mM MgSO₄). Samples were flash frozen in liquid nitrogen and stored at −80°C. Protein was extracted after 3 cycles of freezing (−80°C) and thawing (on ice) in lysis buffer. Samples were homogenized in five cycles of 30 seconds each using a motorized homogenizer (Kontes) followed by a five-minute incubation on ice. Samples were allowed to settle for one hour on ice after which the supernatant was removed, flash frozen, and stored at −80°C. Total protein concentration was calculated using the Coomassie Plus Protein Assay Reagent (Thermo Scientific).

Analytical Ultracentrifugation.

Crude larval lysates used for analytical ultracentrifugation were prepared as described.[10] Experimental samples were prepared by adjusting lysates to 0.5 mg/ml or 0.25 mg/mL total protein in 1X lysis buffer (50 mM HEPES, pH 7.3, 50 mM KCl, 1 mM EGTA, 1 mM DTT, 50 mM PMSF, protease inhibitor cocktail (Sigma-Aldrich)) for a total volume of 400 μL. FC43 fluorinert heavy oil (20 μL) was loaded into two-channel charcoal/Epon sedimentation velocity cells, followed by the loading of the experimental samples. Cells were loaded into an 8-hole AnTi rotor (Beckman-Coulter) and equilibrated to 20°C in an ProteomeLab XL-A analytical ultracentrifuge (Beckman-Coulter) equipped with fluorescence detection (Aviv Biomedical). Fluorescence was detected with a 488 nm laser for excitation, using an emission filter of 520 nm with a 30 nm band pass. A multispeed method was followed, collecting 100 radial scans each, using a 20 μm step size, at 3,000 rpm, 6,000 rpm, 10,000 rpm, 20,000 rpm, 30,000 rpm, and 50,000 rpm until all monomers had cleared the solution. An example wide-distribution analysis of data from these speeds is shown in Supplementary Figure 1 for GFP-(GA)₅₀. Using the multispeed approach minimizes the potential for sample degradation by reducing the data acquisition time. While the multispeed approach results in some resolution loss due to diffusional broadening of the distributions, the broadening does not affect any of the conclusions in this work. All experiments were conducted under constant photomultiplier voltage. Data were fit utilizing the wide distribution option with SedAnal v. 6.01.[32] Radii of 6.40–6.59 cm with 0.01 cm step sizes were chosen to improve the signal to noise ratio, and 5% smoothing was applied before being graphed in Origin v. 8.6.

Semi-Denaturing Detergent Agarose Gel Electrophoresis.

Agarose gels were prepared using 1X TAE + 0.1% (w/v) SDS, and lysates were loaded using 1–10 μg of total protein.[13] All samples were prepared in 1X SDS sample buffer (125 mM Tris-HCl, 1% (w/v) SDS, 10% (v/v) glycerol, 0.1% Bromophenol Blue, pH 6.8), and incubated at room temperature for ten minutes. Gels were run in 1X TAE containing 0.1% (w/v) SDS at 45 V for 4 hours at 4°C. Proteins were then transferred overnight from the agarose gel to a nitrocellulose membrane through capillary action. Blots were processed in blocking buffer (5% (w/v) dry milk in 0.1% (v/v) PBS-Tween) for 30 minutes, and washed two times with PBS-Tween. Blots were incubated using a 1:8000 dilution of rabbit-anti-GFP primary antibody (Invitrogen) in PBS-Tween + 1% (w/v) BSA overnight at 4°C. Blocking buffer was used to wash blots, followed by incubation using a 1:12,000 dilution of HRP-IgG-antirabbit secondary antibody (Cell Signaling Technology) in blocking buffer) for 30 minutes at room temperature. Blots were then washed 4 times with PBS-Tween before being imaged by chemiluminescence using an Alpha-Innotech imager.

Confocal microscopy.

For imaging studies, third instar D. melanogaster larvae were transferred to a glass slide containing a drop of 70% glycerol and heat-shocked at 70°C for 15 seconds. The larvae were mounted and imaged on a Nikon Eclipse 80i confocal microscope equipped with EZ-C1 software, and using a 488 nm laser with a 10x objective. Focus and gain were adjusted to minimize background fluorescence, and maximize GFPt-agged protein detection. Images of whole larvae were collected at 4X, 10X, 20X magnification. For C. elegans imaging, nematodes were deposited onto a slide containing a pad of 0.5% agarose and supplemented with 5 μL 100 mM levamisole to induce paralysis. Confocal imaging was carried out similarly to that described for the D. melanogaster larvae.

Results and discussion

Our aim was to establish the ability of AU-FDS and SDD-AGE to characterize aggregation profiles of the various DPRs as produced in vivo in D. melanogaster 3^rd instar larvae. We apply both SV and SDD-AGE to characterize the aggregation states of three dipeptide repeat (DPR) constructs – (GA)₅₀, (GP)₄₇, and (GR)₅₀ – with GFP genetically fused to the N-terminal end of the DPRs, in a similar manner to our previous work in studies of the in vivo aggregation of variants of the huntingtin protein.[7,9,10,13]

Sedimentation velocity analysis.

Sedimentation velocity experiments were carried out using wide-distribution analysis (WDA) of multispeed method (MSM) data to capture a broad range of protein aggregation, encompassing s-values from 1 S to 1,000 S in a single experiment. We first compared the sedimentation of GFP-(GA)₅₀ in D. melanogaster to a control animal, expressing only eGFP (Fig. 1A). The sedimentation of eGFP gives rise to a prominent peak at about 2.2 S, consistent with the expected monomeric molecular weight of this protein. Beyond this, the only other well-defined peak is seen at 70 S, consistent with that seen in our earlier work, [7,9] and likely reflecting limited aggregation, a known property of such fluorescent proteins.[33,34] We note that eGFP contains a two-point mutation (S65T and F64L) that is 35 times brighter than wildtype GFP and is thought to have similar intrinsic aggregation profiles to GFP.[33] Such aggregation of the fluorescent protein in the GFP-(GA)₅₀ crude extract is not obvious as this region is occluded by aggregation of the polyGA itself, with a broad peak centered at 120 S, and shoulders at around 50 S and 300 S. This pool of aggregation is consistent in size range with the smear in the 3–5 mDa range observed by SDD-AGE (Fig. 2). In addition, a well-populated monomer peak for GFP-(GA)₅₀ is observed at around 2.2 S. The SV profile for this DPR is similar to that observed in our previous work on polyQ-GFP constructs, showing aggregation in the size range of 80 S to 300 S using the same MSM-WDA approach, suggesting a stable limited aggregation state.[7] A comparison across a two-fold DPR concentration difference confirms that the sedimentation profile remains largely the same, thus suggesting that hydrodynamic nonideality is not affecting the sedimentation rate [35] and that the aggregates are not susceptible to dissolution due to dilution effects at these concentrations (Fig. 1B). A similar aggregation profile is also observed in the MSM-WDA experiment for (GA)₅₀-GFP expressed in C. elegans (Fig. 1D), again suggesting that the intermediate aggregates are similar in nature and furthermore, do not depend on detailed differences in the cellular environment across model organisms.

Figure 1.

MSM-WDA analysis of DPRs expressed in D. melanogaster and C. elegans. All data profiles were subject to 5% smoothing. (A-C) D. melanogaster: (A) GFP-(GA)₅₀ (solid line) vs eGFP (dashed line); (B) GFP-(GA)₅₀ at two concentrations: 0.25 mg/mL (dashed line) and 0.5 mg/mL (solid line); (C) GFP-(GA)₅₀ (solid line), GFP-(GP)₄₇ (dashed line), and GFP-(GR)₅₀ (dotted-dashed line); (D) C. elegans (GA)₅₀-GFP.

Figure 2.

SDD-AGE analysis of DPR proteins expressed in D. melanogaster and C. elegans. Samples were fractionated on a 1.5% (w/v) agarose gel and transferred onto nitrocellulose. The blot was probed using an anti-GFP antibody. The Ctrl lane represents the da-Gal4 driver line alone, which does not express GFP.

We compare the sedimentation profiles of three DPRs: GFP-(GA)₅₀, GFP-(GP)₄₇, and GFP-(GR)₅₀ in Fig. 1C. All three dipeptides have peaks centered around 2.2 S, consistent with what is expected for the monomer. Some aggregation is also observed in the (GFP-GR)₅₀ in the 50 S to 500 S range, but at only low levels, similar to what we see in the SDD-AGE experiments (Fig. 2). In contrast to the SV profile for GFP-(GA)₅₀ and GFP-(GR)₅₀, no appreciable aggregation is seen for GFP-(GP)₄₇, consistent with that observed in the SDD-AGE experiments (Fig. 2).

SDD-AGE analysis.

SDD-AGE analysis revealed pools of monomer/small oligomer populations, intermediate aggregates, and higher order aggregates likely representing material extracted from inclusion bodies (Fig. 2). The bulk of GFP-(GA)₅₀ aggregation appears as a smeared band encompassing molecular weights above 250 kDa in size (in the range of 3–5 mDa) (Fig. 2). This pattern is consistent with work from Chang et al.[29] and our previous work on huntingtin and polyQ aggregation, which are known to form inclusion bodies made up of amyloid-like fibrillar material.[7,9] The same pattern of aggregation for GFP-(GA)₅₀ is seen in C. elegans (Fig. 2), suggesting that the aggregation process is independent of the model organism, as also supported by the SV experiment described above. A small amount of protein failed to enter the gel and may reflect larger-scale aggregation, which is not obvious in the Drosophila sample, possibly due to higher levels of expression in C. elegans. In contrast, the electrophoresis of GFP-(GP)₄₇ was confined to a narrow size range, reflecting either a very well-defined oligomeric state or an aberrant mobility of the monomeric form of this DPR. The latter interpretation is more likely since proline-rich peptides are highly extended, resulting in slower migration through the agarose pores. There was no hint of material in the well for this DPR.

Most interesting were the results for GFP-(GR)₅₀, which show a distinct tight band near the bottom of the gel, most likely representing the monomer form of the DPR, and a smear of similar size distribution seen for GFP-(GA)₅₀. These intermediate species, showing a series of three slightly more dense but compact distributions look strikingly similar to the peaks that we observed in the SV experiments described above. The ramifications of these findings are discussed below.

Conclusion

The aim of this work was to explore differential protein aggregation in C9orf72 DPRs using AU-FDS as an analytical tool to measure the size heterogeneity of the aggregate pool over a wide size range. Using D. melanogaster as our model system, we show that GFP-(GP)₄₇ does not aggregate appreciably, as evidenced by biophysical and biochemical approaches. This conclusion is buttressed by the observation by us (Supplementary Figure 2B,E) and others [36] that GFP-(GP)₄₇ is diffusely expressed as evidenced in confocal microscopy approaches. In contrast, GFP-(GA)₅₀ aggregates significantly, and our findings are consistent with confocal microscopy images revealing the presence of puncta, likely involving intermediate aggregation via an amyloid-like fibril assembly process (Supplementary Figure 2A,D).[29,36–38] The aggregation seen for GFP-(GR)₅₀ may also be associated with the puncta observed in confocal microscopy images (Supplementary Figure 2C,F) although the makeup of the aggregation is unlikely to be related to what we have observed for GFP-(GA)₅₀ aggregation.[39,40] This difference may be explained through monomer recruitment into subcellular structures, such as nucleoli and stress granules, involving principally ionic interactions. We infer the localization based on a comparison of our own confocal images (Supplementary Figure 2C,F) with previously published work on the same fly lines.[30] Finally, SV and SDD-AGE results with the GFP-(GA)₅₀ in C. elegans also are consistent with confocal images of this transgenic nematode, showing evidence of protein aggregation in the form of green fluorescent puncta (Supplementary Figure 2G).

Therapeutic interventions may be developed that target the DPRs themselves, and thus it is incumbent on researchers to understand the molecular forms of these DPRs that are presented to any potential therapeutic agent, whether it is the DPR monomer species, intermediate aggregate species, or species directly involved in inclusion bodies or other membrane-less bodies.[41,42] It is important that such agents are designed to specifically interact with the aggregate forms that are principally responsible for cellular toxicity. These techniques may then be used to track shifts in the population distribution following interventions, with important translational implications - e.g. how would a shift in the population distribution from inclusion bodies towards intermediate aggregates change the disease phenotype? We argue that the application of biochemical and biophysical tools to characterize protein aggregation not only in ALS and FTD, but in other neurodegenerative diseases involving problems with protein misfolding, is of paramount importance.

Highlights.

• Proteostasis is an important element in organismal health

• Protein aggregation is an underlying problem in several neurodegenerative diseases

• New analytical ultracentrifuge methods can be used to study protein aggregates

• Sedimentation velocity and SDD-AGE are used to study C9orf72 dipeptide repeats

• C9orf72 dipeptide repeats are a familial cause of amyotrophic lateral sclerosis

Abbreviations.

ALS

amyotrophic lateral sclerosis

AU-FDS

analytical ultracentrifugation with fluorescence detection system

BSA

bovine serum albumin

DPR

dipeptide repeat

DTT

dithiothreitol

eGFP

enhanced green fluorescent protein

EGTA

ethylene glycol tetraacetic acid

FTD

frontotemporal dementia

GFP

green fluorescent protein

HD

Huntington’s disease

HEPES

4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid

HRP

horseradish peroxidase

MSM

multi-speed method

NGM

nematode growth media

PBS

phosphate-buffered saline

PFA

paraformaldehyde

PMSF

phenylmethylsulfonyl fluoride

polyQ

polyglutamine

SDD-AGE

semi-denaturing detergent agarose gel electrophoresis

SDS-PAGE

sodium dodecyl sulfate polyacrylamide gel electrophoresis

SV

sedimentation velocity

TAE

Tris-Acetate-EDTA

TBS

Tris-buffered saline

WDA

wide distribution analysis