A kinetic assessment of the C. elegans amyloid disaggregation activity enables uncoupling of disassembly and proteolysis

Abstract

Protein aggregation is a common feature of late onset neurodegenerative disorders, including Alzheimer's disease. In Alzheimer's disease, misassembly of the Aβ peptide is genetically linked to proteotoxicity associated with disease etiology. A reduction in Aβ proteotoxicity is accomplished, in part, by the previously reported Aβ disaggregation and proteolysis activities–under partial control of heat shock factor 1, a transcription factor regulating proteostasis in the cytosol and negatively regulated by insulin growth factor signaling. Herein, we report an improved in vitro assay to quantify recombinant fibrillar Aβ disaggregation kinetics accomplished by the exogenous application of C.elegans extracts. With this assay we demonstrate that the Aβ disaggregation and proteolysis activities of C.elegans are separable. The disaggregation activity found in C.elegans preparations is more heat resistant than the proteolytic activity. Aβ disaggregation in the absence of proteolysis was found to be a reversible process. Future discovery of the molecular basis of the disaggregation and proteolysis activities offers the promise of delaying the age-onset proteotoxicity that leads to neurodegeneration in a spectrum of maladies.

Introduction

Protein aggregation is genetically and biochemically linked to the development of human neurodegenerative disorders, such as Alzheimer's disease (AD), Parkinson's disease (PD), Huntington's disease (HD), the familial amyloidoses, and the prion disorders.1–7 Aggregation of the 40- and 42-amino-acid amyloid beta peptides (Aβ₄₀ and Aβ₄₂) putatively causes AD.8 Whether Aβ proteotoxicity arises from within the neuron, outside the cell, or both, how toxicity is effected and what the structure(s) of the cytotoxic agent is, are key unanswered questions. A prominent risk factor common to all of these neurodegenerative diseases is aging.9–14 Insulin/Insulin growth factor-1 signaling (IIS) is perhaps the most prominent signaling pathway regulating aging. Reduced IIS is linked to longevity in worms,15 flies, mice,16 and humans.17, 18 In worms, lifespan extension facilitated by IIS reduction (either by mutating the sole insulin/IGF receptor, daf-2, or by the application of RNAi directed against daf-2) is dependent upon the activities of the transcription factors DAF-16 and HSF-1.19, 20

Why aging is the prominent risk factor in human neurodegenerative diseases is not completely understood. Some insight regarding the connection between aging and neurodegeneration is provided by recent studies demonstrating that delaying aging in organismal models of HD and AD by IIS suppression ameliorates aggregation-associated proteotoxicity.9, 12, 19, 21, 22 The molecular basis for the link between the reduction in IIS and diminished proteotoxicity is likely because aging signaling pathways and their transcriptomes have a strong influence on the protein homeostasis (or proteostasis) network, that is, the biological pathways that maintain our proteome and minimize aggregate accumulation.9 For example, both the heat shock response controlling proteostasis in the cytosol, and longevity mediated by reduced IIS require the transcription factor, HSF-1. As is the case for lifespan extension, both the DAF-16 and HSF-1 transcriptomes are required to ameliorate Aβ₄₂ proteotoxicity via IIS suppression.22

Speculation23 that a specialized mammalian disaggregation activity, possibly involving chaperones, enables aggregate degradation has some experimental support.22, 24 Active and passive immunization studies resulting in the clearance of Aβ deposits from AD animal models suggest that inducing degradation pathways can be therapeutically beneficial.25, 26 In one potential model of the disaggregase activity, the disassembly of fibrillar Aβ aggregates is envisioned to be driven by proteolysis of Aβ monomers, which would lead to the disassembly of Aβ aggregates by depleting the monomers below their critical concentration, causing dissociation of the fibrils to repopulate the disappearing monomer population.27, 28 Alternatively, disaggregation via a specific cellular pathway or utilizing a macromolecular complex could take apart Aβ fibrils, independent of proteolysis. This scenario would enable reaggregation of Aβ into fibrils if Aβ monomer proteolysis was inhibited and the disaggregase activity was depleted. Discovering the molecular underpinnings of the C.elegans fibril disaggregase pathway(s)/activity(ies) and/or the degradation pathway(s)/activity(ies) could reveal novel therapeutic strategies for human neurodegenerative diseases.9, 22 In this manuscript, we focus primarily on the disaggregase activity.

The apparent importance of disaggregation and degradation in countering Aβ₄₂ aggregation-associated proteotoxicity and the linkage of this biological pathway(s) or activity(ies) to aging-associated signaling pathways22 motivated us to optimize and further scrutinize the Aβ disaggregation assay to discern whether the disaggregation and proteolysis activities are separable. In this study, the Aβ amyloid fibril-thioflavin T (ThT) fluorophore complex prepared in vitro was treated with C.elegans homogenate to discern the rate of fibril disassembly and the degree of Aβ proteolysis in the presence or absence of selected protease inhibitors. We utilized the decrease in ThT fluorescence to follow disaggregation (verified by atomic force microscopy) and the decrease in immunoreactivity to Aβ antibodies and/or an HPLC assay to monitor the degradation of Aβ. These studies demonstrate that disaggregation does not require proteolytic degradation, as these activities are separable.

Results

C. elegans homogenate possesses Aβ₄₀ fibril disaggregation activity

To assess the disaggregation activity within the C. elegans AD model, we prepared cellular extracts by homogenizing worms grown on bacteria to day one of adulthood. The debris from Aβ worm homogenization was cleared by low speed centrifugation (as detailed in the Methods section) and the resulting postdebris supernatant (PDS) was used in subsequent experiments. Since the disaggregation machinery might be inducible, we used worms that express Aβ₄₂ in their smooth body wall muscles under control of the unc54 promoter that encodes the myosin heavy chain gene and is expressed from embryogenesis through adulthood (strain CL 200621), hereafter referred to as Aβ worms. We assumed that Aβ₄₂ overexpression in these worms would be sufficient to induce this activity. These Aβ worms were grown on bacteria to day one of adulthood, at which stage no proteotoxicity is observed.21, 22 Worms not expressing Aβ₄₂ (wild-type worms, strain N2) were used as controls. Worm PDS was added to preformed Aβ₄₀ (or Aβ₄₂) fibrils in vitro to assess its disaggregation activity.

Aβ fibrils were generated from monomerized Aβ, prepared from Aβ₄₀ (or Aβ₄₂) peptide dissolved in hexafluoroisopropanol (HFIP) overnight, then lyophilized, dissolved at pH 10.5, and filtered through a low molecular weight membrane filter (Microcon YM-10, Millipore) to remove covalent and noncovalent aggregates.29 Fibrillar aggregates were then formed by incubating the initially monomeric Aβ peptide (50 μM) at 37°C at pH 7.4 for 3 days, during which time the sample was agitated using an overhead rotary shaker (20 rpm). Amyloidogenesis was monitored by ThT fluorescence (Supporting information Fig. S1A). ThT is a fluorophore that exhibits a higher fluorescence quantum yield and a red shift upon binding to fibrils or spherical aggregates.30 Aggregates were characterized by circular dichroism spectroscopy to observe the expected increase in β-sheet structure (Supporting information Fig. S1B), and by atomic force microscopy (AFM) to confirm the fibril morphology (Supporting information Fig. S1C). Fibrillar aggregates were sonicated for 30 min using a water bath sonicator to afford a uniform (200–500 nm) length distribution of fibrils (Supporting information Fig. S1D). Disaggregation of the presonicated fibrils was monitored in a fluorescence plate reader by the decrease in ThT fluorescence over time, with data being collected every 10 min following 5 s of shaking.31 Disaggregation of Aβ fibrils was confirmed by AFM.22

The ability of worm PDS to disaggregate Aβ₄₀ fibrils in phosphate buffer (150 mM NaCl, 50 mM Na-phosphate, pH 7.4) at 37°C was scrutinized by the addition of Aβ worm PDS at final total PDS protein concentrations ranging from 25 to 500 μg/mL [Fig. 1(A)]. ThT (20 μM) fluorescence decreased in the presence of worm PDS, whereas incubation with buffer alone (black circles) did not significantly change the ThT fluorescence. Disaggregation time courses for all PDS reactions were fitted by mono-exponential decay functions [Fig. 1(A), lines]. Disaggregation time constants were 4.6 ± 0.6, 3.4 ± 0.3, 2.5 ± 0.2, and 1.5 ± 0.1 h for PDS exhibiting a total protein concentration of 25, 50, 100, and 500 μg/mL, respectively. While increasing PDS concentrations accelerated disaggregation as expected, increasing PDS concentrations also increased residual ThT fluorescence, either as a consequence of fluorescence induced by ThT binding to worm protein(s) or as a result of competing activities in the PDS, such as proteolysis which could disable the disaggregase activity.

Figure 1. Aβ₄₀ disaggregation activity A: Disaggregation activities of C. elegans PDS as a function of total protein concentration (25–500 μg/mL). Lines show mono-exponential decay function fits to the data (open symbols). B: Disaggregation activities of PDS (20 μg/mL) from Aβ₄₂-expressing (CL for CL2006) and wt (N2) C. elegans fed bacteria expressing hsf-1 RNAi or empty vector controls (EV). Neither the disaggregation rate nor its reduction by hsf-1 downregulation depend on Aβ₄₂ expression. C: PDS-mediated Aβ₄₀ disaggregation is not discernibly inhibited when ATP in the PDS is hydrolyzed by pretreatment with apyrase (AP, 0.1 and 0.5 U, 15 min, 30°C); average relative ThT signals of quintuplicate samples ± SD. D: PDS-mediated (PDS, 25 μg/mL) Aβ₄₀ disaggregation kinetics in the presence of 0.025–0.2% SDS. ThT signals of detergents in buffer were subtracted from the Aβ signals. E: Relative ThT fluorescence signals after 10 h incubation with PDS in the presence of 0.025–0.2% SDS, Tween-20, or NP-40. F: Separation of PDS-mediated disaggregation and proteolysis products of Aβ₄₀ by denaturing PAGE visualized by Western blot analysis (mAb 6E10) of fibrillar Aβ₄₀ before and after PDS treatment. Aβ samples were sonicated for 0, 15, 30, or 60 min before disaggregation. Arrows indicate apparent monomer (M), dimer (D), and trimer (T) bands as well as high-MW aggregate (A) bands. Control: fibrillar Aβ₄₀ incubated for 4 d in the absence of PDS.

Next, we compared the disaggregation activities of Aβ worms to those of wt control (N2) worms to determine whether the C. elegans disaggregation activity or its inhibition by hsf-1 downregulation depends on the expression of the Aβ peptide. PDS (25 μg/mL total protein) prepared from Aβ₄₂-expressing (CL2006) worms or N2(wt) worms grown on hsf-1 RNAi or on noncoding (EV) RNAi was added to presonicated fibrillar Aβ₄₀ and Aβ disaggregation was monitored by ThT fluorescence for 36 h [Fig. 1(B)]. Downregulation of hsf-1 expression resulted in a reduced disaggregation activity in both wt and Aβ worms (cf. open and closed symbols). Though small differences between the disaggregase activities could be observed in both worm strains, it is clear that the disaggregase activity is constitutive. That said, Aβ expression could lead to more HSF-1 activity and slightly enhanced activity.

We then proceeded to further characterize the disaggregation activity by exploring the influence of NaCl concentration and the length of time the Aβ fibrils were sonicated before being treated with Aβ worm PDS. NaCl concentration had little influence on the disaggregase activity (Supporting information Fig. S2A), whereas prolonged presonication of Aβ fibrils hastened disaggregation (Supporting information Fig. S2B). While sonication is not required to achieve disaggregation (Supporting information Fig. S2B), the number of accessible fibril ends is a significant factor in determining the disaggregation rates. On the other hand, NaCl, which is known to promote fibril formation and would be expected to electrostatically stabilize Aβ amyloid fibrils,32 does not have a measurable influence.

Disaggregation could be an ATP driven process. To test this hypothesis, we preincubated Aβ worm PDS with apyrase (0.1 and 0.5 U) for 15 min at 30°C before adding the treated PDS to Aβ₄₀ fibrils (15 μM_monomer) in PBS. Under these conditions, we have demonstrated that the accessible ATP is hydrolyzed (see Methods). No dependence of the disaggregation activity on apyrase concentration was observed, as ascertained by monitoring ThT fluorescence intensity at 75 h (50% complete) and 120 h (80% complete) into the disaggregation time course, [Fig. 1(C)]. Likewise, no dependence of the disaggregation activity on ATP was observed in the presence of protease inhibitor cocktail or upon addition of the competitive inhibitor ATP-γ-S. However, these experiments do not definitively exclude a requirement for ATP, as vesicular ATP could drive the process. Additional experiments are ongoing to further scrutinize the ATP dependence of the disaggregase activity.

Worm PDS was subjected to centrifugation to separate the soluble and membrane components. We found that the majority of the disaggregation activity could be pelleted at 100,000 × g (Supporting information Fig. S2C), suggesting that the disaggregation activity can associate with membranes. We therefore assessed the impact of the nonionic detergents NP-40 and Tween-20, and the ionic detergent SDS [Fig. 1(D,E)] on the disaggregation activity of PDS at detergent concentrations ranging from 0.025 to 0.2%. The detergents SDS and NP-40 largely eliminate disaggregation activity at concentrations ≥0.1% (w/v), whereas Tween-20 induced a moderate reduction in disaggregation activity over the whole concentration range [Fig. 1(E)]. No clear correlation of the detergents' effects with their respective critical micelle concentrations could be inferred.

SDS accelerated disaggregation at concentrations below 0.1%, suggesting that SDS may weaken Aβ fibril structures by partial denaturation and thus promote disaggregation. Independent of the disaggregation activity, SDS concentrations ≥ 0.1% led to a gradual increase in the Aβ fibril fluorescence signal [Fig. 1(D,E)], suggesting that SDS may promote the accessibility of Aβ fibrils for the ThT dye. Background fluorescence signals resulting from ThT binding to detergent micelles were subtracted from all ThT kinetic measurements. Detergent-only ThT fluorescence remained constant over the course of the experiment (Fig S3).

The ability of the detergents to inhibit the disaggregase activity at higher concentrations is compatible with several interpretations, including the hypothesis that the disaggregation activity is membrane associated and/or that the disaggregase has a quaternary structure that is detergent sensitive. It is also possible that detergents disrupt the interaction between the disaggregation machinery and the Aβ fibrils. How the detergents influence the disaggregation activity is an area of ongoing investigation.

The disaggregated Aβ peptide is proteolyzed by worm PDS

To quantify the Aβ₄₀ concentration before and after disaggregation by PDS, Aβ₄₀ fibrils were sonicated for 0, 15, 30, or 60 min and then subjected to disaggregation by PDS [Fig. 1(F)]. SDS soluble and insoluble Aβ₄₀ was quantified by Western blot analysis using two separate monoclonal Aβ antibodies (clones 6E10 and 4G8) to visualize Aβ before (t = 0) and after being subjected to worm PDS for 4 days [Fig. 1(F), 6E10 data shown, see Supporting information Fig. S2D for 4G8 data]. The amounts of monomeric (M), apparent dimeric (D), and apparent trimeric (T) forms of SDS soluble Aβ₄₀ were dramatically decreased after treating the Aβ₄₀ fibrils with PDS for 4 days, implying that after disaggregation, Aβ₄₀ peptides were proteolyzed by components of the worm PDS. Prolonged sonication did not affect the amount of Aβ₄₀ detected before disaggregation, but reduced the residual monomer, dimer, and trimer bands observed after treatment with PDS, consistent with more complete disaggregation and proteolysis when starting with smaller fibrils [Fig. 1(F), Supporting information Fig. S2B]. Note the steeper decline in the ThT signal of Aβ₄₀ fibrils subjected to prolonged sonication before treating them with PDS (Supporting information Fig. S2B).

The C. elegans disaggregation activity is separable from the proteolytic activity

As mentioned earlier, different models could explain the PDS-mediated Aβ₄₀ disaggregation and proteolysis activity observed. First, proteolysis of monomers could drive disaggregation by lowering the monomer concentration below the critical concentration for Aβ fibril formation.28 A second possibility is that direct proteolytic digestion of fibrillar Aβ could also result in disaggregation. Since these models require proteolysis for disaggregation, inhibiting proteolysis should preclude disaggregation if these models apply. An alternative model recognizes that disaggregation and proteolysis could be separable activities and allows for disaggregation in the absence of proteolysis. Both activities could either be mediated by different components or pathways of the C. elegans proteome or by the same component(s) or pathway(s), having distinct active sites.33

To distinguish between these two mechanistic categories, we tested whether disaggregation and proteolysis could be uncoupled utilizing protease inhibitors. Fibrillar aggregates of either synthetic Aβ₄₂ or Aβ₄₀ peptides were treated with worm PDS in the absence or presence of protease inhibitors. Sodium azide (0.02%) was added to preclude potential proteolytic activity caused by bacterial growth during disaggregation time courses of up to 75 h. Both Aβ₄₀ and Aβ₄₂ fibrillar aggregates are readily detected by AFM in ThT-positive samples before, but not after incubation with PDS, either in the presence or absence of Roche complete protease inhibitor cocktail (PIC) [Fig. 2(A) and Ref. 22] demonstrating disaggregation. Very similar ThT-monitored disaggregation time courses were observed in the absence and presence of PIC for Aβ₄₂ fibrils [Fig. 2(B)], as well as for the first 24 h of Aβ₄₀ disaggregation reaction [Fig. 4(A)], suggesting that proteolysis is not required for disaggregation.

Figure 2. Disaggregation activity is distinct from proteolysis. A: Atomic force microscopy images of fibrillar Aβ₄₂ and Aβ₄₀ incubated in the presence or absence of C. elegans PDS (25 μg/mL) and Roche complete protease inhibitor cocktail (PIC). Scale bars = 200 nm, height scale = 20 nm. Incubation with PDS leads to a loss of readily detectable fibrillar structures both in the presence and absence of PIC. B: Kinetic measurement of Thioflavin T fluorescence of Aβ₄₂ incubated for 75 h with PDS in the presence or absence of PIC. C: Western blot of (mAb 6E10) of monomeric Aβ₄₀ peptide (non agg.), Aβ₄₀ aggregates (agg. control), and the disaggregation products resulting from PDS treatment (25 μg/mL) in the absence or presence of PIC. Arrows indicate apparent monomer (M), dimer (D) and high-MW aggregate (A) bands. D: Coomassie staining reveals monomeric and tetrameric bands of non-aggregated (non agg.) Aβ₄₂. Western blot of Aβ₄₂ aggregates (agg. control), and the disaggregation products resulting from PDS treatment (25 μg/mL) in the absence or presence of PIC. Aβ₄₂ is degraded and proteolyzed after incubation with C. elegans PDS for 72 h in the absence of PIC; bands labeled as in C, in addition a soluble aggregate band was observed (A_sol). E: Aβ aliquots from disaggregation assays were analyzed by dot blots and filter retardation assays, respectively, after incubation of Aβ₄₀ for 72 h to quantify the amounts of total Aβ and SDS-resistant Aβ aggregates. F: Quantification of filter retardation assay and dot blot data. Error bars indicate standard deviations in three independent disaggregation experiments. SDS-resistant, aggregated Aβ is reduced after incubation with C. elegans PDS. Total amounts of Aβ peptide remain constant after incubation in the presence of PIC or phenylmethylsulfonylfluoride (PMSF), but decrease in the absence of protease inhibitors.

To test whether the addition of PIC prevented proteolysis of Aβ₄₀ and Aβ₄₂ after PDS-mediated disaggregation for 72 h, the Aβ aggregates and monomers were denatured and resolved by SDS-PAGE after PDS treatment. Nearly complete proteolysis was observed in PDS samples lacking PIC, whereas monomeric Aβ₄₀ [Fig. 2(C)] or an equilibrium mixture of monomeric (M) and tetrameric (T) Aβ₄₂ structures as well as SDS-soluble aggregates (A_sol) [Fig. 2(D)] was observed in the presence of PIC. Collectively, these experiments demonstrate that disaggregation can occur upon inhibition of proteolysis.

To more accurately quantify the different Aβ₄₀ species, especially aggregates, and to avoid possible under-representation of SDS-insoluble Aβ₄₀ that might not enter the gel or that might fail to be transferred, we used membrane-based filter retardation assays and dot blots as complementary assays._. Dot blots without added SDS were used to quantify total Aβ₄₀, including monomeric, oligomeric, and aggregated species. In parallel, Aβ samples were denatured by boiling in 2% SDS and filtered through a cellulose acetate membrane (0.2 μm), where only large SDS-insoluble Aβ aggregates are retained on the membrane [Fig. 2(E)].5 Aβ₄₀ retained on the membrane was detected using the 6E10 antibody and quantified from triplicate samples. Our results indicate that disaggregation by worm PDS reduced SDS-insoluble Aβ₄₀ aggregates by 70–80% [Fig. 2(E,F)]. However, proteolysis of Aβ₄₀ was largely prevented by the addition of PIC or phenylmethylsulphonyl fluoride (PMSF) [Fig. 2(E,F)]. Fig. 2(E) also demonstrates that heat inactivation of the PDS-associated disaggregase activity (95°C, 10 min) prevents disaggregation and proteolysis, as reported previously.22

The disaggregation activity is more temperature resistant than the protease activity

To evaluate the sensitivity of the disaggregation activity to thermal denaturation, we incubated PDS at 37, 70, 80, or 95°C for 10 min before its addition to fibrillar Aβ₄₀. Only a moderate decline in disaggregation activity is observed upon treatment at 70 or 80°C [Fig. 3(A), cf. blue and green lines to black line], whereas heating to 95°C substantially diminished disaggregation activity (orange line).

Figure 3. Disaggregation activity is heat resistant A: Heat-treated C. elegans PDS (25 μg/mL) incubated with fibrillar Aβ₄₀ in the absence of protease inhibitor. PDS was heated to 37, 70, 80 or 95°C for 10 min before incubation with Aβ₄₀ fibrils. B: Dot blot quantification of total Aβ₄₀ using mAB 6E10 after 72 h incubation. C: Aβ₄₀ SDS-resistant aggregate quantification after 72 h incubation (filter retardation, mAB 6E10). Bar graphs show normalized averaged dot blot signals of three independent experiments, error bars indicate standard deviations. Proteolytic activities are inactivated at 70°C, whereas 95°C is needed to inactivate the disaggregation activity D: Fibrillar Aβ₄₂ was incubated with heat-treated C. elegans PDS as in A in the presence or absence of PIC. Disaggregation products were analyzed by Western blotting as in Fig. 2(D). SDS-insoluble aggregate bands (A) and 20 kDa tetramer bands (T) were analyzed quantitatively. Bar graphs show intensities of aggregate (white) and soluble (gray) Aβ₄₂ peptide bands relative to the amounts of Aβ before incubation (control) from four independent experiments, * denotes P < 0.05; ** denotes P < 0.01. E: C. elegans PDS was pretreated without (blue) or with proteinase K (100 μg/mL) (red) for 1 h at 37°C before inactivation of the proteinase K activity by heating to 80°C for 10 min (conditions that leave the disaggregase intact) and then added to fibrillar Aβ₄₀ (15 μM; Aβ monomer).

The relative amounts of total and SDS-resistant Aβ₄₀ aggregates [Fig. 3(B,C), respectively] were quantified by dot blot and filter retardation assays and normalized to the untreated fibrillar Aβ control. Heat treatment at 80°C inactivated proteolysis [Fig. 3(B)] but not the disaggregation activity, as evidenced by the fact that the amount of SDS-resistant aggregates were significantly reduced [Fig. 3(C)] when compared with PDS heat treated at 95°C. Analogous results were independently obtained for Aβ₄₂ [Fig. 3(D)] by quantifying the soluble (M, T) and aggregate (A) bands from Aβ₄₂ Western blots after disaggregation [cf. Fig. 2(D)] as a complementary assay.

The higher heat resistance of the disaggregation activity in comparison to the proteolytic activity enabled us to test whether the disaggregation activity itself can be compromised by proteolysis. Wild-type worm PDS was pretreated without or with proteinase K (100 μg/mL) for 1 h at 37°C before inactivation of the proteinase K activity by heating to 80°C for 10 min (conditions that leave the disaggregase intact). The capacity to disassemble ThT-binding Aβ40 fibrils was greatly reduced in the protease- and heat-treated PDS when compared with PDS that had been heat treated but had not been subjected to proteinase K treatment [Fig. 3(E)], demonstrating that a protein(s) is largely responsible for the disaggregase activity(ies) in C. elegans.

Disaggregation is reversible in the absence of proteolysis upon exhaustion of the disaggregase activity

If the disaggregation activity is enabled by a pathway or a specific protein or protein complex that is separable from proteolysis, Aβ monomers released by the disaggregase activity could reaggregate when proteases are inhibited, provided that the disaggregase activity is significantly reduced or extinguished. Spontaneous reaggregation of Aβ₄₀ after 25 h was indeed observed [Fig. 4(A)] in the presence of the PIC (blue) and PMSF (red). AFM imaging of the Aβ₄₀ samples after 75 h of incubation with Aβ worm PDS in the presence of PIC or PMSF confirmed the re-emergence of fibrillar aggregates. Compare Fig. 4(B) with Fig. 2(A), bottom right, Aβ₄₀ with PDS in the absence of PIC.

Figure 4. Disaggregation is reversible in the absence of proteolysis. A: C. elegans PDS was incubated with Aβ₄₀ fibrils for 75 h (as in Fig. 1) in the absence (orange) or presence of PIC (blue) or PMSF (red). Inhibition of proteolysis did not prevent disaggregation, but led to reaggregation of Aβ₄₀ upon prolonged incubation. No disaggregation was observed in the presence of heat-inactivated (95°C) PDS in the presence of PIC (green) or in buffer controls (black). B: Atomic force microscopy image of Aβ₄₀ fibrils after 75 h incubation with PDS in the presence of PIC or PMSF, showing re-emergence of fibrils. Scale bars = 200 nm, height scale = 20 nm. C: Aβ₄₀ peptide (15 μM) was separated into soluble and insoluble fractions by centrifugation at 200,000 × g before and after aggregation and after incubation with PDS (25 μg/mL) in the presence or absence of PIC. Both the pellet and the supernatant were treated with HFIP/TFA to denature Aβ₄₀ allowing it to be quantified by HPLC analysis. D: Addition of untreated PDS (5% v/v) to soluble Aβ₄₀ (15 μM) [buffer and incubation as in Fig. 1(A)] prevents aggregation of Aβ into ThT-binding aggregates (triplicate samples, green lines), whereas heat-inactivated PDS (95°C, triplicate samples, red lines), and PDS in the presence of PIC (triplicate samples, blue lines) ultimately allowed Aβ₄₀ aggregation, apparently after the consumption of the disaggregase activity.

To verify the mass balance and the aggregation state of the Aβ peptide during the disaggregation and reaggregation processes, the amounts of aggregated and monomeric Aβ₄₀ peptide were quantified by HPLC analysis. The Aβ₄₀ peptide (15 μM) was incubated in PBS at 37°C for 75 h with rotary agitation (20 rpm) to form fibrillar aggregates and then incubated for a further 24 or 75 h in the presence of PDS (1%), either in the presence or absence of PIC. Aliquots of the peptide were ultracentrifuged (200,000 × g, 20 min) at 24 or 75 h, supernatants were removed and pellets were solubilized in HFIP/TFA overnight. Supernatant and pellet fractions were then analyzed by reverse phase HPLC and quantified by UV absorption at 215 nm. Peptide fractions were quantified relative to the initial amount of soluble Aβ₄₀ peptide [Fig. 4(C)]. While more than 95% of the Aβ peptide was found in the pellet fraction after aggregation, incubation with PDS for 24 h reduced the quantity of Aβ to ∼40% (no PIC). While most of the soluble Aβ was degraded by proteolysis in the absence of PIC, addition of PIC largely prevented proteolysis of soluble Aβ. Interestingly, ∼80% of Aβ was again found in the insoluble fraction after prolonged (75 h) incubation with PDS and PIC, consistent with reaggregation.

To further verify that proteolysis prevents reaggregation of Aβ₄₀ after incubation with worm PDS, heat-inactivated PDS (5% v/v) was added to freshly monomerized Aβ₄₀ (15 μM) and aggregation was monitored by ThT fluorescence [Fig. 4(D)]. The Aβ₄₀ peptide aggregated [Fig. 4(D), red lines] in the presence of heat-inactivated PDS (10 min, 95°C) or PDS with added protease inhibitor cocktail [PIC, Fig. 4(D), blue lines], on a similar time scale as in the absence of PDS (Supporting information Fig. S1A). In contrast, proteolytically active PDS largely prevented aggregation (green lines). The ThT signal of PDS incubated without Aβ₄₀ did not increase with time (orange lines).

Taken together, our results demonstrate that disaggregation and proteolysis are distinct C. elegans PDS activities that can be dissociated and monitored separately.

Discussion

The C. elegans Aβ disaggregation activity appears to be a primary mechanism for the detoxification of Aβ aggregates in the worm AD model.22 The disaggregase activity is partially upregulated by activation of the HSF-1 transcription factor, which is negatively regulated by the insulin growth factor-1 receptor signaling pathway that also influences longevity and youthfulness. The more detailed characterization of the disaggregase activity outlined herein provides a means to systematically fractionate the cellular components and/or the underlying molecular determinants that disaggregate and degrade Aβ fibrils and perhaps oligomeric aggregates. Herein we have demonstrated that disaggregation and proteolysis are separable cellular activities, both critical for the clearance and detoxification of Aβ aggregates. Both activities may be part of the same macromolecular complex or pathway, yet more experiments are required to identify their origins. The widely available genetic tools and the ease with which gene expression within the nematode C. elegans can be manipulated by RNAi enables candidate genes of the Aβ disaggregase activity to be evaluated in this organismal model.

Disaggregation and proteolysis are separate cellular activities

Three unrelated results indicate that disaggregation and proteolysis are separable activities. First, the Roche protease inhibitor cocktail (PIC) and the protease inhibitor PMSF do not affect disaggregation, but largely abolish proteolysis. Second, the disaggregation activity is more resistant to heat inactivation than is the proteolysis activity, as an 80°C pretreatment only moderately slows disaggregation, but largely abolishes proteolysis of Aβ. Third, our data show that spontaneous reaggregation occurs after nearly complete disaggregation when proteolysis is inhibited. Taken together, these results clearly indicate that these two cellular activities are separable, and that the disaggregase activity is protein based, as assessed by the ability of proteinase K to disable the disaggregase activity. Furthermore, our data suggest that proteolysis of amyloidogenic peptides after disaggregation may be required for the detoxification of amyloid aggregates to prevent their reaggregation into toxic structures.

Candidates to mediate disaggregation

The cellular components or pathway(s) that play a role in the C.elegans disaggregation/proteolysis activities are currently unknown. The thermal denaturation resistance of the disaggregase activity is high when compared with the protease activity. Nevertheless, the disaggregase activity can be inactivated by proteolysis, which suggests that it is mediated by a stable protein, protein complex, or pathway, possibly involving membranes or vesicle association. This hypothesis is supported by the finding that detergents compromise the disaggregation activity even at concentrations below those which would denature proteins.

In yeast, Hsp104/ClpB, a AAA ATPase family member, is known to mediate disaggregation33, 34 with a dependence on Hsp70.35 Owing to the lack of an obvious Hsp104 homolog in mammals, the identity of the disaggregation-mediating protein, protein complex or pathway in mammals remains unknown, and is the subject of ongoing investigations.

The disaggregation assay described herein provides the means to identify the protein(s) and/or pathways responsible for disaggregation and/or proteolysis of Aβ fibrils by analyzing worm homogenate fractionated by chromatographic and related strategies. Using PDS from worms also enables investigation of candidate genes using RNAi approaches that are straightforward to apply in C. elegans proteotoxicity models. The disaggregase/proteolysis assay reported herein also enables small molecule screens to be conducted to discover compounds that enhance or inhibit these critical activities. Disaggregase enhancers may be useful for ameliorating aging-associated intracellular Aβ aggregation and proteotoxicity thought to cause neurodegeneration in Alzheimer's disease.

MATERIALS AND METHODS

Materials

Protein concentration was determined using a BCA kit (Pierce #23223). Complete protease inhibitor cocktail (PIC, #1836170) was purchased from Roche (Basel, Switzerland). Synthetic Aβ₄₀ and Aβ₄₂ peptides were purchased from Synpep (Dublin, CA). Monoclonal anti-Aβ antibodies 4G8 and 6E10 were purchased from Signet (Dedham, MA). All other materials were purchased from Sigma.

C. elegans

CL2006 and control (wt) worms (N2)21 were obtained from the Caenorhabditis Genetics Center (Minneapolis, MN). The worms were grown at 20°C as previously described.22

In vitro Aβ disaggregation assay

C. elegans worms grown to day one of adulthood were washed twice with M9 buffer (Sigma) and once more with PBS at room temperature. The worms were then resuspended in 300 μL ice cold PBS, transferred to a tissue grinder (885482, Kontes, Vineland, NJ) and homogenized. Crude homogenates were centrifuged in a desktop microfuge (Eppendorf 5810R, 3000 rpm, 3 min) to prepare post debris supernatants (PDS). PDS were transferred to new tubes and total protein concentrations were measured with a BCA kit (Pierce, Rockford, IL).

A stock solution of monomeric Aβ₄₀ peptide (200 μM) was prepared as described.29 Aβ₄₀ peptide (50 μM) was aggregated in phosphate buffer (300 mM NaCl, 50 mM Na-phosphate, pH 7.4) at 37°C in a 1.5 mL reaction tube under constant agitation using a rocking platform (20 cycles/min) (Fig. 1) or an overhead shaker (20 rpm, Figs. 2–4) for 4 days. Aβ₄₀ fibrils were sonicated for 30 min in a water bath sonicator (FS60, Fisher Scientific, Pittsburg, PA) and characterized by far UV CD spectroscopy and atomic force microscopy (Fig. S1). Aβ₄₀ peptide was then diluted to a final concentration of 15 μM Aβ and 150 mM NaCl in phosphate buffer (50 mM Na-phosphate, pH 7.4) containing ThT (20 μM) and PDS worm homogenate (25 μg/mL) with or without protease inhibitors as indicated. Synthetic Aβ₄₂ was pretreated by an analogous procedure, but using a 30 kDa cut-off filter for the second filtration step during monomerization. All final Aβ concentrations were 15 μM unless indicated otherwise.

Three aliquots (100 μL) of each sample were incubated at 37°C in low-binding, 96-well, clear-bottomed plates (Corning). Samples were agitated for 5 s before each reading and ThT fluorescence was measured using a Gemini SpectraMax EM fluorescence plate reader (Molecular Devices, Sunnyvale, CA) every 10 min. Fluorescence backgrounds from PDS and buffer were subtracted from the ThT signal. Fluorescence data were fit to single exponential decay functions with a time constant τ and a linear term allowing for a sloped postdisaggregation baseline: F (t) = mt + b + A exp(-t/τ). τ-values are reported as averages +/− standard deviations determined from at least three independent samples. ThT signals were normalized by setting the fluorescence at t = 0 h to one.

Roche complete protease inhibitor cocktail (PIC) was diluted from aqueous stock solutions (25 x) to final concentrations of 1 x dilution.

ATP hydrolysis

To hydrolyze ATP, PDS samples were preincubated for 15 min at 30°C with apyrase (New England Biolabs) at a final concentration of 0.1 or 0.5 U before adding to the Aβ aggregates. Under these conditions, 0.25 U apyrase were sufficient to completely hydrolyze the ATP present in the PDS as tested in a luciferase assay (data not shown). Aβ aggregates, PIC and azide (0.02%) were added after hydrolysis as indicated.

Filter retardation assays and dot blotting

Equal volumes of denaturation buffer (Tris buffered saline (TBS) pH 8, 4% SDS) were added to Aβ samples (20 μL). Samples were boiled for 5 min and filtered through a cellulose acetate membrane filter (Bio-Rad) using a 96-well vacuum apparatus (Bio-Rad) as described.5 For dot blots, Aβ samples (20 μL) were mixed with equal volumes of TBS and filtered through a nitrocellulose membrane filter (Bio-Rad) using the same vacuum apparatus.

Western blotting

SDS-PAGE was performed using 16.5% Tris-Tricine buffered acrylamide gels and the peptides were then blotted onto a PVDF membrane. Western blots, dot blots and filter retardation blots were probed using 6E10 (1:2,000) and 4G8 (1:4,000) antibodies and developed using an ECL system. For reprobing, PVDF membranes were stripped by incubation in 300 mM NaOH (5 min, RT), followed by neutralization by several rinses in TBST (10 mM Tris-HCl pH 7.5, 150 mM NaCl, 0.3% Tween-20). Aβ monoclonal antibodies clone 4G8 (#9220) and clone 6E10 (#9320) were purchased from Signet (Dedham, MA), secondary antibodies from Pierce.

Atomic force microscopy

Aliquots (20 μL) were removed from the Aβ₄₀ samples and placed on freshly cleaved mica (1 × 1 cm) and mounted onto a metal or glass sample holder. AFM samples were prepared and imaged as described22 using a Nanoscope III (Veeco, Santa Barbara, CA) or Nanowizard II AFM setup (JPK, Berlin, Germany).

HPLC quantification

Aβ peptide aliquots (100 μL, 15 μM) were centrifuged (200,000 × g 20 min), supernatants were removed and stored at 4°C and pellets were solubilized overnight in 50μL of hexafluoroisopropanol/trifluoroacetic acid (HFIP/TFA 1:1, v/v) after sonicating for 5 min. Water (50 μL) was added to the HFIP samples. Supernatant and pellet fractions were then analyzed by reverse phase HPLC (μRPC C2/C18SC 2.1/10, GE Healthcare) in a water (0.1% TFA)/acetonitrile (0.1% TFA) gradient and quantified by UV absorption at 215 nm. UV signals were normalized to the equivalent amount of soluble Aβ₄₀ peptide.

Acknowledgements

The authors thank S. Kostka and G. Grelle for expert technical support.