Protease-activated alpha-2-macroglobulin can inhibit amyloid formation via two distinct mechanisms

Highlights ► α2M is an extracellular chaperone able to inhibit protein aggregation. ► Protease–α2M complexes can degrade amyloidogenic substrates or act as a chaperone. ► Activated α2M may play an important role in preventing protein deposition in vivo.

a b s t r a c t a 2 -Macroglobulin (a 2 M) is an extracellular chaperone that inhibits amorphous and fibrillar protein aggregation. The reaction of a 2 M with proteases results in an 'activated' conformation, where the proteases become covalently-linked within the interior of a cage-like structure formed by a 2 M. This study investigates, the effect of activation on the ability of a 2 M to inhibit amyloid formation by Ab  and I59T human lysozyme and shows that protease-activated a 2 M can act via two distinct mechanisms: (i) by trapping proteases that remain able to degrade polypeptide chains and (ii) by a chaperone action that prevents misfolded clients from continuing along the amyloid forming pathway. a 2 M is best known for its ability to trap a broad range of proteases within a cage-like quaternary structure via covalent-linkage of the protease to intramolecular thioester bonds on a 2 M [1]. This reaction results in a conformationally altered form commonly known as ''activated'' or ''fast'' a 2 M, the latter term relating to enhanced mobility via native gel electrophoresis. Activation of a 2 M results in the exposure of a cryptic receptor recognition site for the lowdensity lipoprotein receptor-related protein (LRP) [1]. In addition to proteases, small nucleophiles can activate a 2 M by interacting directly with its thioester bonds [2].
Along with protease trapping, many other biological functions have been proposed for a 2 M; including roles in immunomodulation, cancer progression and extracellular proteostasis [3][4][5]. a 2 M can bind to a range of endogenous disease-associated proteins including the amyloid b peptide (Ab 1-42 ) [6], prion proteins [7] and b 2 -microglobulin [8], which are the main components of deposits found in Alzheimer's disease (AD), spongiform encephalopathies and dialysis-related amyloidosis, respectively [9]. Moreover, a 2 M is found to be co-localized in vivo with amyloid deposits in AD and the spongiform encephalopathies [7,10]. Recent work has shown that native a 2 M can act as an ATP-independent molecular chaperone by suppressing stress-induced amorphous protein aggregation [5]. The mechanism by which this occurs appears to involve the formation of stable, soluble complexes between a 2 M and the misfolded client proteins [5]. Native a 2 M has also been shown to suppress the fibril formation of a range of amyloidogenic proteins and peptides [11,12]. It has been proposed that a 2 M can protect against pathogenic misfolded proteins by promoting their removal from the extracellular space [6,13,14]. However, trypsin-activated a 2 M (trypsin-a 2 M) is reportedly unable to prevent the amorphous aggregation, in vitro, of some proteins [5].
Nevertheless, after binding to misfolded proteins, a 2 M retains the ability to become activated, and a 2 M-trypsin-misfolded protein complexes are recognized by LRP [5], representing a potential route for the targeted disposal of misfolded proteins in vivo. Activated a 2 M can protect cells from Ab toxicity in vitro through specific binding and subsequent LRP mediated uptake and degradation of Ab 1À40 [6,10,15]. While it is clear that activated a 2 M can bind to Ab peptide, its ability to prevent the fibrillar aggregation of amyloid forming peptides or proteins has not been tested. To address this issue, we investigate the effect of activated a 2 M on the fibril formation of the amyloidogenic Ab 1-42 peptide and of a non-natural variant of human lysozyme (I59T) that possesses many attributes associated with the natural amyloidogenic variants linked to systemic amyloidosis [16].

Materials and methods
Chemicals and reagents were purchased from Sigma-Aldrich Ltd. unless otherwise stated.

Proteins and peptides
a 2 M was purified from human plasma by zinc chelate affinity chromatography and size exclusion chromatography (SEC) as previously described [5]. Purified a 2 M was stored at 4°C (for less than 2 months) and routinely examined prior to use by native polyacrylamide gel electrophoresis (PAGE) to ensure that the preparation had not become partially degraded, activated or cross-linked, modifications that can occur with prolonged storage [4,17,18]. Ab  was purchased from Biopeptide Co. Inc. or Bachem AG. Solutions of Ab 1-42 peptide were prepared by a TFA/HFIP dissolution method [19]. The non-natural variant of human lysozyme, I59T, was expressed and purified as previously described [16]. Tris-acetate gels with Tris-glycine native running buffer (Life Technologies Ltd.). The reaction was allowed to continue for up to an additional 45 min to ensure completion. Unreacted trypsin was removed by SEC and SDS-PAGE analysis using NuPAGE Novex 4-12% Bis-Tris gels with MES running buffer (Life Technologies Ltd.) confirmed that no cleavage outside the bait region had occurred. To produce enzymatically inactivated trypsin-a 2 M (i.e. (i)trypsin-a 2 M), trypsin-a 2 M was incubated (2 h, 25°C) with excess Complete™ protease inhibitor cocktail (Roche Diagnostics Ltd.) and samples were desalted using Zeba™ desalting columns (Thermo Fisher Scientific). Ammonium chloride (NH 4 Cl) activation was performed by incubating a 2 M with 400 mM NH 4 Cl in PBS (14 h, 25°C) and subsequently desalting as described. was incubated with stirring at 60°C in a Cary Eclipse spectrofluorimeter (Agilent Ltd.) and ThT fluorescence intensity was monitored with excitation and emission wavelengths of 440 nm and 480 nm (slit-widths 5 nm). All samples incubated with native a 2 M, trypsin-a 2 M, (i)trypsin-a 2 M, or NH 4 Cl-activated a 2 M contained a molar ratio of substrate-to-a 2 M of 10:1, based on the molecular weights of the a 2 M tetramer (720 kDa), the Ab 1-42 monomer (4.5 kDa) or the I59T monomer (14.7 kDa). All experiments were performed in triplicate.

Transmission electron microscopy (TEM)
Fibril solutions (5 ll) were applied to carbon-coated nickel grids, stained with 2% (w/v) uranyl acetate, and imaged on a FEI Tecnai G 2 transmission electron microscope (Multi-Imaging Unit in the Department of Physiology, Development and Neuroscience, University of Cambridge, UK). Images were analyzed using the SIS Megaview II Image Capture system (Olympus).

Results
Native a 2 M has previously been shown to inhibit the amorphous and fibrillar aggregation of a range of proteins by increasing their solubility [5,11,12,16]. To determine if activated a 2 M can also prevent amyloid formation, we compared the effect of native a 2 M and trypsin-a 2 M on the fibril formation of I59T lysozyme and the amyloidogenic peptide Ab 1-42 . Previously reported conditions for generating trypsin-a 2 M vary greatly [2,5,20]; therefore, in this study we used an optimized method to obtain preparations of trypsin-a 2 M that were completely activated but not degraded ( Supplementary Fig. 1). The aggregation behavior of I59T lysozyme is well established and this system has been used to study the effects on fibril formation of the extracellular chaperones clusterin, haptoglobin and native a 2 M [12,21]. In this study, the kinetics of aggregation show a lag phase of ca. 50 min, followed by a rapid growth phase that reaches a plateau after ca. 150 min (Fig. 1a, black line). a 2 M, present at a molar ratio of 10:1 (lysozyme-toa 2 M), results in a dramatic decrease in thioflavin-T (ThT) fluorescence over the course of the assay (Fig. 1a; red line). When trypsin-a 2 M is incubated with I59T lysozyme the ThT fluorescence is again, significantly suppressed ( Fig. 1a; blue line). At the endpoint of the fibril formation, the presence of both native a 2 M and trypsin-a 2 M results in over a 90% decrease in ThT signal relative to the I59T lysozyme sample alone (Fig. 1b).
TEM images of the ThT assay endpoint samples demonstrate that while I59T lysozyme alone forms fibrillar structures there is no evidence for such structures when I59T lysozyme is incubated under the same conditions with native a 2 M or trypsin-a 2 M (Fig. 1c). SDS-PAGE analysis of the endpoint supernatants reveals that no detectable I59T lysozyme remains in solution when incubated alone, whereas in the presence of native a 2 M, a large majority (>90%) of lysozyme remains soluble (Fig. 1d). The I59T lysozyme also remains in the soluble fraction when incubated with trypsina 2 M and shows no evidence of proteolytic degradation (Fig. 1d).
Conversely, the pellet fractions (solubilized with 8 M urea), shows a large proportion of I59T lysozyme in the I59T alone sample (Fig. 1e, lane 1p) and only trace amounts (less than 10%) of lysozyme present in samples incubated with native a 2 M and trypsina 2 M (Fig. 1e, lane 2p and 3p). This finding is consistent with the fraction of the maximum ThT signal observed at the aggregation endpoints (Fig. 1b). In separate experiments, incubation of monomeric I59T lysozyme with trypsin or trypsin-a 2 M does not result in the appearance of any degraded protein in the soluble fractions after 120 min of incubation under the aggregation conditions used ( Supplementary Fig. 2a), in addition, trypsin alone has no effect on I59T fibril formation (Supplementary Fig. 2b). However, it is noted that small quantities of protein fragments (less than 5% total protein) are apparent in the SDS-PAGE analysis of the pellet samples after 300 min incubation. These fragments may be the result of residual trypsin-a 2 M activity, but they appear to be aggregation prone as they are only apparent in small quantities in the insoluble pellet sample. Taken together these results reveal that, native a 2 M and trypsin-a 2 M are able to suppress I59T fibril formation predominantly via chaperone action.
We next evaluated whether trypsin-a 2 M could also suppress Ab 1-42 fibril formation. Under the conditions used here, aggregation of Ab 1-42 shows a lag of ca. 70 min, followed by a rapid growth phase and a plateau at ca. 150 min (Fig. 2a, black line). Consistent with previous studies [12], the presence of native a 2 M at a 10:1 (Ab 1-42 -to-a 2 M) molar ratio dramatically reduces the time-dependent increase in ThT fluorescence ( Fig. 1a; red line). At the same molar ratio, the presence of trypsin-a 2 M also results in a suppression of ThT fluorescence (Fig. 2a, solid blue line). This suppression in ThT signal is over 80% for both the presence of native a 2 M and trypsin-a 2 M at the endpoint of the assay (Fig. 2b). In all samples containing Ab  there is a small ThT fluorescence signal at the start of the assay, likely to be due to some ThT positive aggregates being present in the stock peptide solutions. This level remains constant over the time course for the samples containing native a 2 M, but decreases slightly in the presence of trypsin-a 2 M. We suspect that this may be due to the ability of trypsin-a 2 M to degrade these ThT positive species.
TEM images of the ThT assay endpoint samples show that fibrillar aggregates are formed by Ab 1-42 incubated alone; however, in the presence of either native a 2 M or trypsin-a 2 M, the number of well-defined fibrils is reduced and most aggregates appear to be amorphous (Fig. 2c). Analysis of the endpoint supernatants by SDS-PAGE reveals that incubation of Ab 1-42 with trypsin-a 2 M results in proteolysis of the peptide (Fig. 2d, lane 3). This result is consistent with previous work showing that a 2 M-trapped proteases remain active against small substrates including Ab 1-42 [22]. Therefore, it appears that trypsin-a 2 M prevents Ab 1-42 fibril formation, under these conditions at least partly via degradation of the Ab 1-42 peptide to form smaller species that remain soluble. Interestingly, this mechanism for inhibiting fibril formation may not be restricted to just Ab  . We have also observed that trypsin-a 2 M can suppress the fibril formation of reduced and carboxymethylated a-lactalbumin by a process which involves proteolysis of the full-length protein ( Supplementary Fig. 3).
Given that trypsin-a 2 M can degrade polypeptides which can enter the activated a 2 M cage, it is necessary to inactivate the bound trypsin to examine, in isolation, whether trypsin-a 2 M possesses chaperone activity similar to native a 2 M. Trypsin-a 2 M and trypsin-a 2 M after treatment with a small molecule protease inhibitor ((i)trypsin-a 2 M) migrate similarly when analyzed by native PAGE, suggesting that protease inactivation does not grossly affect the structure of the covalent complex ( Supplementary Fig. 1). Incubation of fluorophore-labelled Ab 1-42 with trypsin-a 2 M shows that pre-treatment of the latter with protease inhibitors prevents detectable proteolysis of Ab 1-42 (Fig. 2e), however, the (i)trypsina 2 M retains the ability to inhibit Ab 1-42 aggregation (Fig. 2a,

Discussion
In the work presented here, we show that activated a 2 M, despite a large conformational change upon activation, retains the ability to suppress fibril formation. From earlier work, it is clear that a 2 M has distinct binding sites for proteases and misfolded proteins as the binding of a misfolded client protein does not prevent protease trapping [6]. In the current study, we demonstrate that the presence of a bound protease, regardless of whether or not the protease is pharmacologically inhibited, does not significantly reduce chaperone activity of a 2 M. Moreover, a 2 M remains an active chaperone after direct activation using small molecules.
In vivo, activated a 2 M is rapidly cleared from circulation [2] and typically represents only 0.17-0.7% of the total a 2 M in blood plasma of adults [23]. The activated a 2 M plasma concentration is, however, increased in many disease states including pancreatitis, multiple sclerosis and sepsis [23][24][25]. Moreover, the onset of some diseases, such as periodontitis, diabetic retinopathy and inflammatory joint disease results in increased activated a 2 M levels in other extracellular fluids [26][27][28]. Although enhanced concentrations of activated a 2 M have been largely attributed to increased protease trapping, it has been reported that interaction with proteases only partially accounts for the total activated a 2 M present in synovial fluid [28]; higher levels of both protease-activated and amine-activated a 2 M may therefore be significant for facilitating clearance of aberrant clients via LRP. Interestingly, aggregates of Ab 1-40 and amylin have been shown to activate the plasmin protease system [29]. Thus it is possible that concentrations of plasmin-activated a 2 M may also be increased in response to the accumulation of misfolded proteins.
In conclusion, we provide evidence that protease-activated a 2 M has two distinct mechanisms for inhibiting amyloid formation: (i) via protease-a 2 M-mediated degradation of amyloidogenic substrates and (ii) by a chaperone action that prevents misfolded clients from continuing along the amyloid forming pathway. In the absence of proteases, activated a 2 M is able to inhibit fibril formation via the latter function only. It is tempting to speculate that the chaperone activity of protease-activated a 2 M may target misfolded proteins to the trapped protease, thereby providing a specific mechanism for degradation of amyloidogenic proteins in extracellular fluids. Clearly, further studies are required to substantiate this proposition; however, a greater understanding of the mechanisms by which a 2 M is able to prevent protein aggregation and facilitate the disposal of misfolded peptide and protein molecules could, in future, provide potential therapeutic targets for amyloidosis.