Native aggregation is a common feature among triosephosphate isomerases of different species

Triosephosphate isomerase (TIM) is an enzyme of the glycolysis pathway which exists in almost all types of cells. Its structure is the prototype of a motif called TIM-barrel or (α/β)8 barrel, which is the most common fold of all known enzyme structures. The simplest form in which TIM is catalytically active is a homodimer, in many species of bacteria and eukaryotes, or a homotetramer in some archaea. Here we show that the purified homodimeric TIMs from nine different species of eukaryotes and one of an extremophile bacterium spontaneously form higher order aggregates that can range from 3 to 21 dimers per macromolecular complex. We analysed these aggregates with clear native electrophoresis with normal and inverse polarity, blue native polyacrylamide gel electrophoresis, liquid chromatography, dynamic light scattering, thermal shift assay and transmission electron and fluorescence microscopies, we also performed bioinformatic analysis of the sequences of all enzymes to identify and predict regions that are prone to aggregation. Additionally, the capacity of TIM from Trypanosoma brucei to form fibrillar aggregates was characterized. Our results indicate that all the TIMs we studied are capable of forming oligomers of different sizes. This is significant because aggregation of TIM may be important in some of its non-catalytic moonlighting functions, like being a potent food allergen, or in its role associated with Alzheimer’s disease.


Results and Discussion
Expression and purification of TIMs from ten different species. In order to characterize the oligomeric profile of the native aggregates of this enzyme a highly purified protein is required. The heterologous expression and purification of the TIMs from ten different species was carried out as mentioned in material and methods (sections 3.1, 3.2 and 3.3) Fig. 1 shows the SDS-PAGE electrophoretic pattern of the purified enzymes. All TIMs showed a single polypeptide band and only an additional low molecular mass band could be observed in the sample of Plasmodium falciparum TIM (PfTIM) (Fig. 1, lane 6). As shown in Table 1 the specific activity and the kinetic parameters for nine TIMs are in the range reported previously for each of them [18][19][20][21][22][23][24] . In the case of Thermus thermophilus TIM (TtTIM), to our knowledge, no in vitro activity has been reported previously. The activity of PfTIM is the lowest for all the TIMs studied, but the Kcat/km value is similar to the ones determined for the rest of the enzymes (Table 1). tiMs from several species are able to form native active oligomers. Previously, Swamy et al. 17 detected high molecular weight aggregates (1320 kDa) from rabbit TIM using blue native electrophoresis (BNE). To know if this aggregation process is common for TIM, we analysed samples purified from different species in this native system. We found that TIMs from several species form high molecular weight aggregates that vary in number and molecular masses. Nevertheless, the resolution of the aggregation pattern in this type of gels was not optimal (Fig. S1, Suppl. Information). The TIMs from four species (Giardia lamblia TIM: GlTIM, Trypanosoma brucei TIM: TbTIM, Trypanosoma cruzi TIM: TcTIM and Leishmania mexicana TIM: LmTIM) did not enter the gel, even though the presence of Coomassie blue in this system, contributes with negative charges that favour protein migration 25 . The theoretical isoelectric point (Ip) indicates that TbTIM, TcTIM and LmTIM possess the most basic values, while GlTIM has an Ip near to the pH of the buffer used in the system (pH = 7.0) ( Table 1). This suggests that the migration of the TIMs in BNE is highly dependent on the charge/mass ratio of the sample. To further investigate this effect, a 7 × 7 cm 6% acrylamide horizontal gel (0.5 cm width) was prepared for the analysis of the samples using native electrophoresis. Figure 2 shows that the migration of the Saccharomyces cerevisiae TIM: ScTIM, PfTIM, Entamoeba histolytica TIM: EhTIM, Homo sapiens TIM: HsTIM, Taenia solium TIM: TsTIM and TtTIM, occurs in the cathode-anode direction (normal polarity), but TbTIM, TcTIM and LmTIM migrate in the anode-cathode direction (inverse polarity). In contrast, GlTIM is not able to migrate at all, probably because of its small superficial charge at pH 7.0. Because the protein-dye interaction of the Coomassie blue used in BNE may shift the apparent molecular mass of the aggregates 26 , we decided to analyse the samples using clear native electrophoresis (CN-PAGE), using different polarities.
A modification of the HEPES-imidazole system, used by McLellan 27 , allowed the resolution of aggregates from ScTIM, PfTIM, EhTIM, HsTIM, TsTIM and TtTIM, along with the molecular mass markers used in normal polarity: Type I ferritin from horse spleen (SIGMA) and bovine serum albumin (SIGMA, heat shock fraction) ( Fig. 3A left panel). The aggregates of GlTIM, TbTIM, TcTIM and LmTIM were resolved with inverse polarity (Fig. 3A right panel). Both systems, with normal and inverted polarity, were buffered at pH 7.0. To improve the definition of the bands, we varied the concentration of HEPES and imidazole in the gels, keeping the same proportion of both salts. The resolution of the low molecular mass bands increased with increasing concentration of the salts in the gel (Fig. 3B,C). In all cases, the anode and cathode buffers used in the electrophoresis always remained the same (40 mM HEPES and 15.3 mM imidazole).
All the TIMs tested produced oligomers larger than a dimer, although estimating the size of the aggregates was difficult. We identified 2 to 7 TIM aggregates with high molecular masses in the normal polarity system. The two molecular mass markers used, albumin and ferritin, form oligomers of known size, with masses ranging from 66 to 1320 kDa 28 . The molecular masses and the relative migration of these markers and their aggregates was plotted for each gel and all yielded a linear correlation, even at different salt concentrations (mean R 2 = 0.986) (Figs S2-S4, Suppl. Information). These linear regressions were based on the relative migration of the markers and the molecular masses previously reported for their aggregates. The molecular masses of the aggregates of the TIMs in each gel condition were estimated from these linear regressions.   distribution in the profile of the aggregates was observed with inverted polarity electrophoresis, where LmTIM, TcTIM and TbTIM form 4, 3 and 3 bands of aggregates, respectively ( Fig. 3 right panels). We were unable to calculate the apparent molecular mass of these aggregates, since the markers (ferritin and albumin) do not penetrate the gel when subjected to an inverted polarity. In the case of GlTIM, we could not discern if the absence of migration is due to the small mobility of the high molecular mass aggregates or to the low overall surface charge of the protein at pH 7.0. To discard possible non-specific, hydrophobic-driven associations due to the exposure of non-polar residues that are normally-hidden, we tested the stability of TIM aggregates from three species using three non-ionic detergents: n-dodecyl-β-D-maltoside, digitonin and Triton X-100. These detergents are used to extract large membrane-embedded complexes, without destroying their native associations [29][30][31][32] . Figure 4A shows the stability of the aggregates of HsTIM, EhTIM and PfTIM in the three detergents, which were used in two concentrations (0.5% and 1.0%). No evident dissociation can be observed in any condition, which means that these aggregates are not formed because of non-specific hydrophobic associations, but are due to native interactions between the purified monomers. The thermal stability of aggregates of TbTIM was analyzed following the change in fluorescence of the bound dye SYPRO-Orange. No evident change in the fluorescence signal could be observed in the initial points (up to 45 °C), indicating that the freshly purified enzyme does not expose hydrophobic surfaces (Fig. S5, Suppl. Information). We also tried to dissociate HsTIM, EhTIM and PfTIM aggregates with increasing concentrations of Lithium Dodecyl Sulphate (LDS). This SDS-analogue is suitable for electrophoresis at low temperature and low pH 33,34 . All the aggregates from the analysed TIMs can be dissociated with low LDS concentration, even though they differ in the amount necessary to complete their dissociation into single monomers (Fig. 4B). Thus, TIM aggregates have different structural stabilities, which form reversibly and do not involve covalent bonds (e.g. disulphide bonds). To confirm the latter, purified ScTIM, PfTIM, EhTIM and HsTIM were incubated 30 min at 25 °C in presence of 10 mM of Dithiothreitol (DTT) and subjected to CNE. No evident difference can be observed in the electrophoretic profiles of the aggregates, indicating that no apparent disulphide bonds are formed in the native aggregation process of these TIMs (Fig. S6, Supp. Information). Only GlTIM has been reported to form disulphide bonds between a cysteine partly exposed to the medium (C202) 35 . However, to avoid this problem we used GlTIM C202A, a mutant unable to form disulphide bonds 35 .  www.nature.com/scientificreports www.nature.com/scientificreports/ In the case of TIM, the catalytic activity is one of the parameters that can indicate how well or how poorly the enzyme is folded 36 . To elucidate if TIM, forming high molecular complexes, is able to perform catalysis, we determined in-gel activity, after separating the aggregates by CN-PAGE. Since TIM is inactive in the HEPES-imidazole buffer, the gels with the separated enzymes were incubated for 1.5 h in a developing buffer (see material and methods section 3.5). All aggregates were active and transformed MTT into formazan, even though there are differences in the intensities of the bands for different species (Fig. 5). This indicates that TIM oligomerization is not due to partial unfolding of the protein, and that these high molecular aggregates are catalytically active.

Stable tiM aggregates can be separated by liquid chromatography and their size can be estimated with DLS. Stable high molecular mass protein oligomers can be separated by different techniques 37 ;
however, in some cases, meta-stable oligomers can be formed, whose half-life can last from a few seconds to hours. Depending on the association and dissociation constants of the oligomers, these may be detected with some particular analytical method 9 . In gradient gels, proteins migrate through several networks of pores of different sizes; the movements the enzymes undergo and the obstacles they encounter during their migration, could favour their aggregation. Another factor could be the duration of the electrophoresis. To discard that aggregates of TIM are formed as artifices of gel electrophoresis and to characterize them further, a sample of TbTIM (from the SP Sepharose fast flow exchange column) was separated by an analytical Source 15 S column (with a matrix volume of 1 mL). This new separation yielded four different peaks, which eluted at different concentrations (between 50 and 100 mM) of NaCl (Fig. 6A, colour arrow heads). SDS-PAGE analysis of the eluted fractions revealed that all are pure TbTIM (Fig. 6B). The analysis of the elution fractions by CN-PAGE showed the mix of mainly two different species in different proportions (Fig. 6C). The specific activity of the four peaks was 5620, 5300, 5160 and 4999 µmol mg −1 min −1 , respectively. These values agree with the specific activity shown in Table 1, confirming that the mixes of TIM aggregates are active, with nearly equivalent enzymatic activities.
The reversible and rapid equilibrium of self-association, can be evidenced in separations of a purified protein by size exclusion chromatography or sedimentation, resulting in the appearance of multiple peaks formed by www.nature.com/scientificreports www.nature.com/scientificreports/ mixtures of oligomers 9 . Nevertheless, in our hands the TIM oligomers could not be isolated using gel filtration chromatography. We could only isolate dimers and monomers, even though this technique has shown an excellent resolution between monomers that differ by only one residue 38 . We also performed an analysis of particle size using dynamic light scattering (DLS) spectroscopy for the peaks from the Source column. Previous work has shown that the hydrodynamic radius determined by size exclusion chromatography and pulse field gradient NMR for ScTIM is 2.4 and 2.9 nm, respectively [39][40][41] . Our DLS analysis of TbTIM showed mainly two types of particles with hydrodynamic radii of 3.2 ± 0.3 and 116.3 ± 12 nm, which may correspond to aggregates of 1 and 38 dimers respectively (Fig. 6C). Additionally, the temperature of denaturation (Tm) of each peak was determined using thermal shift assay. The largest species had a Tm 2.5 °C higher than the smallest species (Fig. 6A, insert), indicating a correlation between structural stability and size of the aggregates.

Time favours the formation of non-reversible fibrillar aggregates in TbTIM.
In order to visualize the nature of TIM aggregates in more detail, we analysed two kinds of purified TbTIM by transmission electron microscopy (TEM): one corresponding to freshly purified protein, and the second with protein stored at 4 °C for 20 days. The microscopic images revealed the presence of multiple globular aggregates with sizes ranging from 20 to 100 nm (corresponding to aggregates of 6 to 30 dimers) (Fig. 7A, green arrow heads). Higher oligomers and fibrillar aggregates appeared with time in the stored sample (Fig. 7B, purple arrow heads), and are characteristic of amyloid fibril formation processes 42,43 . The accumulation and deposition of non-covalent homopolymers of proteins is associated with more than 25 pathologies, which include Alzheimer's disease, Parkinson's disease, Huntington's disease, and type II diabetes (reviewed in 16 ). www.nature.com/scientificreports www.nature.com/scientificreports/  www.nature.com/scientificreports www.nature.com/scientificreports/ TIM is one of the highly nitrotyrosinated proteins in Alzheimer's disease 14,44 . To further characterize the capacity of TbTIM to form fibrillar aggregates, freshly purified dimeric enzyme was incubated at 37 °C in presence of an oxidative/nitrative stress agent (heme peroxidase-H 2 O 2 -NO 2− ) 13 . At different times samples were taken and analysed by CN-PAGE. Figure 8A, (red arrow head) shows the decrease of free dimers, at increasing times of exposure to the stress agent, indicating the formation of high molecular aggregates, which probably do not enter the gel. The additional band is observed on top of the lane, whose intensity remains apparently constant in time, may correspond to the horseradish peroxidase, present in this assay, which has little or no migration in the gel of inverse polarity (Fig. 8A, green arrow head). To follow the formation of the high molecular weight aggregates of TIM, the fluorescence of the dye SYPRO orange and the variation of the Tm value were measured for aggregates (Fig. 8B). This dye binds to amylin amyloid fibrils, but not to pre-fibrillar intermediates 45 , therefore, the increase of the fluorescence indicates the fibrillary aggregation of TbTIM. In addition, because the amyloid fibrillar aggregates have high temperature stability 46 an increase in the Tm value is correlated to the native aggregation process, as shown in Fig. 6. Furthermore, typical fibrillar aggregates were observed by negative staining using electron microscopy after 24 h incubation (Fig. 8C, orange arrow heads). Finally, larger fibrillar aggregates (>1 µm), following a 48 h incubation, were stained with thioflavin T and observed under UV light in a fluorescence microscope (Fig. 8D). Thioflavine T interacts with a variety of amyloid fibrils in suspension, generating an amyloid-specific fluorescent signal [47][48][49] . Taken together, these data clearly show the propensity of TbTIM to form fibrillar amyloid aggregates. This conversion of the free dimeric form into fibrillar aggregates is directly related to the loss of enzymatic activity. This process is clearly accelerated in presence of oxidative/nitrative stressing conditions, giving rise to large fibrillar aggregates that are evident in TEM images after 24 h incubation (Fig. 8C), and that exhibit larger sizes than the fibrillar aggregates formed by the cold-stored enzyme (>360 h) (Fig. 7B). Additionally, the size of the stress-induced fibrillar aggregates is considerably larger (>1 µm) as judged by fluorescence microscopy (Fig. 8D). In general, it has been proposed that all peptides have the capability to form amyloid www.nature.com/scientificreports www.nature.com/scientificreports/ fibers, nevertheless, the propensity to form these kind of structures depends on multiple factors like charge, sequence, hydrophobicity, etc. 50 . Most of the amyloid fibers precursor peptides have highly unstable structures and are considered intrinsically disordered proteins 16 . Nevertheless, TIM possesses a defined structure and a high capacity of self-association that forms two kind of aggregates: native globular reversible aggregates and amyloid fibrillar aggregates in vitro. In contrast, nitrotyrosinated TIM associates with Tau and Abeta peptides to start the fibrillar association in vivo 14,44 .

Bioinformatic search of regions prone to aggregation in tiM. Based on experimental information and
computer-based approaches, multiple aggregation-prone amylogenic regions are present in many proteins 51 and have been identified in several sequences of TIM. Here, the amylogenic regions in the TIMs studied were predicted using the WALTZ 52 algorithm, a program that calculates the propensity to form parallel or antiparallel beta fibrillar structures for a given sequence. To analyse the amylogenic zones, the sequences of all TIMs were divided into eight regions, as previously reported by Rodríguez-Bolaños et al. 18 . Three conserved amylogenic regions were predicted for regions 2, 3 and 6 ( Fig. S7 Suppl. Information); the first region extends from residue 33 to 51, and is formed by external loop 1, beta sheet 2, internal loop 2 and the beginning of alpha helix 2; the second region includes residues 63 to 73, corresponding to external loop 2; and the third region extends from residues 59 to 69, and includes the beta sheet 6, internal loop 6 and the start of alpha helix 6 ( Fig. S8 Suppl. Information). It is also known that regions 1, 2 and 3 of TbTIM and TcTIM are involved in dimerization. The internal loops play a fundamental role in the correct positioning of the monomers, and are therefore also related to self-association 53 . It has also been observed that external loops 1 and 2 have an important role in the aggregation of the enzyme. Interference in communication between external loop 1 and the amino-terminus of helix 2, and of external loop 2, has been reported to promote aggregation of TcTIM and TbTIM after refolding 18 . In addition, previous studies described the presence of three regions in some TIM sequences that have some similarity (approximately 8-22%) with Abeta. Of these segments, the one with the highest similarity corresponds to amino acids 173 to 213 (in the sequences of Escherichia coli and Culex tarsalis) which have 20% identity with Abeta. This isolated fragment is capable of forming amyloid aggregates 54 . There are a few reports about proteins with the TIM barrel motif that form fibrillary aggregates, however, it has also been described that HsTIM forms amyloid deposits in patients with Alzheimer's disease 14,51,55 . The sequences of TbTIM and HsTIM share 50% identity and also exhibit 15 and 18% identity with Abeta, respectively (Suppl. Tables 5 and 6, Suppl. Information). Overall, the alignment between Abeta and the sequences of the TIMs studied shows an identity that ranges between 15 and 25% (Suppl. Table 5, Suppl. Information). GlTIM, LmTIM, EhTIM and TcTIM have the highest identity scores, 25, 23, 20 and 20%, respectively. The area where this similarity is found corresponds mainly to the hinged "lid" loop or catalytic loop, and in all cases consists of regions 6 and 7 which correspond to helix 6, external loop 6, beta sheet 7 and internal loop 1. This could possibly indicate that the fibrillary aggregation process is conserved among the TIMs from different species.
Concluding remarks. The formation of reversible aggregates by the self-association capacity of TIM, could favour their adaptation to some particular changes in the cell or in the extracellular environment. All TIMs we studied are capable of forming oligomers of different sizes, and their aggregation profiles appear to be influenced by their surface charges, that give rise to different groups of heterogeneous aggregates. It has recently been described that TIM has multiple functions in addition to its participation in glycolysis (reviewed in 56 ); this self-associative capacity could regulate some of its non-catalytic activities. In other pathological situations, the presentation of protein aggregates to the immune system could trigger the formation of autoantibodies 57 . TIM has been described as a potent allergen, which may also be due to its ability to self-associate. Since TbTIM is able to form fibrillary irreversible aggregates like HsTIM, this may be a conserved feature among triosephosphate isomerases.

Material and Methods
Heterologous expression of the tiMs from several species. The  hydrophobic interaction column (GE Healthcare, 30 mL column volume), previously equilibrated with the same buffer, but containing 2.0 M (NH 4 ) 2 SO 4 . Protein elution was performed using a 2.0 to 0 M gradient of (NH 4 ) 2 SO 4 . Four millilitre fractions were collected and analysed by SDS-PAGE; those containing TIM were pooled and dialyzed overnight against 3 L 100.0 mM TEA, 10.0 mM EDTA, pH 7.4 at 4 °C. After removal of the precipitated material (35 000 × g/30 min) the samples were loaded onto a Source 15Q anion exchange column (24 mL column volume), previously equilibrated with the same buffer. Protein elution was performed with a linear gradient of 0-500 mM NaCl. Four mL fractions were collected and analysed by SDS-PAGE; those fractions enriched with TIM were pooled and concentrated with an Amicon Ultra-15 Centrifugal Filter 10 kDa (EMD Millipore) to a final volume of 5 mL, and then precipitated with 70% (NH 4 ) 2 SO 4 and kept at 4 °C.
Purification of TcTIM and TbTIM. The cell pellet was suspended in 30 mL of lysis buffer (50.0 mM MES, 300.0 mM NaCl, 1.0 mM DTT, 0.5 mM EDTA, pH 6.3). The sample was sonicated for 1 min at a power setting of 5 W, with a rest of 2 min, for a total of 10 cycles. This suspension was centrifuged for 60 min at 200 000 × g. The supernatant was diluted with buffer A (50.0 mM MES pH 6.3) to a final concentration of 20.0 mM NaCl. The sample was then loaded onto a SP Sepharose fast flow column (30 mL volume column, GE Healthcare), previously equilibrated with buffer A. Protein elution was performed using a linear gradient of 0-500 mM NaCl. Three mL fractions were collected and analysed by SDS-PAGE. Those fractions enriched with TIM were pooled and (NH 4 ) 2 SO 4 was gradually added to reach the final concentration of 70% (w/v). The sample was incubated with gentle agitation for 16 hours at 4 °C. After this step, the purification with the Butyl toyopearl column and the subsequent (NH 4 ) 2 SO 4 precipitation were performed as described in section 3.2.
All the precipitated purified enzymes were dialyzed against 2 L of 100 mM TEA, 10 mM EDTA, pH 7.4, just before use. Protein concentration measured reading the absorbance at 280 nm, and using a molar extinction coefficient ε = 34950 M −1 cm −1 58 . native electrophoresis. Horizontal clear native electrophoresis was performed in a 7 × 10 Gel Box chamber (Labnet Int.). A 7 × 7 cm 6% acrylamide horizontal gel (0.5 cm width) with sample wells located in the middle of the gel was prepared, to allow migration of the samples in both directions during the electrophoretic separation. The buffers for the gels and the electrophoresis were 40.0 mM HEPES and 15.3 mM imidazole. The electrophoresis was performed at 80 V during 1 h at 4 °C. Clear native polyacrylamide gel electrophoresis (CN-PAGE) was performed using a Mini-PROTEAN II protein electrophoresis chamber (BIO-RAD). Only one gel was used for each run. All separations were carried out at a constant voltage of 150 V at 4 °C for 5 hours. The upper and lower reservoirs were filled with fresh buffer containing 40.0 mM HEPES and 15.3 mM imidazole. Before each run, the chamber with the gel and buffers was brought to the desired temperature. The gradient gels used had 4 to 10% polyacrylamide. For some experiments, different HEPES-imidazole ratios were used in the gels, but the concentration of the buffer in the anode and cathode chambers was always the one mentioned above. The HEPES and imidazole concentrations in the gels were as follows: HEPES 40.0 mM, imidazole 15.3 mM, or HEPES 88.9 mM, imidazole 34.0 mM, or HEPES 133.2 mM imidazole 50 mM. Ten micrograms of each purified enzyme were mixed with 5 µL of 16% (v/v) glycerol in 100 mM TEA, 10 mM EDTA, pH 7.4 in a total volume of 30 µL; these samples were the loaded onto each lane.
Blue native electrophoretic separation (BNE-PAGE) under native conditions was performed as reported by 59 using an acrylamide gradient of 4 -12%. One microliter of 5% Coomassie Serva blue G solution was added to the sample before loading the gel. Separation was performed at 4 °C and the electrophoretic front was monitored and not allowed to leave the gels.
After migration, all gels were fixed with 50% methanol and 10% acetic acid solution for 15 min and stained with 50% ethanol, 10% acetic acid and 0.1% Coomassie R-250 blue solution for 3-4 h with constant agitation. The gels were destained for 5-6 h with 10% acetic acid solution until the background was clear.

Determination of in vitro and in-gel enzymatic activity.
In vitro catalytic activity was measured following the inter-conversion of glyceraldehyde 3-phosphate (GAP) to dihydroxyacetone phosphate (DHAP) at 25 °C with the aid of a coupled enzymatic reaction with glyceraldehyde phosphate dehydrogenase (αGDH). The activity was followed by changes in absorbance at 340 nm due to the oxidation of NADH. The reaction mix con- For the determination of the activity in-gel, the enzymatic reaction was measured in the direction of DHAP to GAP. Enzyme activity was detected by changes in the oxidation of NAD + using a coupled colorimetric reaction (dimethyl thiazolyl diphenyl tetrazolium (MTT) assay). After the electrophoresis, the gels were incubated for 1.5 h at 4 °C in developing buffer (50.0 mM Tris, 5.0 mM MgSO 4 and 5.0 mM NaH 2 AsO 4 •7H 2 O, pH 7.6). After this time the gels were carefully placed in a glass container and uniformly impregnated with a solution containing 11.0 mM phenazine methosulphate (PMS), 19.3 µM MTT, 7.5 µM NAD + , 9.8 mM DHAP and 0.2 mg glyceraldehyde 3-phosphate dehydrogenase (GAPDH). The gel was carefully incubated for 2 to 5 min, in darkness, avoiding the formation of bubbles between both surfaces. To fix the colour developed by the formation of formazan, the gel was placed in a container with 7.0% acetic acid for 2 h. www.nature.com/scientificreports www.nature.com/scientificreports/ 30-125 mM NaCl in the same buffer. The specific activity of the fractions with the highest UV absorbance in each elution peak was measured. Additionally, the purity and the state of aggregation of the fractions were monitored by CN-PAGE and SDS-PAGE. estimation of the molecular weight of tiM aggregates by dynamic light scattering. The dynamic light scattering was performed in a Zetasizer Nano-Zs spectrum (Malvern Instruments, Ltd., U. K.). A 500 µL sample was deposited in the cuvette, and five measurements of each sample were made with 10 scans for each of them. Filtered (PVDF, 0.22 μM) 50.0 mM MES, pH 6.3, was used as a blank.
Microscopic visualization of tiM aggregates. Transmission electron microscopy. Nine microliter aliquots were adsorbed for 5 min onto freshly glow-discharged carbon-coated copper grids. Excess amount of sample was blotted with filter paper and the grids allowed to dry for another 5 min. The grids were stained with 2% uranyl acetate for 5 min followed by 10 min of drying. Images of the TIM aggregates were observed and photographed with Transmission Electron Microscope JEM-1200X EX II (JEOL, Japan) coupled to a CCD GATAN camera.
Fluorescence microscopy. TbTIM fibrillar aggregates were stained with 1 mM Thioflavin T in a 1:1 (protein:Thioflavin T) proportion. The sample was incubated 10 min in the dark and observed under blue light in a Nikon SMZ1500 fluorescent microscope. The image acquisition was performed with the NIS-Elements F software (Nikon, Japan).
Formation of TbTIM fibrillar amyloid aggregates by peroxynitration. The peroxynitrite production was performed as described previously 13 . Briefly, purified TbTIM (0.4 mg/mL) was incubated in 50 mM MES pH 6.3 with 20 µM radish heme peroxidase (Sigma), 1 mM H 2 O 2 and 1 mM NaNO 2 , at 37 °C with constant agitation (250 rpm). Samples were taken at different times and analysed by CN-PAGE, thermal shift assay and their enzymatic activity was determined.
The thermal shift assay was performed by following the fluorescence of the SYPRO-orange dye in a Real-Time PCR StepOnePlus System (Applied Biosystems, Massachusetts, USA). Each sample (8.4 µg) was loaded onto a 96-well plate and mixed with a 1:1000 dilution of SYPRO-orange dye in 50 mM MES pH 6.3 in a final volume of 20 µL. The melting curve was obtained using a 25-99 °C temperature gradient, with 490 nm as the exciting light wavelength and 575 nm as the detection wavelength.
prediction of regions prone to aggregation in the tiM sequences. The prediction of regions prone to aggregation was made using the WALTZ program 52 (http://switchlab.org/bioinformatics/waltz). These regions were calculated for all TIM sequences using a high-specificity limit to prevent false positives. Sequence alignments were performed with Clustal Omega 60 (https://www.ebi.ac.uk/Tools/msa/clustalo/).