Aggregation and Cellular Toxicity of Pathogenic or Non-pathogenic Proteins

More than 20 unique diseases such as diabetes, Alzheimer’s disease, Parkinson’s disease are caused by the abnormal aggregations of pathogenic proteins such as amylin, β-amyloid (Aβ), and α-synuclein. All pathogenic proteins differ from each other in biological function, primary sequences, and morphologies; however, the proteins are toxic when aggregated. Here, we investigated the cellular toxicity of pathogenic or non-pathogenic protein aggregates. In this study, six proteins were selected and they were incubated at acid pH and high temperature. The aggregation kinetic and cellular toxicity of protein species with time were characterized. Three non-pathogenic proteins, bovine serum albumin (BSA), catalase, and pepsin at pH 2 and 65 °C were stable in protein structure and non-toxic at a lower concentration of 1 mg/mL. They formed aggregates at a higher concentration of 20 mg/mL with time and they induced the toxicity in short incubation time points, 10 min and 20 min only and they became non-toxic after 30 min. Other three pathogenic proteins, lysozyme, superoxide dismutase (SOD), and insulin, also produced the aggregates with time and they caused cytotoxicity at both 1 mg/mL and 20 mg/mL after 10 min. TEM images and DSC analysis demonstrated that fibrils or aggregates at 1 mg/mL induced cellular toxicity due to low thermal stability. In DSC data, fibrils or aggregates of pathogenic proteins had low thermal transition compared to fresh samples. The results provide useful information to understand the aggregation and cellular toxicity of pathogenic and non-pathogenic proteins.

NaCl, pH 7.2). Protein samples (140 µL) were incubated with 15 µL of ANS stock solution and the fluorescence of ANS was excited at 385 nm and the emission intensities were observed at 500 nm. A 20 μL sample was also withdrawn for protein concentration determination using the Bradford protein assay. cell culture. PC-12 (ATCC ® CRL-1721 ™ ) cell lines from the American Type Culture Collection (ATCC) (Manassas, VA) were maintained at 37 °C under a humidified atmosphere of 5% CO 2. PC-12 cells were maintained in RPMI-1640 medium containing 5% (v/v) fetal bovine serum and 10% (v/v) heat-inactivated horse serum, supplemented with 1% penicillin (100 U/mL) and streptomycin (100 µg/mL). cytotoxicity of protein aggregates. An MTT (3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) reduction assay was performed to measure the toxicity of protein aggregates. Protein samples were prepared as described in the protein sample aggregation section. At designated time points (10 min, 20 min, 30 min, and 60 min), protein samples were neutralized to pH 7.4 and they were added to PC-12 cells (2 × 10 4 cells/well in 96 well plates) in RPMI 1640. The cells were treated with protein samples for 24 hours followed by the MTT assay. Next, 10 μL of MTT solution (5 mg/mL in PBS) was added to each well, and the cells were incubated for 4 hours. Yellow MTT solution is reduced to purple formazan in living cells and the absorbance of formazan crystals was measured at 570 nm (λ abs = 570 nm) using an Infinite ® 200 Pro microplate reader. Percentage cell viability was calculated by comparing the absorbance of the control cells (cells with medium only without any protein aggregates) to that of protein aggregates-treated cells (sample). Percentage cell viability = (sample with MTT -sample)/ (control with MTT -control) × 100. transmission electron micrograph (teM). Six proteins, BSA, catalase, pepsin, lysozyme, insulin, and SOD, in pH 2 solution (1.0 mg/mL) were incubated at 65 °C for 2 hours. Then, 20 μL of protein solution were placed on carbon-coated copper grids (Ted-Pella Inc, Redding, CA, USA), and the protein samples on the grid were negatively stained with 2% aqueous ammonium molybdate. After sufficient air-dry, grids were examined in a FEI Tecnai transmission electron microscope (TEM) at an accelerating voltage of 200 kV. Circular dichroism (CD). Six proteins, BSA, catalase, pepsin, lysozyme, insulin, and SOD, in pH 2 solution (1.0 mg/mL) were incubated at 65 °C for 2 hours. Secondary structure of fresh proteins at pH 7.0 or proteins at pH 2.0 for 2 hours was measured using a Jasco model J-815 circular dichroism (CD) spectropolarimeter (Jasco, Tokyo, Japan). CD spectra in the far UV range (190-260 nm) were obtained using a 2 mm quartz cell and a Xe lamp as a light source. A deconvolution method (BeStSel) of CD spectrum was used to estimate the secondary structures of fresh proteins and protein aggregates at pH 2.0 for 2 hours 48 .

Results and Discussion
Protein may cause cellular toxicity when aggregated. Fibrillar aggregates of proteins are associated with more than 20 degenerative diseases such as Alzheimer's disease, Parkinson's disease, and type II diabetes respectively 1,2 . Protein aggregation is influenced by physicochemical and biological factors such as temperature, pH, and salt concentration 49 . However, what leads to protein toxicity has not been clearly elucidated.
Six proteins, lysozyme, insulin, superoxide dismutase (SOD), bovine serum albumin (BSA), catalase, and pepsin, were selected and similarities and differences in aggregation mechanisms were investigated in this research. Three proteins, lysozyme, insulin, superoxide dismutase (SOD), were in the list of pathogenic proteins. Catalase, an antioxidant enzyme, was selected as a counterpart of SOD, another antioxidant enzyme. Pepsin, a digestive enzyme, was chosen as a counterpart of lysozyme, an antimicrobial enzyme. BSA, the most abundant protein in plasma, was used for a counterpart of insulin, a hormone that regulates the blood glucose level. In Fig. 1, six proteins (1 mg/mL) were incubated at pH 2 with 150 mM NaCl and 65 °C, and the aliquots were withdrawn at the following time points (0, 10, 20, 30, 60 and 120 minutes). Aggregation patterns were monitored by fluorescence of Thioflavin T (ThT) and 8-Anilinonaphthalene-1-sulfonic acid (ANS). Apparently all proteins seem to follow a nucleation-dependent aggregation model. However, three proteins, BSA, catalase, and pepsin, increased ThT fluorescence emission without a lag time and they reached the saturation level in 30 minutes. The ThT fluorescence differences between fresh and aged samples of BSA, catalase, and pepsin, were small compared to other three proteins. The intensity differences of ThT fluorescence between 0 minute sample and 30 minute sample were 28.7 ± 3.1 a.u (mean ± standard deviation) for BSA, 319.0 ± 41.4 a.u for catalase, and 12.7 ± 9.1 a.u for pepsin. Other three proteins, lysozyme, insulin, and SOD, increased ThT fluorescence emission with 10 minute lag time and they reached the saturation level in 30-60 minutes. The intensity differences of ThT fluorescence between 0 minute sample and 60 minute sample were 14,360.0 ± 778.8 a.u for insulin, 4,869.3 ± 166.3 a.u for SOD, and 546.7 ± 60.3 a.u for pepsin. The ANS fluorescence of all proteins was increased in 10 minutes and a plateau was reached after 10 minutes. Lysozyme, insulin, and SOD, reached a plateau slightly faster than BSA, catalase, and pepsin. Unlike other proteins, pepsin had a very small change in ANS fluorescence between 0 minute sample and 10 minute sample and it was 46.7 ± 17.6 a.u. The result is not surprising. Pepsin is a digestive enzyme in the stomach and the pH of the human stomach lumen is pH 1.5 to 3.5. A working pH range of pepsin is pH 1-4 and www.nature.com/scientificreports www.nature.com/scientificreports/ the optimum pH is around pH 2.5. At the condition, pH 2 with 150 mM NaCl and 65 °C, that we used, the structural change of pepsin should be affected by the temperature only, not by pH.
In this research, two fluorescent dyes, ThT and ANS, have been applied to characterize protein denaturation and protein aggregation. ANS is weakly fluorescent in hydrophilic environment such as water; however, its intensity is increased in hydrophobic environment such as nonpolar solvents 50,51 . ANS enhances fluorescence signals as it binds to hydrophobic regions on the protein surface 52,53 . Therefore, the ANS method has been used to detect transient states in protein denaturation. The fluorescent properties of ThT have also been well studied. ThT is used to investigate the formation of amyloid fibrils, as ThT does not bind to proteins in the folded, completely unfolded and partially folded or denaturated states of the type of a melted globule, as well as with amorphous aggregates of proteins 54,55 . Both amyloid fibrils and amorphous aggregates enhance the signal of light scattering; however, only amyloid fibrils increase the ThT fluorescence 54,55 . In protein aggregation, unfolded proteins could be formed by unfavorable environments, and the unfolded proteins are prone to form amorphous aggregates or fibrils via inappropriate interaction with other proteins since hydrophobic regions are exposed to the hydrophilic cell environment. For the protein aggregation in this research, both ANS and ThT fluorescence were monitored to understand the kinetics of protein denaturation and amyloid fibril formation.
In Fig. 2, aliquots of six proteins (1 mg/mL) at pH 2 with 150 mM NaCl and 65 °C were withdrawn at 10, 20, 30, and 120 minutes, neutralized to pH 7.4, and added to PC-12 cells (ATCC ® CRL-1721 ™ ) that was a cell line derived from a pheochromocytoma of the rat adrenal medulla. After 24 h of incubation, cellular viability of protein aggregates was measured by the reduction of MTT (3-(4,5-Dimethylthiazol-2-yl)-2,5-Diphenyltetrazolium Bromide). Fresh or aged BSA, catalase, and pepsin did not induce any cytotoxicity to PC-12 cells. There was no statistical difference in MTT reduction between cells only in medium and cells treated with fresh or aged BSA, catalase, and pepsin. Other three proteins, lysozyme, insulin, and SOD, were not toxic when they were fresh; www.nature.com/scientificreports www.nature.com/scientificreports/ however they induced the cytotoxicity when incubated for more than 30 minutes. For insulin, protein for 30 minute incubation caused 45.8 ± 12.6% toxicity to PC-12 cells and insulin for 120 minute incubation increased the cytotoxicity to 63.6 ± 4.2%.
Transmission electron micrograph (TEM) was used to visualize protein species formed at pH 2 and 65 °C for 2 hours. In Fig. 3, the morphological structures of BSA (Fig. 3A), catalase (Fig. 3C), and pepsin (Fig. 3E) at pH 2 and 65 °C for 2 hours were not different from those of fresh proteins. Although BSA (Fig. 3A) and catalase (Fig. 3C) did not produce fibrils, the denaturation of proteins resulted in seed-or protofibril-types of compounds. Due to the optimum pepsin activity at pH 1.5-1.6, the denaturation kinetics of pepsin might have been very slow in this condition (Fig. 3E), or pepsin was stable thermodynamically. Insulin (Fig. 3B) at pH 2 and 65 °C for 2 hours formed protofibrils or fibrils. The width of single protofibrils (n = 33) was 7.7 ± 1.8 nm. The length of most protofibrils varied from 161.0 nm to 638.9 nm. Protofibrils were arranged side by side to produce fibrils. In the same condition, SOD (Fig. 3D) and lysozyme (Fig. 3F) also formed protein aggregates. In Fig. 3D, misfolding of SOD initiated the formation of the seed for aggregation. The seed (n = 33) had uniform spherical shape with diameter of 9.6 ± 1.8 nm. The seeds elongated together and they became hairpin-like protofibrils. The oligomeric lysozyme protofibrils were also found in Fig. 3F. The width of single protofibrils (n = 33) was 8.5 ± 1.1 nm.
Based on Figs. 1-3, it is clear that insulin (1 mg/mL) at pH 2 and 65 °C for 120 minutes forms fibrils that are toxic to PC-12 cells. Fresh insulin that is mainly composed of insulin monomers is not toxic to PC-12 cells. The cellular toxicities of insulin at the time points of 10 minutes and 30 minutes are less than that of insulin fibrils at 120 minutes. In this result, there is no clear evidence that intermediates of insulin are more toxic than fibrils. However, the cellular toxicity of oligomeric β-amyloid (Aβ), another pathogenic protein for Alzheimer's disease, has been studied relatively well. Soluble oligomeric forms of Aβ, termed amyloid-derived diffusible ligands (ADDLs) are known to be toxic species 56,57 . Cytotoxicity of Aβ may be maximal for the oligomers of intermediate size 58 . From monomer or dimer, cytotoxicity increased with oligomer size; however, cytotoxicity decreased with size when Aβ oligomers are bigger than 12-mers or higher 59,60 . In this research, other two pathogenic proteins, SOD and lysozyme, at pH 2 and 65 °C for 120 minutes forms protofibrils or premature fibrils (Fig. 3D,F) that induce cellular toxicity; however, they are less toxic than insulin fibrils. BSA and pepsin do not form any oligomers or fibrils at pH 2 and 65 °C for 120 minutes (Figs. 3A,E). In the same condition, there are small species in catalase (Fig. 3C); however, those small species are also shown in fresh catalase, which may result from the denaturation of catalase. BSA, catalase, and pepsin in this condition do not cause any cellular toxicity. Figures 1-3 demonstrate that cytotoxicity of proteins depends on the size of aggregates or fibrils as well as the type of proteins.
Proteins in the physiological condition are thermodynamically stable and they are in a low energy state. Based on the state of protein folding, thermodynamic stability of proteins will change. Differential scanning calorimetry (DSC) is one of useful tools to measure thermal denaturation of proteins. In Fig. 4, thermal stabilities of six proteins were studied by DSC. Fresh proteins as folded proteins were used for control groups in the DSC analysis. Aged proteins (1 mg/mL) were also prepared at pH 2 (150 mM NaCl) and 65 °C for 120 minutes and the protein solution was freeze-dried. DSC thermogram of all powder proteins showed one thermal transition between 40 °C and 55 °C. Aged samples of lysozyme, insulin, SOD, and pepsin had low thermal transition compared to fresh samples. Other two proteins, BSA and catalase, demonstrated opposite thermal transition. Aged samples of BSA and catalase had high thermal transition compared to fresh samples. In the protein aggregation experiment, all samples were incubated at 65 °C that was above thermal transition temperatures of all proteins 61,62 . During the 2 hour incubation time, all proteins were denatured; however, protein denaturation of BSA and catalase was reversible as shown in TEM images. Reversible proteins were thermodynamically stable and thermal transitions of aged BSA and catalase were higher than those of fresh proteins. However, denaturation of other proteins, lysozyme, insulin, and SOD in this condition was irreversible and unfolded proteins were not refolded. Aged samples of lysozyme, insulin, and SOD were thermodynamically unstable, and thermal transitions of aged samples were lower than those of fresh proteins. All pathogenic proteins, lysozyme, insulin, and SOD, became thermodynamically unstable when they were incubated at 65 °C for 2 hours. Pre-fibrils, oligomers, fibrils or aggregates of  Table 1. Percentage of secondary structure fractions for proteins at fresh (pH 7.0 for 0 hours) and aged (pH 2.0 for 2 hours) conditions. The data were generated by deconvolution of CD spectra in Fig. 5.
Circular dichroism (CD) spectroscopy has been used to examine secondary structure changes of six proteins during aggregation. As seen in Fig. 5, fresh proteins were a mixture of α-helix, β-sheet, turn, and random coil. Although changes in the CD spectra were observed when proteins were incubated at pH 2 and 65 °C for 2 hours, all six proteins did not indicate a significant change in the basic secondary structure elements from the deconvolution of CD spectra in Table 1.
Protein secondary structural information can be derived from CD signals. Two negative bands at 222 nm and 208 nm and a positive band at 193 nm are characteristic α-helix CD spectra 67 . The two negative bands arise from π-π* and n-π* transitions in the amide groups. Negative bands at 218 nm and positive bands at 195 nm are well-defined antiparallel β-sheets 68 . The conversion of the secondary structure into a predominantly antiparallel β-sheet is a pathologic process of pathogenic proteins 69 . In Table 1, only BSA and lysozyme increased β-sheet structure in the pH 2 and 2 hour condition. Other pathogenic proteins, insulin and SOD, increased α-helix structure in the pH 2 and 2 hour condition. Overall, pathogenic proteins formed more ordered structures such as α-helix and β-sheet in the pH 2 and 2 hour condition; however, the differences are not drastic. www.nature.com/scientificreports www.nature.com/scientificreports/ In order to understand the effect of protein concentration on the protein aggregation, we increased the protein concentration of six proteins to 20 mg/mL and we repeated experiments of protein aggregation kinetics, cytotoxicity, and TEM imaging. Protein concentration is one of key factors in protein aggregation 14 . Proteins increase the chance of collision at higher concentrations, which enhances the formation of protein aggregates 16 .
Here, six proteins (20 mg/mL) were incubated at pH 2 with 150 mM NaCl and 65 °C, and the aliquots were withdrawn at the following time points (0, 10, 20, 30, 60 and 120 minutes). In Fig. 6, all proteins increased ThT fluorescence emission without a lag time; however, BSA, catalase, and pepsin reached the top of ThT fluorescence in 10 minutes and the fluorescence were diminished with time after 10 minutes. Other proteins, lysozyme, insulin, and SOD, steadily increased ThT fluorescence emission and they reached the saturation level in 30 minutes. The ANS fluorescence of all proteins was increased up to 10 minutes, confirming the presence of molten globular intermediates or denaturation of proteins.
In Fig. 7, aliquots of six proteins (20 mg/mL) at pH 2 with 150 mM NaCl and 65 °C were withdrawn at 0, 10, 30, and 120 minutes, neutralized to pH 7.4, and added to PC-12. BSA, catalase, and pepsin samples at 10 minutes caused 39.7 ± 9.1%, 19.3 ± 2.0%, and 11.8 ± 2.4% toxicities to PC-12 cells respectively. Interestingly toxicities of aged BSA, catalase, and pepsin disappeared when they were incubated longer than 10 minutes. Other three proteins, lysozyme, insulin, and SOD, induced the cytotoxicity when incubated for more than 30 minutes, which was similar to lower concentration (1 mg/mL) of proteins.
Transmission electron micrograph (TEM) was used again to visualize the higher concentration of protein species formed at pH 2 and 65 °C for 2 hours. In Fig. 8, all proteins except pepsin showed larger protein aggregates. All proteins formed spherical species in this condition and the sizes of spherical species were between 8 nm and 12 nm in diameter. Spherical species were basic unit of protofibrils and they elongated into protofibrils or fibrils via doughnut shapes. The diameter of protofibrils or fibrils was around 10 nm that was the characteristic diameter of fibrils.
In Figs. 6 and 7, even non-pathogenic proteins such as BSA and catalase, induced cellular toxicities at a higher concentration (20 mg/mL). In the incubation condition of pH 2 and 65 °C, all proteins are denatured and the fluorescence of ANS was enhanced as a result. In both 1 mg/mL and 20 mg/mL concentrations, non-pathogenic proteins increased the fluorescence of ANS less than 10 minutes. However, the ThT fluorescence of BSA and catalase at 20 mg/mL was about 10-fold higher than that of BSA and catalase at 1 mg/mL. It is clear that BSA and catalase at 20 mg/mL have more ordered structure than at 1 mg/mL since thioflavin T increases the fluorescence signal when it binds to oligomeric or fibrillar proteins with a high β-sheet or α-helix content. However, the toxic species are not clearly understood. One possible explanation is that BSA or catalase may produce same toxic species at both 1 mg/ mL and 20 mg/mL, and only 20 mg/mL concentration generates enough amounts of toxic species to cause cellular toxicity. Another explanation is that BSA or catalase at 20 mg/mL makes different species from BSA or catalase at 1 mg/mL, and those species could be an ordered form such as micelles that induce cellular toxicity. Interestingly BSA www.nature.com/scientificreports www.nature.com/scientificreports/ or catalase at 20 mg/mL caused the cellular toxicities at the only early incubation time. As time went, BSA or catalase had less ordered structures and they became less hydrophobic since both ThT and ANS signals were reduced. Unstructured hydrophilic compounds may be relatively less toxic than ordered hydrophobic structures.
In summary, we investigated the cellular toxicity of pathogenic or non-pathogenic protein aggregates. Three proteins, lysozyme, insulin, superoxide dismutase, were selected as pathogenic proteins, and other three proteins, catalase, SOD, pepsin, were chosen as non-pathogenic proteins. Six proteins (1 mg/mL) were incubated at pH 2 with 150 mM NaCl and 65 °C, and three pathogenic proteins, lysozyme, insulin, and SOD, enhanced ThT fluorescence emission and they reached the saturation level in 30-60 minutes. Based on TEM images and MTT test, fibrils or aggregates of three pathogenic proteins induced the cellular toxicity. Other three non-pathogenic proteins, BSA, catalase, and pepsin were stable, and they didn't cause the cellular toxicity in this condition. According to DSC analysis, fibrils or aggregates of pathogenic proteins had low thermal transition compared to fresh samples. Low thermal stability is associated with the cellular toxicity. At a higher concentration (20 mg/mL), even non-pathogenic proteins cause cellular toxicity in the early incubation time. Pathogenic proteins induced the cellular toxicity like a lower concentration (1 mg/mL). All results in this research would be beneficial to understand similarities and differences in protein aggregation kinetics, cellular toxicity, and morphological structures of pathogenic and non-pathogenic proteins.