Toxic effects of endoplasmic reticulum stress transducer BBF2H7-derived small peptide fragments on neuronal cells

Aggregation, fibril formation, and deposition of amyloid β (Aβ) protein are believed to be the central pathogeneses of Alzheimer's disease (AD). Numerous studies have shown that fibril formation is promoted by preformed seeds at the beginning of the aggregation process. Therefore, aggregated molecules that promote fibrillization of Aβ protein as seeds could affect the pathology. We recently found that approximately 40 amino acid hydrophobic peptides, BBF2H7-derived small peptide (BSP) fragments, are generated via intramembranous cleavage under endoplasmic reticulum (ER) stress conditions. Interestingly, similar to Aβ protein, the fragments exhibit a high aggregation propensity and form fibril structures. It has been noted that ER stress is involved in the pathogenesis of AD. In this study, we examined the effect of BSP fragments on aggregation and cytotoxicity of Aβ1-40 protein, which is generated as a major species of Aβ protein, but has a lower aggregative property than Aβ1-42 protein. We demonstrated that BSP fragments promote aggregation of Aβ1-40 protein. Aggregates of Aβ1-40 protein mediated by BSP fragments also exhibited potent neurotoxicity. Our findings suggest the possibility that BSP fragments affect accumulation of Aβ proteins and are involved in the pathogenesis of AD.


Introduction
Alzheimer's disease (AD) is the most prevalent form of dementia. AD is characterized by misfolding, fibrillization, and accumulation of the amyloid β (Aβ) peptide, which results in the formation of extracellular senile plaques as a pathological hallmark (Masters et al., 1985). A 37-43 amino acid hydrophobic Aβ peptide is generated by sequential proteolytic processing of the amyloid precursor protein (APP) (Haass et al., 2012). APP is a type-I transmembrane protein with its N-terminus within the lumen/extracellular space and C-terminus within the cytosol. In the first step of the sequential proteolysis, β-secretase cleaves APP at the N-terminal luminal domain. This process removes a large part of the ectodomain of APP and generates a membrane-retained Cterminal fragment (CTF). CTF is processed further to Aβ and the APP intracellular domain by γ-secretase that cleaves within the transmembrane region. Aβ is located between the β-and γ-secretase cleavage sites. Heterogenous proteolysis of CTF by γ-secretase principally generates 40 and 42 amino acid peptides, Aβ 1-40 and Aβ 1-42 , respectively. Although the relative abundance of Aβ 1-40 (~90%) in the brain is much greater than that of Aβ 1-42 (~10%), amyloid plaques consist predominantly of Aβ 1-42 in patients with AD (Iwatsubo et al., 1994;Roher et al., 1993). Aβ 1-42 is more hydrophobic than Negative-stain transmission electron microscopy micrographs of amyloid fibrils increased by addition of Aβ 1-42 or BSP fragments. Mixtures of 5 µM Aβ 1-40 and 0.15 µM Aβ 1-42 , BSP fragments, or 0.3 µM Apelin-36, mixture of 0.3 µM Aβ 1-42 and 0.3 µM BSP fragments, or each peptide alone were incubated for 48 h. After incubation, the peptides were stained with uranyl acetate. Scale bar: 200 nm. (C) Length, width, and periodicity of fibrils derived from a mixture of Aβ 1-40 and Aβ 1-42 , or Aβ 1-40 and BSP fragments (mean ± SD, n = 5). *P < 0.05, **P < 0.01 relative to a mixture of Aβ 1-40 and Aβ 1-42 ; significance was calculated by the Student's t-test. (D) Amounts of fibrils derived from a mixture of Aβ 1-40 and Aβ 1-42 or Aβ 1-40 and BSP fragments (mean ± SD, n = 30). *P < 0.05 relative to Aβ 1-40 alone; significance was calculated by Dunnett's method. (E) Time course of the length of fibrils derived from a mixture of Aβ 1-40 and Aβ 1-42 , or Aβ 1-40 and BSP fragments (mean ± SD, n = 5). *P < 0.05, **P < 0.01 and ***P < 0.001 relative to a mixture of Aβ 1-40 and BSP fragments at a same time point; significance was calculated by the Student's t-test.
Aβ aggregation is promoted in a prion-like manner by the presence of misfolded Aβ seeds that act as a core structure of amyloid fibrils (Walker and Jucker, 2015). Local overproduction of Aβ 1-42 aggregates could induce fibrillization of abundant and normally soluble Aβ 1-40 . This concept is supported by several animal models by showing accelerated Aβ pathology in host organisms after intracerebral injection of brain homogenates containing Aβ aggregates (Kane et al., 2000;Meyer-Luehmann et al., 2006). Additionally, another fibril-forming peptide, amylin, promotes fibrillization and deposition of Aβ in AD model mice (Moreno-Gonzalez et al., 2017). Hence, other aggregative molecules could affect aggregation and deposition of Aβ and modify the pathogenesis of AD.
Several studies have indicated that endoplasmic reticulum (ER) stress is closely related to AD pathology. Numerous studies have reported that various ER stress markers, such as heat shock protein 70 (also known as BiP), phosphorylated PKR-like endoplasmic reticulum kinase, phosphorylated inositol-requiring enzyme 1, and phosphorylated eukaryotic translation initiation factor 2-α, are increased in postmortem brain samples from patients with AD (Hetz and Saxena, 2017). Attenuation of ER stress by treatment with a chemical chaperone, 4phenylbutyrate, significantly reduced the number of plaques in the hippocampus and reversed cognitive deficits in AD mouse models, which indicated a causal link between ER stress and neurodegeneration (Ricobaraza et al., 2011;Wiley et al., 2011). However, the precise mechanisms underlying ER stress and the pathogenesis of AD are not completely understood.
Recently, we found that novel small hydrophobic peptides, namely BBF2H7-derived small peptide fragments (BSP fragments), are produced in response to ER stress (Matsuhisa et al., 2020). BBF2H7, one of the OASIS family members, is a type-II transmembrane protein that is highly expressed in neurons, chondrocytes, and certain tumor tissues (Iwamoto et al., 2015;Kondo et al., 2007;Saito et al., 2009Saito et al., , 2014. The N-terminal segment contains a basic leucine zipper domain and projects into the cytosol. This is followed by a 20 amino acid transmembrane domain and 123 amino acid C-terminal domain that projects into the ER lumen. When unfolded proteins accumulate in the ER, BBF2H7 translocates from the ER to Golgi apparatus. In the Golgi, BBF2H7 is a substrate for two proteases, site-1 protease (S1P) and site-2 protease (S2P). S1P cleaves BBF2H7 in the luminal region. The N-terminal domain of BBF2H7 remains attached to the membrane through the transmembrane segment. The transmembrane region of this segment is then cleaved by S2P. This intramembranous cleavage releases the Nterminal fragments and 40-50 amino acid BSP fragments (Kondo et al., 2007;Matsuhisa et al., 2020). Although the cleavage enzymes are different, BSP fragments are also released by a two-step cleavage process in the luminal and transmembrane domains of the precursor, which is similar to Aβ. Additionally, BSP fragments exhibit a high aggregation propensity and form fibril-like structures (Matsuhisa et al., 2020). Thus, we hypothesized that hyper and prolonged production of BSP fragments by ER stress may affect aggregation and neurotoxicity of Aβ, which modifies AD pathogenesis.
In this study, we examined the effects of BSP fragments on fibrillization and cytotoxicity of Aβ. Furthermore, we investigated the degradation pathway of BSP fragments. Accordingly, we found that coincubation of BSP fragments promotes fibrillization and cytotoxicity of Aβ 1-40 .

BSP fragments promote cytotoxicity of Aβ
Next, we analyzed the effect of BSP fragments on cytotoxicity of Aβ 1-40 using synthetic peptides and human neuroblastoma SK-N-SH cells. Aβ 1-40 at 5 μM was mixed with 0.15 μM Aβ 1-42 , 0.15 μM BSP fragments, or 0.3 μM Apelin-36 and incubated at 37°C for 24 h. The incubated peptides were then applied to SK-N-SH cells for 48 h ( Fig. 2A,  B). Cells treated with 5 μM Aβ 1-40 alone did not show visible morphological changes. However, 0.15 μM Aβ 1-42 slightly decreased the number of cells. Aβ 1-40 preincubated with Aβ 1-42 exhibited more potent cytotoxicity than either Aβ peptide alone, as described previously (Sowade and Jahn, 2017). Interestingly, almost all cells shrank after treatment with Aβ 1-40 coincubated with BSP fragments. A large proportion of cells were intact in the presence of 0.15 μM preincubated BSP fragments alone. Morphological changes were not observed in cells treated with a mixture of Aβ 1-40 and Apelin-36. Taken together, these results indicate that BSP fragments enhance the cytotoxicity of Aβ 1-40 .

BSP fragments also exhibit cytotoxicity
BSP fragments have similar characteristics to those of Aβ. Accordingly, BSP fragments may directly cause cytotoxicity. To investigate the potential contribution of BSP fragments to neurodegeneration, we examined their cytotoxicity. BSP fragments were preincubated at 37°C for 24 h and then applied to SK-N-SH cells for 48 h. A large proportion of cells were intact, but slightly decreased in number in the presence of 5 μM preincubated BSP fragments (Fig. 3A, B). The cells shrank after addition of 10 or 20 μM fragments. Moreover, treatment with the above 10 μM preincubated peptides significantly decreased cell survival. These results indicate that aggregated BSP fragments themselves are cytotoxic molecules.

BSP fragments are degraded through the lysosomal degradation pathway
Next, we analyzed the expression levels of full length and N-terminal fragments of BBF2H7, BSP fragments, and ER stress markers in SK-N-SH cells transfected with BBF2H7 under ER stress conditions induced by ER stressor thapsigargin (Tg) by Western blotting. In mammalian cells, three major ER stress transducers have been well established: PKR-like endoplasmic reticulum kinase, inositol-requiring enzyme 1, and ATF6 (Harding et al., 1999;Tirasophon et al., 1998;Yoshida et al., 2000). These transducers induce expression of C/EBP homologous protein (CHOP), spliced form of X-box binding protein 1 (XBP1s), and BiP/Grp78, respectively (Ron and Walter, 2007). Therefore, we analyzed the expression levels of these proteins as ER stress markers. BiP was increased from 12 to 24 h after treatment with Tg (Fig. 4A). Treatment with Tg induced expression of CHOP and XBP1s at 6 h, which decreased gradually from 12 to 24 h. The 80 kDa full length BBF2H7 and 60 kDa N-terminal fragments were elevated gradually from 6 h after treatment, which indicated cleavage of BBF2H7. BSP fragments were also increased after treatment with Tg for 12 h, which followed the increase in N-termini. These data indicate that BSP fragments are produced after induction of CHOP and XBP1s, and that production of BSP fragments might be sustained under ER stress conditions. Furthermore, we investigated the degradation pathway of BSP fragments. SK-N-SH cells transfected with BBF2H7 were treated with Tg with or without proteasome inhibitor MG132 or lysosomal inhibitor bafilomycin A1 (BafA1). Next, we performed Western blotting using anti-BBF2H7 N-terminus and -BSP fragment antibodies. Full length and N-terminal fragments of BBF2H7 were increased gradually in Tg- Results are presented as mean relative cell counts of surviving cells compared with the untreated group ± SD (n = 3). ***P < 0.001 relative to control; significance was calculated by Dunnett's method. treated cells (Fig. 4B). BSP fragments appeared at 12 h after treatment. Treatment of cells with Tg and MG132 resulted in a dramatic increase in BSP fragments. Levels of full length BBF2H7 and N-terminal fragments were also elevated in these cells. BBF2H7 is degraded by the ubiquitin-proteasome system under normal conditions (Kondo et al., 2012). Stabilization of full length BBF2H7 by inhibition of the proteasome leads to an increase in the N-terminal and BSP fragments. Cotreatment with BafA1 did not increase the levels of N-terminal fragments compared with Tg-treated cells, which indicated that cleavage of BBF2H7 was not affected. However, levels of BSP fragments were increased significantly by lysosomal inhibition. These data indicate that BSP fragments were degraded in lysosomes. Additionally, we performed immunofluorescence staining of SK-N-SH cells transfected with BBF2H7 (Fig. 4B). Immunoreactivities of BSP fragments did not overlap with those of lysosome-associated membrane protein 2 (LAMP2; a lysosomal marker) in cells treated with Tg. BSP fragments accumulated in LAMP2positive lysosomes after inhibition of lysosomal functions by cotreatment with BafA1. The anti-BSP fragment antibody used in this study also detects full length BBF2H7. BSP fragments were not produced from BBF2H7 mutated in its S1P recognition site (BBF2H7 S1Pmut. ; RNLL to ANLV) (Matsuhisa et al., 2020). Then, we analyzed the immunoreactivities of BSP fragments in SK-N-SH cells transfected with BBF2H7 S1Pmut. . The signals of BSP fragments did not overlap with those of LAMP2 in cells expressing BBF2H7 S1Pmut. in spite of treating the cells with Tg and BafA1. Therefore, these findings suggest that BSP fragments are constitutively degraded in lysosomes. Lysates were subjected to Western blotting using anti-BBF2H7 N-terminus, anti-KDEL (to detect BiP), anti-XBP1s, and anti-CHOP antibodies. To detect BSP fragments, lysates were immunoprecipitated using an anti-BSP fragment antibody followed by Western blotting with the same antibody. (B) Western blotting of BBF2H7 N-terminal and BSP fragments in SK-N-SH cells expressing BBF2H7. Cells were treated with 1 µM Tg alone or together with 10 µM MG132 or 100 nM bafilomycin A1 (BafA1) for the indicated times. (C) Immunofluorescence staining of BSP fragment (Green) and lysosome-associated membrane protein-2 (LAMP2; Red) in SK-N-SH cells expressing BBF2H7 or BBF2H7 S1Pmut. . Cells were treated with 1 µM Tg alone or together with 100 nM BafA1 for 12 h. Right most panels show higher magnification images in the boxes shown in merged panel. Scale bars: 10 µm (BSP fragment, LAMP2, and lower magnification merged panels) and 1 µm (higher magnification images).

Discussion
We have shown that BSP fragments promote fibrillization of Aβ 1-40 (Fig. 1B). Formation of amyloid fibrils is described by the seeding-nucleation polymerization model (Jarrett and Lansbury, 1993). First, monomeric Aβ is incorrectly folded and forms small oligomeric aggregates. The aggregates act as "seeds" in fibrilization of Aβ. This aggregation process is thermodynamically unfavorable. Then, the seeds rapidly recruit and misfold the other Aβ monomers. The misfolded Aβ monomers bind to the seeds and form fibrils. Formation of seeds is a rate-limiting step in fibrillization. Several studies have shown that the aggregates derived from different proteins such as prion protein, tau, and α-synuclein act as preformed seeds, which accelerate fibril formation of Aβ (Morales et al., 2013). It has been proposed that the amyloid structure promotes misfolding of other amyloidogenic proteins (Makarava and Baskakov, 2012). Our previous study revealed that BSP fragments easily aggregate and form fibril structures similar to Aβ and other amyloidogenic proteins (Matsuhisa et al., 2020). Aggregated BSP fragments could recruit monomeric Aβ and change its conformation to a misfolded structure as preformed seeds, which accelerates fibril formation.
BSP fragments themselves also decreased neuronal cell viability (Fig. 3A, B). Although the cytotoxic mechanisms of Aβ are still not fully determined, some researchers have proposed that Aβ cause membrane damage through the formation of stable pores on the membrane, which result in severe neurotoxicity (Kotler et al., 2014). Other amyloidogenic proteins including 37 amino acid peptide amylin also exhibit cytotoxicity through the same mechanism (Cheng et al., 2013). Hydrophobic interactions between amyloidogenic proteins including Aβ and membranes is essential for their cytotoxicity. A previous study has suggested that the fibril-forming propensity is partly related to the assembly of Aβ within the membrane (Yip and McLaurin, 2001). BSP fragments also contain a highly hydrophobic N-terminal region and form a fibril structure similar to Aβ (Matsuhisa et al., 2020). Therefore, it is possible that BSP fragments disrupt cell membrane integrity and cause cell death via an interaction between their N-terminal hydrophobic region and membranes. Additionally, Aβ has been reported to exhibit cytotoxicity through production of reactive oxygen species (ROS) (Uttara et al., 2009). Aβ binds to metal ions, such as Cu, Zn, and Fe, via the imidazole ring of His residues (Cheignon et al., 2018). The Aβ-metal complex plays an important role in production of ROS. Another amyloidogenic peptide, amylin, also binds to Cu ions via the His residue and the complex produces ROS (Seal and Dey, 2018). BSP fragments contain a His residue in the luminal region. Taken together, another possible cytotoxic mechanism of BSP fragments is that the fragments form a complex with metal ions via the His residue and produce ROS, which results in cell death.
BSP fragments accumulate and are metabolically degraded in the lysosome (Fig. 4). Many previous investigations have found lysosomal dysfunction in patients with AD (Whyte et al., 2017). It is natural that BSP fragments accumulate in the lysosome because of impaired lysosomal degradation in the brain of AD patients. It is known that Aβ resides at the outer membrane of multivesicular bodies (MVBs) in the neurons of AD patients (Takahashi et al., 2002). However, it has been frequently observed that Aβ accumulates in the lysosome of patients with AD (Zheng et al., 2012). MVBs fuse with the lysosome and then their contents undergo degradation. Thus, dysfunction of the lysosome reduces degradation of Aβ, which results in accumulation in the lysosome. Lysosomal accumulation of BSP fragments suggests the possibility that the fragments act as seeds for Aβ fibrils and promote fibril formation in the lysosome. Therefore, it would be interesting to analyze the promoting effects of BSP fragments on aggregation of Aβ in vivo.
Interestingly, ER stress markers appeared even in morphologically healthy neurons of patients in the early stages of AD (Hoozemans et al., 2009). It has also been reported that the number of neurons containing phosphorylated PKR-like endoplasmic reticulum kinase, which is an ER stress sensor, correlates with the pathological stage in post-mortem AD brains (Hoozemans et al., 2009). These observations indicate that ER stress is induced in the early stages of AD and the ER stress response is activated as the disease progresses. Because production of BSP fragments is dependent on ER stress, the fragments could be sustainably produced in neurons from the early stages of AD. Taken together, it is conceivable that long term ER stress and lysosomal dysfunction could synergistically cause overproduction and accumulation of BSP fragments, which accelerates fibril formation of Aβ.
In conclusion, our results suggest the possibility that BSP fragments promote deposition of Aβ as a core structure of amyloid fibrils. We believe that our findings may contribute to a better understanding of the pathological mechanisms and the development of novel therapeutic approaches of AD. In vivo analyses of the significance of BSP fragments in AD pathogenesis may provide novel insights into the aggregation mechanisms of Aβ.
The pcDNA3.1(+) vector expressing BBF2H7 was constructed previously (Kondo et al., 2007). Transfection of the expression vector was performed using ScreenFectA (Wako) in accordance with the manufacturer's protocol. Cells transfected with the expression vector were used in experiments at 24 h after transfection.

Transmission electron microscopy (TEM) for observation of fibril formation of peptides
Fibril formation and acceleration of Aβ 1-40 fibrillization of BSP fragments were analyzed using a modification of previously published protocols (Yanagida et al., 2009). Briefly, synthetic BSP fragments were dissolved in dimethyl formamide to a concentration of 10 mM. The resultant peptides were mixed in 50 mM potassium phosphate buffer (pH 7.4). Diluted peptides were then incubated at 37°C for 0, 3, 6, 12, 24, or 48 h. After incubation, the peptides were centrifuged for 10 min at 20,600×g, and precipitated fibrils were analyzed by TEM.
Precipitated fibrils were suspended in 10 μl distilled water. Samples (3 μl) were applied to carbon-coated Slidefilm SLC-C15 (STEM, Tokyo, Japan) and incubated for 3 min at room temperature. Excess samples were absorbed with filter paper, after which an equal volume of uranyl acetate solution was added. After incubation for 2 min at room temperature, the solution was removed and the grid was dried in air. Samples were examined under a JEM-1400 transmission electron microscope (JEOL, Tokyo, Japan).

Cell survival assay
BSP fragments, Aβ peptides, or a mixture of Aβ, BSP fragments, or Apelin-36 were preincubated at 37°C for 24 h in 50 mM potassium phosphate buffer (pH 7.4). SK-N-SH cells were treated with preincubated peptides for 48 h. The number of surviving cells was counted on the basis of morphological changes.

Statistical analysis
Statistical comparisons were made using the Student's t-test, Tukey-Kramer method, or Dunnett's method. The statistical significance of differences was determined by P < 0.05.