EDEM1 Regulates Amyloid Precursor Protein (APP) Metabolism and Amyloid-β Production

Endoplasmic reticulum (ER) degradation-enhancing α-mannosidase-like protein 1 (EDEM1) is a quality control factor directly involved in the endoplasmic reticulum-associated degradation (ERAD) process. It recognizes terminally misfolded proteins and directs them to retrotranslocation which is followed by proteasomal degradation in the cytosol. The amyloid-β precursor protein (APP) is synthesized and N-glycosylated in the ER and transported to the Golgi for maturation before being delivered to the cell surface. The amyloidogenic cleavage pathway of APP leads to production of amyloid-β (Aβ), deposited in the brains of Alzheimer’s disease (AD) patients. Here, using biochemical methods applied to human embryonic kidney, HEK293, and SH-SY5Y neuroblastoma cells, we show that EDEM1 is an important regulatory factor involved in APP metabolism. We find that APP cellular levels are significantly reduced after EDEM1 overproduction and are increased in cells with downregulated EDEM1. We also report on EDEM1-dependent transport of APP from the ER to the cytosol that leads to proteasomal degradation of APP. EDEM1 directly interacts with APP. Furthermore, overproduction of EDEM1 results in decreased Aβ40 and Aβ42 secretion. These findings indicate that EDEM1 is a novel regulator of APP metabolism through ERAD.


Introduction
Alzheimer's disease (AD) is irreversible, progressive, and the most common brain disorder that results in dementia, mainly in the elderly. A very characteristic and dominant neuropathological hallmark of this disease is associated with the accumulation of the amyloid-β peptide (Aβ) in the senile plaques [1][2][3]. These plaques are deposited extracellularly, in the grey matter of the brain; however, an aggregation-prone Aβ is produced inside the cells [4][5][6][7].
Aβ is formed by sequential proteolytic cleavage of a large, type 1 transmembrane glycoprotein, the amyloid-β precursor protein (APP), that is ubiquitously expressed in the neuronal and non-neuronal cells (for a review see for example references [8,9]). During its translation, APP is incorporated into the endoplasmic reticulum (ER) membrane, thus entering the constitutive secretory pathway. Along this pathway, APP maturates by being subjected to extensive post-translational modifications that affect its sorting and trafficking (for a review see reference [10]). The immature, N-glycosylated APP is mostly located in the ER. The mature APP, which is additionally O-glycosylated, as well as subjected to other modifications (e.g., tyrosyl-sulfation, palmitoylation, and phosphorylation), is mainly detected in the trans-Golgi network (TGN) or at the plasma membrane. However, different forms of APP are much more diverse, as there are about ten variants of this protein arising from alternative splicing of exons 7 and 8. The length of these isoforms terminally misfolded proteins from productive folding cycles, before targeting them for degradation by ERAD [42][43][44]. Removal of the α1,2 mannose residues from the exposed N-glycans is the signal for ERAD indicating misfolded glycoproteins in mammalian cells (for a review see references [33,45,46]). Interestingly, EDEM1 has a mannosidase activity, as demonstrated in vivo [47][48][49] and in vitro [50]. However, interactions of EDEM1 with ERAD substrates, as well as with ERAD regulators, are complex and multifaceted [51][52][53][54][55][56][57][58]. In our previous studies, we demonstrated that EDEM1 can recognize an untypical ERAD substrate, the ricin-A chain, and promotes its retrotranslocation from the ER to the cytosol [59][60][61]. Importantly, appropriate substrate protein structure, as well as its high hydrophobicity, are relevant for interactions with EDEM1 [62,63].
In this study, we investigated the effects of EDEM1 on APP metabolism and Aβ secretion. We report that EDEM1 regulates cellular levels of APP by promoting transport of the immature form of APP from the ER to the cytosol, where APP is degraded by the proteasome. This results in decreased generation of Aβ 40 and Aβ 42 . Together, our findings provide direct evidence that the APP metabolism is dependent on EDEM1, a key regulator of the ERAD process.

Intracellular Level of APP Is Significantly Reduced in the HEK293 Cells Overexpressing EDEM1
Human embryonic kidney 293 (HEK293) cells produce endogenous forms of APP 695 , APP 751, and APP 770 , as was previously described in [12]. The immature form of APP 695 is poorly detected by Western blotting, whereas it is clearly visible in lysates of cells with overproduction of this protein (Figure 1). The endogenous immature form of APP 751 , together with mature APP 695 and the mature form of APP 751 are expressed at relatively high levels ( Figure 1). Nevertheless, these levels are significantly higher in the APP 751transfected cells than in the cells transfected with control cDNA (Figure 1). It seems very likely that APP must be specifically recognized in the ER before its transport to the cytosol for proteasomal degradation. The ER degradation-enhancing αmannosidase-like protein 1 (EDEM1) has been demonstrated to recognize and extract terminally misfolded proteins from productive folding cycles, before targeting them for degradation by ERAD [42][43][44]. Removal of the α1,2 mannose residues from the exposed Nglycans is the signal for ERAD indicating misfolded glycoproteins in mammalian cells (for a review see references [33,45,46]). Interestingly, EDEM1 has a mannosidase activity, as demonstrated in vivo [47][48][49] and in vitro [50]. However, interactions of EDEM1 with ERAD substrates, as well as with ERAD regulators, are complex and multifaceted [51][52][53][54][55][56][57][58]. In our previous studies, we demonstrated that EDEM1 can recognize an untypical ERAD substrate, the ricin-A chain, and promotes its retrotranslocation from the ER to the cytosol [59][60][61]. Importantly, appropriate substrate protein structure, as well as its high hydrophobicity, are relevant for interactions with EDEM1 [62,63].
In this study, we investigated the effects of EDEM1 on APP metabolism and Aβ secretion. We report that EDEM1 regulates cellular levels of APP by promoting transport of the immature form of APP from the ER to the cytosol, where APP is degraded by the proteasome. This results in decreased generation of Aβ40 and Aβ42. Together, our findings provide direct evidence that the APP metabolism is dependent on EDEM1, a key regulator of the ERAD process.

Intracellular Level of APP Is Significantly Reduced in the HEK293 Cells Overexpressing EDEM1
Human embryonic kidney 293 (HEK293) cells produce endogenous forms of APP695, APP751, and APP770, as was previously described in [12]. The immature form of APP695 is poorly detected by Western blotting, whereas it is clearly visible in lysates of cells with overproduction of this protein (Figure 1). The endogenous immature form of APP751, together with mature APP695 and the mature form of APP751 are expressed at relatively high levels ( Figure 1). Nevertheless, these levels are significantly higher in the APP751-transfected cells than in the cells transfected with control cDNA (Figure 1). Figure 1. HEK293 cells express different APP isoforms. Western blotting of APP and actin in cell lysates of HEK293 transfected with control, APP695, or APP751 cDNAs. The different APP isoforms are indicated by arrows on the left side of the membrane. Immature and mature forms of APP770 were not analyzed in this work and therefore location of these isoforms is not indicated. Molecular mass markers are shown on the right.
To verify whether the intracellular levels of both the endogenous and overproduced forms of APP could be regulated by high EDEM1 production, we examined the levels of APP695 and APP751 in HEK293 cells by ELISA. For this purpose, cells were transfected with EDEM1 cDNA or cotransfected with APP695 and EDEM1 or APP751 and EDEM1 expression vectors (for transfection conditions, see Section 4). The ELISA analysis of total APP in EDEM1-transfected cells shows a decrease in endogenously produced protein by approximately 25%, relative to control cells (Figure 2A), and an approximately 40% decrease in cells overproducing APP695 or APP751 ( Figure 2B,C). It should be noted that in cells with overproduction of APP695 ( Figure 2B) or APP751 ( Figure 2C), total amount of all APP isoforms was analyzed. were not analyzed in this work and therefore location of these isoforms is not indicated. Molecular mass markers are shown on the right.
To verify whether the intracellular levels of both the endogenous and overproduced forms of APP could be regulated by high EDEM1 production, we examined the levels of APP 695 and APP 751 in HEK293 cells by ELISA. For this purpose, cells were transfected with EDEM1 cDNA or cotransfected with APP 695 and EDEM1 or APP 751 and EDEM1 expression vectors (for transfection conditions, see Section 4). The ELISA analysis of total APP in EDEM1-transfected cells shows a decrease in endogenously produced protein by approximately 25%, relative to control cells (Figure 2A), and an approximately 40% decrease in cells overproducing APP 695 or APP 751 ( Figure 2B,C). It should be noted that in cells with overproduction of APP 695 ( Figure 2B) or APP 751 ( Figure 2C), total amount of all APP isoforms was analyzed.

Figure 2.
Total amount of APP is decreased in EDEM1-transfected HEK293 cells. ELISA analysis of (A) endogenously expressed APP, (B) overproduced APP695, or (C) overproduced APP751 in lysates from cells with or without EDEM1 cotransfection. The values are expressed as mean ± SD, n = 3, ** p ˂ 0.01, *** p ˂ 0.001, Student's t test. EDEM1 overexpression and equal β-actin levels were assessed by Western blotting, as indicated in Figure 3. To analyze reduction in the level of individual forms of APP695 and APP751 by EDEM1 overproduction, we assessed the level of APP by Western blotting in lysates from EDEM1transfected HEK293 cells or cells cotransfected with APP695 and EDEM1 or APP751 and EDEM1. As shown in Figure 3, overproduction of EDEM1 significantly reduced the amount of both the immature and mature forms of APP695 and APP751. The most spectacular effect was observed for the immature N-glycosylated form of APP695, where a more than eight-fold decrease in the amount of this form was detected ( Figure 3A). A huge decrease was also found for the overproduced mature form of APP751 ( Figure 3B). The reduction in endogenously expressed forms of APP695 and APP751 was significant ( Figure  3C), but not as great as that for overexpressed APP. This may suggest that EDEM1 is especially involved in the APP metabolism when a large amount of this protein is accumulated in the ER. Probably, regulation of the amount of immature forms of APP695 and APP751 by EDEM1 has a direct effect on the intracellular level of mature forms. However, the EDEM1-related reduction in the level of immature and mature forms of APP695 and APP751 is slightly different ( Figure 3A,B), which may suggest that their intracellular levels may still depend on other distinct regulatory proteins. To analyze reduction in the level of individual forms of APP 695 and APP 751 by EDEM1 overproduction, we assessed the level of APP by Western blotting in lysates from EDEM1transfected HEK293 cells or cells cotransfected with APP 695 and EDEM1 or APP 751 and EDEM1. As shown in Figure 3, overproduction of EDEM1 significantly reduced the amount of both the immature and mature forms of APP 695 and APP 751 . The most spectacular effect was observed for the immature N-glycosylated form of APP 695 , where a more than eight-fold decrease in the amount of this form was detected ( Figure 3A). A huge decrease was also found for the overproduced mature form of APP 751 ( Figure 3B). The reduction in endogenously expressed forms of APP 695 and APP 751 was significant ( Figure 3C), but not as great as that for overexpressed APP. This may suggest that EDEM1 is especially involved in the APP metabolism when a large amount of this protein is accumulated in the ER. Probably, regulation of the amount of immature forms of APP 695 and APP 751 by EDEM1 has a direct effect on the intracellular level of mature forms. However, the EDEM1-related reduction in the level of immature and mature forms of APP 695 and APP 751 is slightly different ( Figure 3A,B), which may suggest that their intracellular levels may still depend on other distinct regulatory proteins.
It should be noted that overproduction of EDEM1 does not affect APP gene transcription (Figure 4), and thus decrease in the amount of APP by EDEM1 overproduction is regulated at the post-transcriptional level.   shown in the graphs. Values obtained for lysates of cells without EDEM1 overexpression are indicated as 1. Values representing EDEM1-transfected cells are plotted relative to 1. The values are expressed as mean ± SD, n ≥ 3, *** p ˂ 0.001, Student's t test.
It should be noted that overproduction of EDEM1 does not affect APP gene transcription (Figure 4), and thus decrease in the amount of APP by EDEM1 overproduction is regulated at the post-transcriptional level.

Overproduction of EDEM1 Regulates the APP Level through Proteasomal Degradation
Since the 26S proteasome is largely responsible for intracellular degradation of APP [39,40,[64][65][66], we studied the effect of a proteasome inhibitor on APP accumulation under EDEM1 overproduction. For this purpose, lactacystin or epoxomicin, selective and irreversible proteasome inhibitors, were applied. As shown in Figure 5, proteasome inhibition significantly abolishes the effect of EDEM1-induced intracellular reduction on the amount of immature and mature forms of APP695 and APP751. This includes both the endogenous and overproduced forms of APP. Interestingly, the intracellular levels of immature overproduced APP695 and APP751 in cells with EDEM1 overproduction which are treated with a proteasome inhibitor almost fully correspond to the amounts of these forms in the cells without EDEM1 overproduction ( Figure 5A,B).

Overproduction of EDEM1 Regulates the APP Level through Proteasomal Degradation
Since the 26S proteasome is largely responsible for intracellular degradation of APP [39,40,[64][65][66], we studied the effect of a proteasome inhibitor on APP accumulation under EDEM1 overproduction. For this purpose, lactacystin or epoxomicin, selective and irreversible proteasome inhibitors, were applied. As shown in Figure 5, proteasome inhibition significantly abolishes the effect of EDEM1-induced intracellular reduction on the amount of immature and mature forms of APP 695 and APP 751 . This includes both the endogenous and overproduced forms of APP. Interestingly, the intracellular levels of immature overproduced APP 695 and APP 751 in cells with EDEM1 overproduction which are treated with a proteasome inhibitor almost fully correspond to the amounts of these forms in the cells without EDEM1 overproduction ( Figure 5A,B).
Thus, it can be concluded that the EDEM1-dependent intracellular reduction in APP level is mainly associated with the proteasomal degradation of this protein in the cytosol. This effect is particularly noticeable in the case of immature, overproduced forms of APP.

APP Retrotranslocation from the ER to the Cytosol Is Regulated by EDEM1 in HEK293 Cells
It has previously been demonstrated that APP is retrotranslocated from the ER to the cytosol in CHO APP751 cells [36]. To assess whether EDEM1 overproduction affects APP 695 retrotranslocation out of the ER in HEK293 cells, we applied a permeablisation assay that we have previously used to study the ricin toxin A-chain retrotranslocation from the ER to the cytosol [59,60,62]. The HEK293 cells overproducing APP 695 or cells cotransfected with APP 695 and EDEM1 were semipermeablised with a mild detergent (digitonin) to obtain separate cytosolic and membrane fractions. It was shown before that treatment of cells with the proteasome inhibitor epoxomicin enables cytosolic accumulation of the immature form of APP 695 , whereas without proteasome inhibition this APP form was not detectable [36]. These findings were confirmed in our studies ( Figure 6A). For cells treated with epoxomicin, our results indicate an approximately two-fold increase in APP 695 retrotranslocation from the ER to the cytosol in EDEM1 overproducing cells when compared to control cells ( Figure 6B). It should be noted that a higher amount of the immature form of APP 695 was observed in the cytosolic fraction and there was a concomitantly reduced level of APP 695 in the membranes of EDEM1 overexpressing cells when compared to the corresponding cytosolic and membrane fractions of cells without EDEM1 overproduction ( Figure 6B). Importantly, accumulation of APP 695 in the cytosol was not due to an uncontrolled leakage of APP out of the ER. To verify this, we examined localization of calreticulin, a soluble ER protein, and calnexin, an ER membrane protein, in cells subjected to permeabilisation. As shown in Figure 6A,B, these proteins were not released from the ER into the cytosol, indicating that the ER membranes remained intact. Thus, it can be concluded that high production of EDEM1 promotes retrotranslocation of the immature form of APP 695 to the cytosol.

Downregulation of EDEM1 Causes a Significant Elevation in APP Amounts in the HEK293 and SH-SY5Y Cells
To analyze the relationship between EDEM1 and APP in more detail, we first examined the levels of endogenous forms of APP in HEK293 cells with reduced intracellular amounts of EDEM1. For this purpose, cells were treated with a heterogeneous mixture of siRNA, esiRNA specific for EDEM1 (esiEDEM1), or a control unspecific esiRNA (silencing GFP, esiGFP). Alternatively, cells were transfected with short hairpin RNA, shRNA against EDEM1 (shEDEM1) [59]. Our previous experiments showed that both esiEDEM1 and shEDEM1 effectively downregulate the expression of EDEM1 [59,60]. Additionally, in each experiment, EDEM1 mRNA levels were quantified by qRT-PCR and were estimated relative to GAPDH mRNA, or protein levels were analyzed by Western blotting ( Figure 7A). All analyzed data come from experiments in which highly efficient downregulation (90-95%) of EDEM1 was achieved. A reduced level of EDEM1 causes a significant elevation in intracellular levels of APP, both in the immature and mature forms ( Figure 7B). Interestingly, the level of the endogenous immature form of APP 695 increased approximately 20-fold in cells with downregulated EDEM1, reflecting the levels obtained in cells with overproduction of APP 695 (compare with Figure 1).
To verify the generality of APP regulation by the EDEM1 chaperone protein, we downregulated EDEM1 in SH-SY5Y neuroblastoma cells. Endogenous APP has been shown to be expressed at relatively high levels in these cells [12]. For EDEM1 level reduction, cells were treated with esiEDEM1 or control esiGFP. We estimated that, as in HEK293 cells, a concentration of 40 nM esiEDEM1 caused an approximately 95% reduction in EDEM1 mRNA levels, as quantified by qRT-PCR and analyzed relatively to GAPDH mRNA ( Figure 7C). Intracellular levels of both the immature and mature forms of APP were significantly increased in SH-SY5Y cells after EDEM1 downregulation ( Figure 7D). Similar to HEK293 cells, an approximately 20-fold elevation in the amount of the immature form of APP 695 was observed. These observations suggest that EDEM1 is a significant regulator of intracellular APP protein levels.

EDEM1 Colocalizes and Interacts with APP
We first used confocal immunofluorescence imaging to examine the colocalization of EDEM1 and APP. Immunostaining of HEK293 cells cotransfected with APP 695 and EDEM1 showed colocalization of both proteins (Figure 8), with the degree of EDEM1 colocalization with APP 695 equal to 0.87 (Mander's overlap coefficient, mean ± SD, n = 5, *** p < 0.001). As was previously reported in [59], EDEM1 is localized in the ER. This protein was found to colocalize with the ER protein disulfide isomerase (PDI) (Supplementary Materials Figure S1), with the degree of EDEM1 colocalization with PDI equal to 0.7 (n = 3, *** p < 0.001). It can be concluded that most of the EDEM1 that is localized in the ER colocalizes with APP 695 ; however, the degree of APP 695 colocalization with EDEM1 is decreased by 33% (n = 5, *** p < 0.001) in comparison to EDEM1 colocalization with APP 695 . This is due to the fact that intacellular APP 695 does not entirely localize in the ER.  ture form of APP695 was observed. These observations suggest that EDEM1 is a significant regulator of intracellular APP protein levels.

EDEM1 Colocalizes and Interacts with APP
We first used confocal immunofluorescence imaging to examine the colocalization of EDEM1 and APP. Immunostaining of HEK293 cells cotransfected with APP695 and EDEM1 showed colocalization of both proteins (Figure 8), with the degree of EDEM1 colocalization with APP695 equal to 0.87 (Mander's overlap coefficient, mean ± SD, n = 5, *** p ˂ 0.001). As was previously reported in [59], EDEM1 is localized in the ER. This protein was found to colocalize with the ER protein disulfide isomerase (PDI) (Supplementary Materials Figure S1), with the degree of EDEM1 colocalization with PDI equal to 0.7 (n = 3, *** p ˂ 0.001). It can be concluded that most of the EDEM1 that is localized in the ER colocalizes with APP695; however, the degree of APP695 colocalization with EDEM1 is decreased by 33% (n = 5, *** p ˂ 0.001) in comparison to EDEM1 colocalization with APP695. This is due to the fact that intacellular APP695 does not entirely localize in the ER. To further investigate possible interactions between EDEM1 and APP695, and to confirm that EDEM1 is directly involved in APP transport to the cytosol, we applied a coimmunoprecipitation assay. A band corresponding to APP was detected in precipitates from cells cotransfected with APP695 and EDEM1, but not in precipitates from cells transfected only with APP695, EDEM1, or a control cDNA (Figure 9). This indicates the specificity of the observed interactions between APP and EDEM1. To further investigate possible interactions between EDEM1 and APP 695 , and to confirm that EDEM1 is directly involved in APP transport to the cytosol, we applied a coimmunoprecipitation assay. A band corresponding to APP was detected in precipitates from cells cotransfected with APP 695 and EDEM1, but not in precipitates from cells transfected only with APP 695 , EDEM1, or a control cDNA (Figure 9). This indicates the specificity of the observed interactions between APP and EDEM1.

Intracellular Level of EDEM1 Influences Aβ Secretion
We next investigated the effect of EDEM1 overexpression on Aβ40 and Aβ42 secretion in HEK293 cells cotransfected with APPSwe/Ind and EDEM1 or APPdeltaCT and EDEM1. The Swedish mutation of the APP protein (APPSwe) is known to enhance the abnormal cleavage of cellular APP by β-secretase, resulting in an increase in the total Aβ production and secretion, specifically of the Aβ40 and Aβ42 forms [67,68]. On the other hand, the APP Indiana mutation (APPInd) increases the Aβ42/Aβ40 ratio in conditioned media [69,70]. Finally, in cells transfected with an APP construct with a deletion of the C-terminal intracellular domain of APP (APPdeltaCT) [71], the detected Aβ is entirely derived from the secretory pathway. Since the intracellular domain of APP is necessary for proper endocytosis of APP, production of Aβ that comes from the endocytic pathway is blocked in cells transfected with APPdeltaCT [72,73].

Intracellular Level of EDEM1 Influences Aβ Secretion
We next investigated the effect of EDEM1 overexpression on Aβ 40 and Aβ 42 secretion in HEK293 cells cotransfected with APP Swe/Ind and EDEM1 or APP deltaCT and EDEM1. The Swedish mutation of the APP protein (APP Swe ) is known to enhance the abnormal cleavage of cellular APP by β-secretase, resulting in an increase in the total Aβ production and secretion, specifically of the Aβ 40 and Aβ 42 forms [67,68]. On the other hand, the APP Indiana mutation (APP Ind ) increases the Aβ 42 /Aβ 40 ratio in conditioned media [69,70]. Finally, in cells transfected with an APP construct with a deletion of the C-terminal intracellular domain of APP (APP deltaCT ) [71], the detected Aβ is entirely derived from the secretory pathway. Since the intracellular domain of APP is necessary for proper endocytosis of APP, production of Aβ that comes from the endocytic pathway is blocked in cells transfected with APP deltaCT [72,73].
We aimed to determine whether reduction in intracellular APP levels induced by EDEM1 overproduction leads to a decrease in Aβ secretion by analyzing the medium obtained from cultures of HEK293 cells with or without overexpression of EDEM1. In cells transfected with the examined APP constructs, EDEM1 overexpression reduced the amounts of secreted Aβ 40 and Aβ 42 ( Figure 10). We observed an approximately 25% reduction in the level of Aβ 40 and a 16% reduction in Aβ 42 in APP Swe/Ind ( Figure 10A,B), and an approximately 20% reduction in the amount of Aβ 42 in APP deltaCT ( Figure 10C). These results indicate that a reduction in APP levels in the cell has a significant effect on Aβ forms production. Moreover, this effect is closely related to intracellular Aβ production, as indicated by data obtained for cells transfected with the APP delta CT construct.

Discussion
Several studies have indicated that the early stages of APP maturation in the ER largely determine the metabolism of this protein. APP can enter different processing pathways, among which the degradation paths significantly affect the fate of APP, including Aβ production. Besides degradation by the ubiquitin-proteasome pathway [39,64,66], APP can be also metabolized in lysosomes [74] and by the stress-responsive chaperoneprotease HtrA2 [36]. It has been suggested that the proteasome is involved in the degradation of misfolded forms of APP that lack the correct structure due to blocked N-glycosylation or due to the action of other factors affecting APP maturation in the ER [66]. Before degradation in the cytosol, APP retrotranslocates out of the ER as the ERAD substrate. EDEM1 is one of the main regulators of ERAD involved in the recognition of aberrant proteins in the ER [43,44].
We have demonstrated here that EDEM1 interacts with APP and significantly regulates the intracellular levels of immature forms of APP695 and APP751, which also affects the amounts of mature forms of APP. Some discrepancies between fold reduction of the immature APP695 and APP751 in the EDEM1-transfected cells (more than 8-fold reduction for APP695 and 4.5-fold reduction for APP751) in comparison to immature APP levels in the corresponding control cells may be due to the fact that immature APP751 was analyzed together with mature APP695 and/or due to possible different interaction partners of APP695 and APP751 within the ER that alter their regulatory mechanisms. It has been demonstrated

Discussion
Several studies have indicated that the early stages of APP maturation in the ER largely determine the metabolism of this protein. APP can enter different processing pathways, among which the degradation paths significantly affect the fate of APP, including Aβ production. Besides degradation by the ubiquitin-proteasome pathway [39,64,66], APP can be also metabolized in lysosomes [74] and by the stress-responsive chaperone-protease HtrA2 [36]. It has been suggested that the proteasome is involved in the degradation of misfolded forms of APP that lack the correct structure due to blocked N-glycosylation or due to the action of other factors affecting APP maturation in the ER [66]. Before degradation in the cytosol, APP retrotranslocates out of the ER as the ERAD substrate. EDEM1 is one of the main regulators of ERAD involved in the recognition of aberrant proteins in the ER [43,44].
We have demonstrated here that EDEM1 interacts with APP and significantly regulates the intracellular levels of immature forms of APP 695 and APP 751 , which also affects the amounts of mature forms of APP. Some discrepancies between fold reduction of the immature APP 695 and APP 751 in the EDEM1-transfected cells (more than 8-fold reduction for APP 695 and 4.5-fold reduction for APP 751 ) in comparison to immature APP levels in the corresponding control cells may be due to the fact that immature APP 751 was analyzed together with mature APP 695 and/or due to possible different interaction partners of APP 695 and APP 751 within the ER that alter their regulatory mechanisms. It has been demonstrated that the immature form of APP 695 interacts with ER proteins to form large complexes [30]. It cannot be excluded that immature APP 695 and APP 751 are involved in interactions with different ER proteins. However, a proteomics-based approach did not identify any ER proteins specific to either of these two isoforms [75]. Anyway, it is important that the exact mechanism of APP 695 and APP 751 recognition by ER regulatory proteins, especially the ERAD process proteins, is fully understood and future studies should aid in resolving this issue.
The unique effect of EDEM1 on the immature form of APP 695 was also confirmed in cells with downregulated EDEM1 gene expression, where we observed a dramatic increase in the amount of this form. This effect was observed not only in the non-neuronal kidney cells, but also in human neuroblastoma cells. These results indicate that depletion of EDEM1 can significantly alter the metabolism of endogenously produced APP.
EDEM1-induced reduction in overproduced APP levels was dramatically restored by pretreatment with specific proteasome inhibitors, lactacystin or epoxomycin. These results suggest that a decrease in APP levels by EDEM1 is associated with APP transport to the cytosol and subsequent proteasomal degradation of this protein. Indeed, our permeabilisation assay revealed increased transport of the immature form of APP from the ER to the cytosol in EDEM1-transfected cells. It has been previously reported that immature APP 751 and the N-terminally truncated form of APP are retrotranslocated to the cytosol [36]. Both of these forms of APP can be degraded by the proteasome and a chaperone protease HtrA2, which is located in the cytosolic side of the ER membrane. The EDEM1-dependent reduction of endogenous levels of APP 695 and APP 751 was not completely restored by the proteasome inhibitor, suggesting that probably another degradation system, in addition to the proteasome, is important in these processes. On the other hand, the results of experiments performed in cells with APP overproduction suggest that in cells with elevated APP levels, this protein is mainly degraded by the proteasome.
In our permeabilisation assays, we used a proteasome inhibitor, which allows us to analyze the amount of immature APP 695 present in the cytosolic fraction. In cells with a functional proteasome, the entire pool of cytosolic APP is degraded very rapidly, as demonstrated previously in [36] and also confirmed in this study. It should be noted that the observed amounts of immature APP present in the cytosol do not reflect the total amount of APP retrotranslocated out of the ER. It was demonstrated that the proteasome activity is required for retrotranslocation for the movement through the retrotranslocation channel, for the ERAD substrate dissociation of the ER protein regulatory complex, or for both steps [76,77]. It is known that proteasome inhibition causes APP retention in the ER under ER stress [64]. Additionally, our mild digitonin treatment released only 20-25% of the cytosolic fraction [59]. A low concentration of digitonin was used to avoid leakage of ER proteins to the cytosol.
It has been reported that the CMV promoter can be significantly upregulated in cells treated with proteasome inhibitors for longer than 8 h, yielding dramatic increases in mRNA and protein levels after 24 h of incubation with proteasome inhibitors [78,79]. In our work, we used EDEM1 cDNA fused to an HA-tag in the pCMV-SPORT2 vector [42]. There were no differences in the cellular EDEM1 levels between cells untreated with proteasome inhibitors and cells treated with proteasome inhibitors for 5 h.
Accumulating observations indicate that ER stress plays an important role in the etiology and pathogenesis of AD [80][81][82]. This was demonstrated in studies using cellular and animal models of this disease, as well as in human postmortem brains. However, the ER-related pathological features of AD appear to be complex and multifaceted. The unfolded protein response (UPR) is triggered in cells subjected to ER stress (for a review see reference [82]). During this response, the chaperone protein BiP/GRP78 is released from the three UPR sensors, leading to their activation. Activation of the first receptor, RNA-dependent protein kinase-like ER kinase (PERK), causes phosphorylation of the translation factor eIF2α, which inhibits protein synthesis. The second effector of this response, transcription factor-6 (ATF6), is responsible for an increase in the amount of ER chaperone proteins, including BiP and calreticulin, which accelerates the folding capacity of the ER. It is known that the immature form of APP 751 interacts with BiP in the ER [29] and that this interaction is elevated upon ER stress [80]. Under ER stress, BiP retains APP in the ER, prevents its transport to distal compartments, where amyloidogenic proteolysis takes place, which overall directly leads to reduced Aβ production [29,80]. Furthermore, both immature and mature forms of APP 695 can bind to calreticulin, forming larger complexes that may contain other ER chaperones [30]. However, it should be noted that the nature of calreticulin interaction with immature and mature APP may differ significantly. It has been shown that the mature form of APP can interact with calreticulin to form a complex on the cell surface or alternately reach the cell surface as a complex [31]. Moreover, calreticulin regulates Aβ production as it interacts with sequences at the γ-cleavage site of APP and binds to presenilin and nicastrin to modulate the γ-secretase activity. These observations suggest that overproduction of BiP and calreticulin may promote different mechanisms leading to reduced Aβ generation. This mechanism is even more complex if we consider the action of EDEM1. EDEM1 production during UPR depends on activation of the inositolrequiring protein 1α (IRE1) [49]. IRE1 controls specific splicing of the transcription factor XBP1, which in turn induces the expression of factors facilitating ERAD. This is the third and final stage of the UPR. It can be assumed that during the ER stress, a high level of EDEM1 promotes the transport of the immature form of APP to the cytosol and its further proteasomal degradation. Taking all of these factors into consideration, it can be suggested that the depletion of APP as a substrate in the amyloidogenic pathway during ER stress may be related to its increased interaction with BiP and possibly calreticulin, and to transient retention of APP in the ER [29,80], but also to increased expression of EDEM1, resulting in the final accelerated transport of APP to the cytosol and its increased proteasomal degradation. This is consistent with our experiments performed with proteasome inhibitors and with the observation that APP is preferentially degraded by proteasome during acute ER stress [64]. In a study using a Drosophila model of chronic ER proteinopathy in the brain, activation of the IRE1/XBP1 pathway was shown to significantly reduce the Aβ 42 levels [83]. Moreover, overexpression of either dEDEM1 or dEDEM2 in neurons had decreased Aβ 42 levels in the brain without significant effects on the Aβ 42 mRNA levels. XBP1 negatively regulates expression and activity of β-secretase, the rate-limiting enzyme involved in the production of the Aβ peptide, by promoting activity of the ubiquitin-ligase HRD1 [84]. Several studies indicate that XBP1 protects against Aβ toxicity and ameliorates pathology connected with AD [85]. Based on these results, it can be concluded that properly functioning ERAD which attenuates ER stress can lower the toxic Aβ generation in the cell.
However, the capacity and efficiency of ERAD are decreased under pathological conditions caused by chronic ER stress [83]. The continuous accumulation of Aβ and Tau proteins can cause ERAD deregulation and chronic ER stress [83,86]. Furthermore, it was demonstrated that despite activation of the UPR, ERAD activity can be reduced in aged tissues [83]. The mechanisms of UPR signaling and ERAD activation appear to be a complex scenario, and different and even initially conflicting effects have been observed in AD models. It was demonstrated that IRE1 activation in the brain, measured as IRE1 phosphorylation, unexpectedly positively correlates with the severity of AD [81]. In this case, targeting IRE1 signaling has significantly reduced Aβ deposition. It cannot be excluded that altered neuronal proteostasis during AD progression triggers abnormal levels of ER stress and activates the damaging UPR signaling processes mediated in part by the IRE1/XBP1 pathway. Indeed, it was shown that long-term activation of IRE1/XBP1 was detrimental in fly neurons [83]. All of this exacerbates AD. It is possible that under chronic ER stress, the level of EDEM1 is reduced which may result in the abnormal production of APP that is targeted to the amyloidogenic pathway. Furthermore, chronic ER stress inhibits the proteasome [87], which may additionally contribute to the intracellular accumulation of APP and increased production of toxic Aβ.
In conclusion, we found that EDEM1 has a significant effect on regulation of the APP 695 and APP 751 metabolism, and production of the Aβ forms. EDEM1 interacts with APP and promotes its retrotranslocation from the ER to the cytosol. This may suggest that this crucial ERAD regulator has an important role in the pathogenesis of AD. Although it was reported previously that overexpression of Drosophila EDEMs did not activate canonical UPR (including chronic UPR activation), interestingly it suppressed behavioral deficits and neurodegeneration in Aβ 42 flies [83]. Thus, selective activation of human EDEM1 may be expected to have a potential therapeutic role in AD.

Total RNA Isolation and Quantitative Real-Time RT-PCR
Total RNA was isolated from HEK293 or SH-SY5Y cells transfected with esiEDEM1 or shEDEM1 and from HEK293 cell transfected with EDEM1-HA by RNA Extracol (Eurx, Gdańsk, Polska), following the protocol provided by the manufacturer. Total RNA was then treated with 2 units/per 10 µg RNA of TURBO DNase (RNase-Free, 2 U/µL, Thermo Fisher Scientific, Waltham, MA, USA) at 37 • C for 30 min, according to the manufacturer's procedure. The TURBO DNase was inactivated at 75 • C for 10 min. The overall quality of RNA preparations was assessed by electrophoresis on a denaturing agarose gel. RNA was quantified using Quant-it RiboGreen RNA Assay Kit (Thermo Fisher Scientific, Waltham, MA, USA) according to the protocol provided by the manufacturer. RNA integrity, purity, concentration, and size were also analyzed by the Agilent 2100 Bioanalyzer (Agilent, Palo Alto, CA, USA) using the RNA 6000 Nano assay (Agilent), according to the manufacturer's instructions and using the Agilent 2100 Bioanalyzer Software (Agilent, Palo Alto, CA, USA). The highest quality RNA, with visibly distinct 18S and 28S ribosomal RNA bands, not degraded and with a RNA Integrity Number (RIN) above 8.0 was used for real-time qRT-PCR. Then, 500 ng of total RNA were reverse-transcribed into cDNA using the Transcriptor First Strand cDNA Sythesis Kit (Roche Life Science, Basel, Switzerland) according to the manufacturer's instructions.
The real-time PCR reactions were performed using the Roche LightCycler TaqMan Master mix in combination with Roche Universal Probe Library (UPL) assays. All assays were designed to span an intron-exon boundary to prevent amplification of DNA. Each reaction consisted of 1 × LightCycler TaqMan master mix, specific primer pairs and fluorescently-tagged probes, both for the reference and the examined gene. The expression of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) mRNA was used as reference control. The primers and fluorescently-tagged probe for the reference gene were used according to the protocol provided by the manufacturer (Roche Life Science, Basel, Switzerland). It has been evaluated previously that down-regulation of EDEM1 or overexpression of EDEM1-HA does not influence the expression of the reference gene [60]. The real-time PCR reaction was performed in the LightCycler 480 System (Roche Diagnostic, Basel, Switzerland) under the following conditions: initial denaturation, incubation at 95 • C for 10 m followed by: 45 cycles of 95 • C denaturation for 10 s, 60 • C annealing for 30 s, and extension at 72 • C for 1 s. Quantification of mRNA expression was carried out using the LightCycler detection system. Real time PCR efficiency was calculated for each reaction, and only results from reactions whose efficiencies were within the range of 90-110% (corresponding to a 1.8-2.2-fold increase per cycle) were exclusively taken for further analysis. The efficiency correction was accounted for in mRNA quantification. The 2 −∆∆ct method was used to determine the relative gene transcript levels after normalization to the reference genes [88].

Cell Lysis and Western Blotting
HEK293 or SH-SY5Y cells were seeded in 6 cm plates (1 × 10 6 /plate) or 6-well plates (5 × 10 5 /well) and transfected with the appropriate cDNAs or esiEDEM1. Then 72 h post transfection, cells were washed with ice cold phosphate-buffered saline (PBS) and lysed in a buffer (0.1 M NaCl, 10 mM Na 2 HPO 4 , 1 mM EDTA, 1% Triton X-100, pH 7.4) supplemented with protease inhibitor mixture (Roche Life Sciences, Basel, Switzerland). Alternatively, before lysis the cells were incubated for 5 h at 37 • C with epoxomicin (1 µM) or lactacystin (10 µm). Lysates were centrifuged to remove cell debris and nuclei for 10 min at 10,000× g. Samples were resolved by reducing SDS/PAGE (12% gels). The proteins were transferred onto an Immobilon-FL membrane (Merck, Darmstadt, Germany) by the Trans-Blot Turbo Transfer System (Bio-Rad, Hercules, CA, USA). Membranes were then incubated with appropriate primary and secondary antibodies. Proteins were detected by chemiluminescence with the Clarity Max ECL Western Blotting Substrate (Bio-Rad, Hercules, CA, USA) or the SuperSignal West Femto Maximum Sensitivity Substrate (Thermo Fisher Scientific, Waltham, MA, USA) and were visualized using Typhoon FLA 9400 (GE Healthcare, Chicago, IL, USA). Signal intensities of the bands were quantified using Image Studio Lite (v.5.2) (LI-COR Bioscences, Lincoln, NE, USA).

Retrotranslocation Assay
HEK293 cells were seeded in 6 cm plates (1 × 10 6 /plate) and cotransfected with APP 695 and EDEM1-HA cDNAs or APP 695 and control cDNAs. Then 72 h post transfection, the cells were incubated for 5 h at 37 • C with epoxomicin (1 µM) and washed with room temperature PBS. For permeabilisation, the cells were incubated for 5 min at room temperature with KOAc buffer (115 mM CH3COOK, 25 mM HEPES, and 2.5 mM MgCl 2 , pH 7.4) containing 3 µg/mL digitonin, followed by a 30-min incubation on ice. After incubation, the buffer containing cytosolic fraction was collected, and the remaining membranes were lysed as indicated above. Both the cytosolic and membrane fractions were centrifuged for 10 min at 10,000× g. APP was immunoprecipitated from both fractions with mouse anti-APP antibodies immobilized on Dynabeads protein G (Thermo Fisher Scientific). Finally, the beads were washed with ice-cold PBS supplemented with 0.35% Triton X-100, and the adsorbed material was resolved by SDS/PAGE (12% gels) under reducing conditions. For the detection of APP by Western blotting, the proteins were transferred onto an Immun-Blot Low Fluorescence PVDF membrane (Bio-Rad, Hercules, CA, USA). In addition, Western blotting was performed for the cytosolic and membrane fractions with anti-calreticulin, anti-calexin, and anti-Hsp70 antibodies and the corresponding secondary antibodies.

Co-Immunoprecipitation Assay
HEK293 cells were seeded in 6 cm plates (1 × 10 6 /plate) and cotransfected with APP 695 and EDEM1-HA, APP 695 and control, EDEM1-HA and control, or control cDNAs. Then 72 h post transfection, the cells were washed with cold PBS and lysed as indicated above. Samples were then sonicated (5 min, 40% output). The lysates were centrifuged to remove cell debris and nuclei at 10,000× g. The supernatants were immunoprecipitated for 1 h at 4 • C using mouse anti-APP antibodies coupled to protein G Dynabeads (Thermo Fisher Scientific, Waltham, MA, USA). The beads were washed 3 times with HBS buffer (pH 6.8) containing 0.1% Tween 20. Samples were resolved by 12% reducing SDS/PAGE and transferred onto a PVDF membrane (Merck, Darmstadt, Germany). For the Western blotting detection of EDEM1 interacting with APP, membranes were treated with anti-HA antibodies. For the control of APP 695 and EDEM1 transfection, whole cell lysates were subjected to Western blotting with anti-APP or anti-HA antibodies.

Immunofluorescence Microscopy
HEK293 cells cotransfected with APP 695 and EDEM1-HA or cells transfected with EDEM1-HA were grown on coverslips. Cells were then washed once with PBS and fixed in 2% (w/v) paraformaldehyde (PFA, Merck, Darmstadt, Germany). Cells were then permeabilised in 0.1% Triton X-100 and blocked in 5% FBS before labelling with mouse anti-HA together with rabbit anti-APP or with mouse anti-HA together with rabbit anti-PDI and treated with appropriate secondary antibodies. DAPI (Thermo Fisher Scientific, Waltham, MA, USA)) was used to stain the nuclei. The cells were mounted in Mowiol (Molecular Probes, Eugene, OR, USA) and examined with a laser scanning confocal microscope LSM800 Zeiss with Airyscan (Carl Zeiss, Jena, Germany) or with an automated inverted microscope Leica DMI4000B (Leica, Wetzlar, Germany). Images were prepared with the ZEISS ZEN Microscope Software (Carl Zeiss) or the Leica Application Suite 3.1.0 (Leica) and were analyzed by the Fiji plugin in the ImageJ software. Mander's coefficient was used for reporting co-localization between EDEM1 and APP, APP and EDEM1, or EDEM1 and PDI. Mander's coefficient ranges from 0 to 1, corresponding to non-overlapping images and 100% co-localization between the two images, respectively.

Analysis of APP and Aβ Forms by ELISA
APP, Aβ 40, and Aβ 42 levels were quantified using commercially available, standard ELISA kits (Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer's protocol. For APP detection, HEK293 lysates of cells transfected with EDEM1-HA or control cDNAs, or cotransfected with APP 695 and EDEM1-HA or APP 695 and control were used. For Aβ detection, conditioned media coming from cells cotransfected with APP 695Swe/Ind and EDEM1-HA or APP 695Swe/Ind and control cDNAs, as well as from cells cotransfected with APP deltaCT and EDEM1-HA or APP deltaCT and control cDNAs were applied. To prevent degradation of APP or Aβ, protease inhibitors (Roche Life Sciences, Basel, Switzerland) were added. The lysates or the conditioned media were loaded onto plates coated with an appropriate APP or Aβ-specific antibodies and treated with corresponding detector antibodies. ELISA plates were developed using a color reaction, and the absorbance was read at 450 nm using the Victor 3 plate reader (Perkin Elmer, Waltham, MA, USA).

Statistics
All experiments were performed independently at least three times. Values are expressed as mean ± SD. Statistical analyses were performed by Student's t test. A p value of 0.05 or less was considered to be statistically significant. p-values of * p < 0.05, ** p < 0.01, *** p < 0.001.