AR12 increases BAG3 expression which is essential for Tau and APP degradation via LC3-associated phagocytosis and macroautophagy

Introduction

AR12 (OSU-03012) is a derivative of the anti-inflammatory agent celecoxib; unlike the parent compound AR12 lacks COX2 inhibitory activity. Our group demonstrated that AR12 reduced the expression of many chaperone proteins and caused endoplasmic reticulum stress signaling with repression of protein translation, i.e., increased eIF2α S51 phosphorylation, and rapidly caused autophagosome formation, followed later by autolysosome formation, i.e., autophagic flux [1–10]. We demonstrated that the key cellular target for AR12 was the ER stress-regulatory chaperone GRP78 (aka BiP, HSPA5), followed by other ATP-dependent chaperones in the HSP90/HSP70 families. AR12 was shown to directly inhibit chaperone ATPase activities. In vivo studies using AR12 in mice, rats and rabbits has shown drug efficacy without damage to normal tissues [8, 11].

In our recent studies in Alzheimer’s Disease, using wild type and genetically modified HCT116 colon cancer cells as a model system expressing either ATG16L1 T300 or ATG16L1 A300, we determined whether drugs that directly inhibit the chaperone ATPase activity or cause chaperone degradation and endoplasmic reticulum stress signaling leading to macroautophagy could reduce the levels of proteins which play a pathogenic role in neurodegenerative diseases [10]. AR12 and the breast cancer drug neratinib (NER) rapidly reduced expression of Tau, amyloid precursor protein (APP), superoxide dismutase 1 (SOD1) and TAR DNA-binding protein 43 (TDP-43) [10, 12–14].

GRP78 is expressed in the ER and on the outer leaflet of the plasma membrane [15–20]. In the ER it prevents ER stress signaling by PERK, IRE1 and ATF6 and acts as a molecular chaperone renaturing misfolded proteins. GRP78 can chaperone Tau and APP, and can prevent, in an ATP-independent fashion, the processing of APP to insoluble amyloid-β [21]. Cell surface GRP78 plays a key role in permitting neurons and microglia to uptake Tau and APP, i.e., GRP78 is a key player in facilitating the ‘prion’-like bystander behavior of Tau [21]. In that regard, GRP78 has already been shown to modulate prion propagation [19]. At first glance, approaches that would enhance GRP78 expression, rather than inhibiting its ATPase, could be considered as an AD therapeutic approach. This would prevent protein denaturation and processing of Tau and APP into tangles. However, over-expression of GRP78 also blocks both ER stress signaling and the ability of cells to perform macroautophagy and autophagic flux. i.e., over-expression of GRP78 can stabilize Tau and APP, but it does so at the cost of preventing their degradation. Over-expression of GRP78 will also increase the amount of Tau and APP being taken up by bystander neurons and microglia.

We know that AR12 and neratinib have overlapping and separate biologies which facilitate autophagosome formation and autophagic flux resulting in the degradation of proteins. We hypothesize that AR12, by inhibiting GRP78, leads to increased ER stress signaling and a reduced capacity to signal into the PI3K pathway, resulting in the inactivation of mTOR, with subsequent autophagosome formation and degradation of Tau and APP. Neratinib through reactive oxygen species, activates ataxia telangiectasia mutated (ATM) which phosphorylates and activates the AMP-dependent protein kinase (AMPK). AMPK, by acting to reduce mTOR activity and by directly activating ULK1, also results in autophagosome formation and degradation of Tau and APP. Furthermore, because neratinib can cause degradation of HER2, which is over-expressed in the brains of AD patients, PI3K/mTOR signaling will be further reduced [12–14, 22, 23]. The present studies were performed to define the biology of AR12 and neratinib in macrophages, microglia, and neuronal cells and whether AR12, alone or combined with neratinib, was competent to cause degradation of APP and Tau proteins in these cell types.

Materials and Methods

Results

The chaperone GRP78 acts both as a chaperone to renature proteins but also plays a pivotal role in the abilities of cells to sense endoplasmic reticulum (ER) stress [15]. GRP78 is located both in the ER and on the outer leaflet of the plasma membrane. In the ER, GRP78 inhibits PKR-like endoplasmic reticulum kinase (PERK), which phosphorylates and inactivates eIF2α on S51. On the cell surface GRP78 plays roles in stabilizing plasma membrane receptors and more recently was shown to play an essential role as a co-receptor for the virus SARS-CoV-2 [9, 16]. We have previously shown that neratinib, via Rubicon-dependent LC3-associated phagocytosis (LAP), caused the internalization and subsequent macroautophagic degradation of growth factor receptors and RAS proteins [12–14]. In microglia, the uptake and degradation of APP has also been linked to LC3-associated endocytosis (LANDO) [24, 25]. Our present studies were designed to determine whether LAP / LANDO played a mechanistic role in the abilities of AR12 and neratinib to cause Tau, APP, and chaperone degradation in neurons and microglia. In HCN2 human neuronal cells and BV2 murine microglial cells, knock down of the essential LAP regulatory protein Rubicon suppressed autophagosome formation though did not significantly alter autophagic flux, i.e., vesicles that were initially GFP+ RFP+ became over time only RFP+ (Figure 2). Knock down of the macroautophagy proteins Beclin1 or ATG5 also abolished autophagosome formation and autophagic flux (Figure 3) [10].

Based on the data in Figures 2, 3, HCN2 neuronal and BV2 microglial cells were transfected with plasmids to express Tau or APP, and co-transfected with siRNA molecules to knock down the expression of Rubicon, Beclin1 or ATG5. AR12 and neratinib reduced the expression of Tau and APP in HCN2 cells, that was blocked by knock down of Rubicon, Beclin1 or ATG5 (Figure 4A). Knock down of Rubicon, Beclin1 or ATG5 did not significantly alter the basal expression levels of Tau or APP (not shown). Knock down of Rubicon, Beclin1 or ATG5 blocked the degradation of Tau and APP in BV2 microglia (Figure 4B). ompared to the amount of APP expressed from a transfected plasmid (100%), endogenous APP expression in the HCN2 cells was only 5%. For Tau expression, it was 6%. AR12 and neratinib, to a significantly greater extent than observed when expressing Tau or APP from plasmids, reduced endogenous Tau and APP levels in HCN2 cells (Figure 4C). To confirm our Rubicon siRNA data, we made use of RAW macrophages that had been genetically deleted for Rubicon. In wild type RAW macrophages, AR12 and neratinib reduced the expression of chaperones, Tau, and APP, and increased eIF2α S51 phosphorylation (Figure 4D). Deletion of Rubicon in the macrophages abolished the degradation of all tested proteins and the increase in eIF2α S51 phosphorylation (Supplementary Figure 1).

We next determined in HCN2 neuronal cells and in BV2 microglia the abilities of AR12 and neratinib to reduce the expression of mutant forms of APP and Tau [26–29]. AR12 and the drug combination reduced the expression of Tau 301L which trended to be less than the reduction of wild type Tau (Figure 5). AR12 and the drug combination was equipotent at reducing the expression of APP, APP715 and APP 692. These findings are important for future in vivo studies as, for example, the Tau P301L mutant is used in transgenic models of Alzheimer’s Disease.

We hypothesized that expression of Tau or APP may alter the behavior of intracellular signaling pathways when cells are treated with AR12 and neratinib. HCN2 cells were treated with AR12 and neratinib for 6h, after which alterations in protein expression and protein phosphorylation were determined. As we have observed in other cell types, neratinib reduced both the expression and the phosphorylation of the plasma membrane receptors ERBB1/2/3 (Figure 6). Neither expression of Tau nor expression of APP significantly altered the levels of drug-induced protein degradation or protein phosphorylation when compared to empty vector transfected cells. Notably, compared to tumor cell types we have previously treated with neratinib as a single agent, we observed a profound increase in the phosphorylation of ULK1 S317 and profound reductions in the phosphorylation of ULK1 S757, mTORC1 S2448, mTORC2 S2481 and p70 S6K T389. Increased ULK1 S317 phosphorylation concomitant with lower ULK1 S757 phosphorylation results in a very high level of ULK1 catalytic activity which drives autophagosome formation. Knock down of AMPKα prevented the alterations in protein phosphorylation observed in mTOR and ULK1 (Figure 7A). Expression of an activated mTOR protein suppressed autophagosome formation and the degradation of APP and Tau (not shown). The most surprising data was that AR12 and neratinib combined to not only reduce p70 S6K T389 phosphorylation but also to reduce p70 S6K protein levels. Knock down of the macroautophagy regulatory proteins Beclin1 or ATG5 prevented p70 S6K degradation (Figure 7B). As p70 S6K signaling has been linked to enhanced Tau phosphorylation, we hypothesize that the portions of p70 S6K complexed with Tau were being degraded by macroautophagy in our cells.

A key AR12 target are the ATP binding domains of HSP90 and HSP70 family chaperone proteins [3–10]. Both AR12 and neratinib can also reduce the protein levels of chaperone proteins by stimulating autophagy. In HCN2 neuronal and BV2 microglial cells, AR12 reduced the expression of GRP78 (cell surface and total), HSP70 and HSP90 (Figures 8, 9). AR12 and neratinib interacted to further reduce the expression of HSP90 and to inactivate eIF2α. Chaperones are complexed with other proteins including BAG3 (associated with HSP70) and AHA1 and CDC37 (associated with HSP90). BAG3, AHA1 and CDC37 have all been linked to AD pathology [30–36]. BAG3 has been shown to enhance Tau degradation by autophagy [24, 25, 30]. HSP90 and AHA1 promote Tau pathogenesis [31]. And CDC37 with HSP90 also acts to maintain Tau stability [34–36]. AR12 alone as well as the drug combination increased BAG3 expression (Figures 8, 9). The drug combination reduced AHA1 levels but did not alter the expression of CDC37. The histone deacetylase HDAC6 regulates HSP90 activity; increased HSP90 acetylation reduces chaperone function [37]. AR12 and the drug combination reduced HDAC6 expression, which will result in increased HSP90 acetylation concomitant with a further reduction in overall HSP90 chaperoning activity (Figures 8, 9).

In neuronal cells and microglia, knock down of Rubicon, Beclin1 or ATG5 prevented the drugs alone or in combination from reducing the expression of chaperone proteins (Figures 10A, 10B). Knock down of Beclin1 or ATG5 prevented AR12 alone or in combination from enhancing eIF2α S51 phosphorylation in microglia whereas knock down of Rubicon did not (Figure 10B) [38–41]. We discovered, however, that knock down of eIF2α significantly reduced autophagosome formation and prevented the degradation of Tau and APP by AR12 and neratinib (Figure 11). This implies eIF2α regulates autophagy, but that autophagy also regulates serine 51 phosphorylation of eIF2α. In agreement with the autophagy data in Figure 11, AR12 and the drug combination reduced expression of p62 and LAMP2 (Figure 12). Thus, we hypothesize that the initial inactivation of GRP78 catalytic function by AR12 facilitates an initial increase in eIF2α phosphorylation which in turn is essential for greater levels of eIF2α phosphorylation, greater levels of macroautophagy and eventually leading to significant amounts of Tau / APP / chaperone protein degradation.

Based on our data showing reduced chaperone expression following drug exposure, we next defined which chaperones played the most important roles in regulating APP and Tau stability; in HCN2 and BV2 cells (Figure 13); in RAW macrophages (Supplementary Figure 2). Over-expression of GRP78, HSP70 or HSP90, or knock down of GRP78, HSP70 or HSP90 surprisingly did not significantly alter the basal expression levels of APP and Tau (not shown). This data demonstrates that over-expression of either GRP78, HSP70 or HSP90 prevented the drug-induced degradation of APP. Over-expression of GRP78, but not of HSP70 or HSP90, prevented the drug combination from reducing Tau expression. Knock down of GRP78, but not of HSP70 or HSP90, enhanced the ability of AR12 alone and the drug combination to reduce APP expression. Knock down of GRP78 also further enhanced the ability of AR12 as a single agent to reduce Tau levels. Our GRP78 data is congruent with our earlier findings when knocking down eIF2α expression. Collectively these findings strongly argue that the chaperone GRP78 and translation regulator eIF2α play key roles in regulating the ability of AR12 and neratinib to reduce Tau and APP protein levels.

The co-chaperone BAG3 has been shown to facilitate the degradation of Tau and APP [30–32]. Over-expression of GRP78, HSP70 or HSP90, or knock down of GRP78, HSP70 or HSP90 did not alter the basal expression level of BAG3 (not shown). Knock down of GRP78, HSP70 or HSP90 significantly enhanced the ability of AR12 to enhance BAG3 expression (Figure 14A, 14B). Over-expression of GRP78, HSP70 or HSP90 significantly reduced the ability of AR12, alone or in combination with neratinib, from enhancing BAG3 levels (Figure 14A, 14B). Thus, reduced chaperone levels facilitate more drug-induced BAG3 expression.

Knock down of BAG3 reduced drug-induced autophagosome formation and autophagic flux (Figure 14C). AR12 and neratinib, as previously observed, profoundly reduced the expression of Tau and APP, and knock down of BAG3 almost abolished the abilities of AR12 and neratinib to cause degradation of Tau (~92%) and APP (~91%) (Figure 14D). Knock down of BAG3 also significantly reduced the abilities of AR12 and neratinib to reduce the expression of GRP78 (total and cell surface) and of HDAC6 (Figure 15A). Notably and in contrast to Tau and APP, a trend of GRP78 degradation was observed even in drug-treated BAG3 knock down cells, with a reduction in degradation of only ~48%. This data suggests that the regulation of APP and Tau expression after AR12 / neratinib exposure, compared to GRP78, is exquisitely dependent upon BAG3 expression.

Knock down of BAG3 prevented the degradation of HSP90, p62 and LAMP2 which is congruent with our autophagy data (Figure 15B). AR12 and neratinib activated PERK and significantly increased eIF2α S51 phosphorylation (Figure 15C). Knock down of BAG3 reduced the ability of the drugs to increase PERK and eIF2α phosphorylation. This data suggests that the initial catalytic inhibition of GRP78 by AR12 that facilitates ER stress signaling, which in turn leads to autophagy, and degradation of GRP78, acts in a feed-forward fashion to fully activated ER stress signaling.

Finally, we attempted to link cause-and-effect for the actions of AR12 upon the expression of BAG3 and the role of autophagy and ER stress signaling. Knock down of Beclin1, ATG5, ULK1, eIF2α or PERK significantly reduced AR12-induced BAG3 expression (Figure 16). This data further supports the concept that the drugs cause a feed-forward signaling loop via ER stress signaling and autophagy to degrade Tau and APP.

Discussion

The present studies were performed to determine whether AR12 and neratinib in microglia and neuronal cells caused the degradation of Tau and APP. Our data in a neuronal cell line, a microglial cell line and an established monocyte cell line, were near identical to our prior findings in HCT116 colon cancer cells. Previously we had shown that AR12 and neratinib reduced Tau and APP levels via macro-autophagy and that cells expressing the autophagy protein ATG16L1 T300 were more capable of autophagosome formation and APP / Tau degradation than cells expressing the ATG16L1 A300 isoform [10, 38–42]. In the present studies, knock down of a regulator of LC3-associated phagocytosis / endocytosis, Rubicon, significantly reduced the abilities of the drugs alone or in combination to reduce the expression of chaperone proteins, Tau and APP. LAP cooperated with macroautophagy in facilitating the degradation of Tau and APP (Figure 17) [43].

In our initial studies, we discovered that AR12 significantly increased BAG3 expression. BAG3 has been shown to play a role in the regulation of eIF2α phosphorylation and promote macroautophagy, HDAC6 function, and a transcriptional regulatory in its own right and with heat shock factor 1 (HSF1) [44–48]. The IRE1 ER stress pathway has been proposed to regulate the transcription of BAG3 in part by the regulation of HSF1 [48, 49]. Over-expression of GRP78 prevents both the phosphorylation of eIF2α and also activation of the IRE1 ER stress pathway and our data demonstrated that expression of GRP78 almost abolished the drug-induced expression of BAG3. This data argues we are inducing a coordinated series of cell signals which promote ER stress signaling to increase the expression of BAG3, Beclin1 and ATG5 which collectively facilitate the formation of autophagosomes which sequester chaperones, Tau, and APP, leading to their degradation.

GRP78 is localized in the ER where it binds to and inactivates PERK and chaperones newly synthesized proteins and in the outer leaflet of the plasma membrane where it mains stability of receptors and itself can act as a docking protein. AR12 reduced the protein levels of ER- and plasma membrane-localized GRP78 however the mechanisms by which this occurred were overlapping but not identical. The degradation of ER-localized required macroautophagy whereas the reduction in membrane-localized GRP78 required LAP and macroautophagy. The ability of the AR12 and neratinib drug combination to increase eIF2α S51 phosphorylation also reflected this pattern where knock down of Rubicon did not prevent eIF2α inactivation. In AD, the ability of cells to mount an ER stress response and clear denatured proteins is impaired and our findings argue that one way to overcome this issue is the application of AR12 and neratinib which complement each other in promoting ER stress signals and protein degradation.

Membrane localized GRP78 is capable of sensing the presence of denatured Tau and APP in the extracellular space liminal to the plasma membrane of neurons and microglia. Extracellular GRP78 chaperones both Tau and APP and extracellular GRP78 is essential for amyloid-β uptake by microglia [50–56]. Previously we noted that cells expressing ATG16L1 T300 expressed 25% higher cell surface levels of GRP78 than cells expressing ATG16L1 A300 [10]. Amyloid-β induces cells to over-express GRP78 and GRP78 is over-expressed in neurons from APP/PS1 mice [57, 58]. n.b. This is the well-described ER stress response that occurs after any perceived overload of denatured protein. The exogenous membrane-associated GRP78, once ingested with the amyloid-β / Tau proteins, was shown to translocate to the ER of the microglia where it acts to block ER stress signaling by PERK, and hence the macroautophagic digestion of denatured Tau and amyloid-β proteins. Thus, neurons and microglia from persons with more surface GRP78 are likely to have an enhanced capability to take up extracellular materials such as Tau and APP which, in contrast to the beneficial effect this has in Crohn’s Disease, in AD is deleterious. And, if Tau is considered to have ‘prion’-like properties, elevated plasma membrane GRP78 levels will result in a greater amount of Tau being propagated / seeded into bystander neurons and microglia.

Signaling by ERBB2 and KRAS play important roles in the development and progression of Alzheimer’s Disease [22, 23, 59–64]. Neratinib as a single agent and trending more so when combined with AR12 reduced both ERBB2 and KRAS levels. The robust changes in protein phosphorylation after drug exposure demonstrated that a strong signal was being sent to the cell to form autophagosomes. This agrees with our data showing that AR12 and neratinib interacted to cause autophagosome formation. The degradation and dephosphorylation of p70 S6K has important consequences for a cell over-expressing Tau and APP. Phosphorylation of ribosomal S6 is mediated by p70 S6K, and this enhances ribosomal protein synthesis. Inactivation of mTORC1 and p70 S6K, combined with enhanced phosphorylation of eIF2α S51 will likely completely shut down further synthesis of Tau and APP. In parallel, with enhanced autophagosome formation and autophagic flux, insoluble aggregates of Tau and APP will be cleared, restoring protein homeostasis to the neuron.

The co-chaperone BAG3 has been linked to the regulation of autophagy and cell viability, for example, Ji et al. demonstrated that BAG3 facilitates the autophagic degradation of Tau [30–32]. AR12 increased BAG3 expression and knock down of BAG3 reduced the ability of AR12 to cause autophagosome formation. Knock down of BAG3 profoundly reduced the ability of AR12 to reduce the expression of Tau and APP. BAG3 also facilitates the clearance of endogenous Tau in primary neurons and it plays a key role in sensing and regulating protein quality control [65]. Our data demonstrated that knock down of PERK / eIF2α significantly reduced the ability of AR12 to increase BAG3 expression. Furthermore, eIF2α is required to increase the expression of Beclin1 and ATG5 and knock down of eIF2α significantly reduces AR12-induced autophagosome formation. Knock down of Beclin1 / ATG5 / ULK1 also significantly reduced the AR12-induced expression of BAG3. These findings collectively support the hypothesis that AR12 via an initial catalytic inhibition of GRP78 results in an initial wave of ER stress signaling which leads to autophagosome formation. This results in a feed-forward positive loop where GRP78 and other chaperones are digested via autophagy resulting in greater levels of ER stress signaling, a significant increase in BAG3 expression, leading to greater autophagosome formation and the digestion of APP and Tau. Studies beyond the scope of the present manuscript will be required to fully define the relationship between the actions of AR12 and the biology of BAG3.

Our initial hypothesis, stated in the Introduction, was oversimplistic. Regulation of GRP78 function by both agents influences the abilities of the drugs to degrade APP and especially Tau. The relevance of BAG3 to the processes of degrading chaperones, APP and Tau was not initially appreciated, and our data highlight the importance of this cochaperone protein. Our prior publication and the data presented in this manuscript strongly suggest that AR12 and neratinib can cause the degradation of Tau and APP in multiple cell types, including microglia and neurons. Future work, based on the availability of funding, will be required to perform in vivo studies in transgenic Tau and APP mice to define whether these drugs have therapeutic efficacy against Alzheimer’s Disease mouse models. Should such studies eventually take place, we hope that this therapeutic combination can be tested for safety and activity in AD patients.

Supplementary Materials

Supplementary Figures