Δ8-THC Protects against Amyloid Beta Toxicity Modulating ER Stress In Vitro: A Transcriptomic Analysis

Alzheimer’s disease (AD) represents the most common form of dementia, characterized by amyloid β (Aβ) plaques and neurofibrillary tangles (NFTs). It is characterized by neuroinflammation, the accumulation of misfolded protein, ER stress and neuronal apoptosis. It is of main importance to find new therapeutic strategies because AD prevalence is increasing worldwide. Cannabinoids are arising as promising neuroprotective phytocompounds. In this study, we evaluated the neuroprotective potential of Δ8-THC pretreatment in an in vitro model of AD through transcriptomic analysis. We found that Δ8-THC pretreatment restored the loss of cell viability in retinoic acid-differentiated neuroblastoma SH-SY5Y cells treated with Aβ1-42. Moreover, the transcriptomic analysis provided evidence that the enriched biological processes of gene ontology were related to ER functions and proteostasis. In particular, Aβ1-42 upregulated genes involved in ER stress and unfolded protein response, leading to apoptosis as demonstrated by the increase in Bax and the decrease in Bcl-2 both at gene and protein expression levels. Moreover, genes involved in protein folding and degradation were also deregulated. On the contrary, Δ8-THC pretreatment reduced ER stress and, as a consequence, neuronal apoptosis. Then, the results demonstrated that Δ8-THC might represent a new neuroprotective agent in AD.


Introduction
Alzheimer's disease (AD) is the most common neurodegenerative disorder and represents the most frequent cause of dementia, affecting more than 50 million people worldwide [1,2]. Considering that the strongest risk factor for AD is aging and that life expectancy is gradually increasing, the number of AD patients is also progressively rising [3,4].
Even though AD is mainly sporadic and shows a late onset (>65 years of age), there are rare cases in which the disease is associated with autosomal dominant inheritance and usually develops earlier (between 24 and 60 years of age) [5]. More than 300 pathogenic mutations in presenilin 1 (PSEN1), presenilin 2 (PSEN2), and amyloid precursor protein (APP) genes have been identified in familial AD cases [5]. Instead, allelic variation in the apolipoprotein E (APOE) gene represents a major genetic risk factor for sporadic AD [6].
The most important neuropathological features of AD are β-amyloid (Aβ)-containing extracellular plaques and tau-containing intracellular neurofibrillary tangles [7,8]. In addition, AD is characterized by the atrophy of the cerebral cortex and by the loss of basal forebrain cholinergic neurons [9].
It is known that AD patients generally manifest prominent amnestic cognitive impairment; however, non-amnestic cognitive impairment is less frequent [8]. AD symptoms can vary depending on the stage of the disease and AD can be classified as preclinical

∆ 8 -THC Counteracted the Aβ 1-42 -Induced Loss of Cell Viability
RA-differentiated SH-SY5Y were pretreated with different doses of ∆ 8 -THC for 24 h and, after, were treated with 10 µM Aβ 1-42 for another 24 h. Using the MTT assay, we evaluated if ∆ 8 -THC exerted toxicity in the range of the doses tested (5-20 µM) and if it was able to counteract Aβ 1-42 toxicity.
The MTT assay results demonstrated that 10 µM Aβ 1-42 treatment reduced the cell viability of RA-differentiated SH-SY5Y cells. ∆ 8 -THC was not cytotoxic at all the concentrations tested. However, 5 µM ∆ 8 -THC was not able to counteract Aβ 1-42 -reduced cell viability. On the contrary, both 10 and 20 µM ∆ 8 -THC were able to restore the cell viability of RA-differentiated SH-SY5Y cells after 10 µM Aβ 1-42 treatment (Figure 1). Transcriptomic analysis was carried out using the concentration 20 µM ∆ 8 -THC. order to evaluate whether ∆ 8 -THC could modulate signaling pathways leading to protective effects.

Δ 8 -THC Counteracted the Aβ1-42-Induced Loss of Cell Viability
RA-differentiated SH-SY5Y were pretreated with different doses of Δ 8 -THC for 24 h and, after, were treated with 10 µM Aβ1-42 for another 24 h. Using the MTT assay, we evaluated if Δ 8 -THC exerted toxicity in the range of the doses tested (5-20 µM) and if it was able to counteract Aβ1-42 toxicity.

Transcriptomic Analysis Revealed That ∆ 8 -THC Counteracted the Aβ 1-42 -Induced ER Stress
In order to evaluate the differential pattern of gene expression in RA-differentiated SH-SY5Y treated with 20 µM ∆ 8 -THC and 10 µM Aβ 1-42 , we performed NGS transcriptomic analysis. The aim was to evaluate the pathways associated with the protective effects exerted by ∆ 8 -THC. Figure 2 represents the distribution of differentially expressed genes (DEGs) between control against Aβ 1-42 (CTRL vs. Aβ 1-42 ), control against ∆ 8 -THC (CTRL vs. ∆ 8 -THC), or Aβ 1-42 against ∆ 8 -THC + Aβ 1-42 groups (Aβ 1-42 vs. ∆ 8 -THC + Aβ    Venn diagram in the center of the plot highlights the amount of DEGs found exclusively in each comparison (outer circles), how many DEGs found in two groups but not in the other (intersection of two circles) or how many DEGs were found in each comparison (center of the diagram). Each donut plot highlights, in turn, the intersection of two comparisons showing in dark red the upregulated and in the light red the downregulated DEGs.
Given that significantly enriched GO are related to ER functions and proteostasis, we focused on related DEGs looking at KEGG pathways "Alzheimer disease" (hsa05010) and "protein processing in endoplasmic reticulum" (hsa04141). In Table 1, we report the common DEGs, altered in both CTRL vs. Aβ 1-42 and Aβ 1-42 vs. ∆ 8 -THC + Aβ 1-42 . In the supplementary Table S1, we report all the inspected DEGs. The position of the bubble in the x axis shows the number of DEGs in the ontology (the more on the right, the higher the number of DEGs). The number of DEGs was normalized over the number of genes included in the ontology term itself so that the terms were comparable to each other. The size of the bubble is a score given by −log(q-value).
Given that significantly enriched GO are related to ER functions and proteostasis, we focused on related DEGs looking at KEGG pathways "Alzheimer disease" (hsa05010) and "protein processing in endoplasmic reticulum" (hsa04141). In Table 1, we report the common DEGs, altered in both CTRL vs. Aβ1-42 and Aβ1-42 vs. Δ 8 -THC + Aβ1-42. In the supplementary Table S1, we report all the inspected DEGs. Table 1. Common DEGs, altered in both CTRL vs. Aβ1-42 and Aβ1-42 vs. Δ 8 -THC + Aβ1-42, related to "Alzheimer's disease" (hsa05010) and "protein processing in endoplasmic reticulum" (hsa04141) pathways.  . The number of DEGs was normalized over the number of genes included in the ontology term itself so that the terms were comparable to each other. The size of the bubble is a score given by −log(q-value). The column fold change shows for each DEG the difference in the level of expression computed by log 2 (Aβ 1-42 /CTRL), log 2 (∆ 8 -THC + Aβ 1-42 /Aβ 1-42 ) or log 2 (∆ 8 -THC/CTRL). The q-Value column was obtained correcting the p-value through Benjamini-Hochberg correction. All values were rounded to the second decimal digit.

∆ 8 -THC Restored the Protein Levels of Bax and Bcl-2
In order to evaluate the effects of ∆ 8 -THC on Aβ 1-42 -induced apoptosis, we evaluated the levels of Bax and Bcl-2. Western blot analysis evidenced a significant increase in Bax in RA-differentiated SH-SY5Y treated with 10 µM Aβ 1-42 . Pre-treatment with 20 µM ∆ 8 -THC was able to reduce Bax protein levels. On the contrary, Bcl-2 protein levels were decreased in Aβ 1-42 treated cells, while ∆ 8 -THC restored its levels ( Figure 4). ∆ 8 -THC treated RA-differentiated SH-SY5Y showed a level of Bax similar to the control, while Bcl-2 increased.

Δ 8 -THC Restored the Protein Levels of Bax and Bcl-2
In order to evaluate the effects of Δ 8 -THC on Aβ1-42-induced apoptosis, we evaluated the levels of Bax and Bcl-2. Western blot analysis evidenced a significant increase in Bax in RA-differentiated SH-SY5Y treated with 10 µM Aβ1-42. Pre-treatment with 20 µM Δ 8 -THC was able to reduce Bax protein levels. On the contrary, Bcl-2 protein levels were decreased in Aβ1-42 treated cells, while Δ 8 -THC restored its levels ( Figure 4). Δ 8 -THC treated RA-differentiated SH-SY5Y showed a level of Bax similar to the control, while Bcl-2 increased.

Discussion
The prevalence of dementia is increasing, and it is expected that about 113 million will be affected in 2050 worldwide [26]. Given that AD is the prevalent form of dementia, it is of main importance to find new therapeutic strategies. Cannabinoids seem promising

Discussion
The prevalence of dementia is increasing, and it is expected that about 113 million will be affected in 2050 worldwide [26]. Given that AD is the prevalent form of dementia, it is of main importance to find new therapeutic strategies. Cannabinoids seem promising for neuroprotective treatments. Some of them have been reported to improve cognitive functions and reduce Aβ [20]. ∆ 8 -THC is present in a very low quantity in plants, and it is mainly produced by cannabidiol. ∆ 8 -THC is a structural isomer of the ∆ 9 -THC, showing a double bond between carbon atoms 8 and 9 rather than carbon atoms 9 and 10. ∆ 9 -THC is responsible for the psychoactive properties of cannabis, such as alterations in mood, perception and cognition. ∆ 9 -THC is one of the cannabinoids most studied, but ∆ 8 -THC has also been attracting attention for the better thermodynamic stability in comparison to ∆ 9 -THC. The two compounds showed similar pharmacokinetics and pharmacodynamics. Both ∆ 8 -THC and ∆ 9 -THC are partial agonists of the cannabinoid receptor CB1, but ∆ 8 -THC showed a lower affinity. Given that the psychoactive effects of ∆ 9 -THC depend on the CB1 receptor, ∆ 8 -THC has a lower psychotropic potency. Both ∆ 8 -THC and ∆ 9 -THC were reported to act as agonists or inverse agonists at the CB2 receptor [22]. A survey of consumers highlighted that ∆ 8 -THC might exert the benefits of ∆ 9 -THC with lower risks [27].
Some cannabinoids have already shown protective effects in both in vitro and in vivo AD models. ∆ 9 -THC was shown to lower Aβ levels in an in vitro AD model in a dosedependent manner, directly binding to the Aβ peptide and inhibiting its aggregation [28]. Moreover, it can alleviate cognitive impairments and reduce inflammatory markers, the numbers of Aβ plaques and degenerated neurons in AD mice [29][30][31]. Additionally, other agonists of CB1 and CB2 receptors were tested in AD models. ACEA, a CB1 receptor agonist, exerted a strong neuroprotective action against Aβ toxicity in vitro and in vivo [32,33]. Additionally, CB2 agonists, such as JWH-133, showed neuroprotective effects in AD models, reducing inflammation, Aβ plaque and deposition, increasing Aβ clearance and improving cognitive performance [34]. CP55940, an agonist of CB1 and CB2, restored mitochondrial membrane potential and reactive oxygen species and reduced extracellular Aβ [35].
However, the potential neuroprotective effects of ∆ 8 -THC have not been investigated yet. To our knowledge, this is the first study that has investigated ∆ 8 -THC effects in an in vitro model of AD.
In this study, we found that ∆ 8 -THC showed no cytotoxicity at all the doses tested. Interestingly, it was able at doses 10 and 20 µM to restore the loss of cell viability induced by Aβ  . The dose of 20 µM ∆ 8 -THC was used for other experiments.
Transcriptomic analysis and GO evaluations evidenced the enrichment of biological processes related to proteostasis and neuronal apoptosis. Then, ∆ 8 -THC could modulate the pathways involved in proteostasis to exert protective effects. For this reason, we focused on DEGs related to ER functions and proteostasis, looking at the KEGG pathways for "Alzheimer's disease" (hsa05010) and "protein processing in endoplasmic reticulum" (hsa04141).
ER plays important roles in protein biosynthesis and in their quality control. In some cases, the maintenance of cellular homeostasis is not possible, causing a reduction in the protein folding capacity of ER and leading to the accumulation of misfolded/unfolded proteins in ER. This process caused the disruption of cellular homeostasis, inducing ER stress. AD is characterized by protein misfolding and aggregation and Aβ accumulation, which are all events that trigger ER stress [36].
Treatment with Aβ 1-42 also altered the expression of several chaperones. Chaperones are a functionally related group of proteins that assist protein folding both in physiological and stress conditions. Among chaperones, heat shock proteins (HSPs) are well known. HSPs have a role in all the phases of proteostasis; they participate in folding, protein synthesis and degradation. Their levels increase during stress exposure, helping to prevent conformational changes and the aggregation of misfolded proteins [37]. We found that several HSPs were dysregulated by Aβ 1-42 treatment. Aβ 1-42 treatment increased the expression of several members of the HSP40 family (DNAJC1, DNAJC10, DNAJC3), HSP70 family (HSPA1A and HSPA1L) and HSP90 family (HSP90AA1 and HSP90AB1). HSPs also played a role in the degradation of proteins by the proteasome. The genes encoding for these HSPs, such as DNAJA2 and DNAJB12, were downregulated by Aβ 1-42 treatment.
We also found the upregulation of HSP90B1, encoding for GRP94, after Aβ 1-42 treatment. GRP94 is a chaperone that directs the folding and/or assembly of proteins. Moreover, GRP94 is one of the few major luminal calcium-binding proteins. It seems to have a role in ER-associated degradation (ERAD) to distinguish misfolded proteins and target them for degradation [38]. ∆ 8 -THC reduced the levels of HSP90B1, DNAJC1 and HSP90AB1 while increasing DNAJA2 expression in cells treated with Aβ 1-42 .
The excess of misfolded proteins induces ER stress. In order to counteract ER stress, cells activate UPR. The UPR starts as cell-protective cascades, with the aim of reducing the ER load of unfolded proteins through the inhibition of protein synthesis and the upregulation of protein folding and degradation. However, prolonged UPR finally leads to cell death. The UPR signaling involves three sensor proteins, which are PERK, ATF6, and IRE1. Abnormal levels of these effectors of UPR were reported in AD brains [39]. We found the upregulation of PERK (EIFAK2 and EIFAK4), ATF6 (ATF6) and IRE1 (ERN1) in cells treated with Aβ 1-42 . PERK activation is reported in AD and is associated with neurodegeneration and memory deficits [40]. Additionally, IRE1 activation is known to participate in AD pathogenesis and to be positively correlated with the progression of AD [41]. IRE1 activation is also associated with the induction of apoptosis [42]. Active ATF6 has also been shown in AD models [42]. We also found the downregulation of WFS1, which influences ER stress, to negatively regulate ATF6α [43]. Interestingly, it was found that the protein level of WFS1 and the number of WFS1+ neurons decreased in both AD-like mouse model brains and human post-mortem AD [44]. In particular, WFS1 deficiency was linked with increased tau pathology and neurodegeneration. WFS1 deficiency may induce chronic ER stress and affect the degradation and clearance of tau aggregates [45]. Interestingly, ∆ 8 -THC pre-treatment reduced the expression of the genes encoding for PERK, ATF6 and IRE1.
Misfolded/unfolded proteins are eliminated by the proteasome through ERAD. ERAD can be divided into four steps that are substrate recognition, dislocation across the membrane, ubiquitination and degradation by the proteasome [46]. The data suggest alterations of ERAD in AD [47][48][49][50]. In this study, we found a dysregulation of genes involved in the ERAD process. Specifically, genes involved in the phase of the recognition of unfolded protein were upregulated by Aβ 1-42 . We found the upregulation of genes encoding for ERManI (MAN1A2 and MAN1B1). Removal of mannose residues is a critical process in targeting misfolded glycoproteins for degradation. This removal is operated by ERManI together with EDEM [51,52]. This trimming permits misfolded glycoproteins to be bound to OS-9 and XTP3-B, which target them to ERAD. OS-9 is upregulated in response to ER stress and is required for the ubiquitination of ERAD substrates, suggesting that it may help transfer misfolded proteins to ubiquitination machinery [53]. OS9 and XTP3B redundantly promote glycoprotein degradation, but XTP3B inhibits the degradation of non-glycosylated proteins, while OS9 antagonizes this inhibition [54]. After treatment with Aβ 1-42, we also found the upregulation of OS-9 and XTP3B (ERLEC1). We also found the downregulation of EDEM1. It can modulate APP metabolism, and its overexpression is associated with a decrease in Aβ secretion [55]. Interestingly, EDEM1 can be upregulated by ∆ 8 -THC pre-treatment. ∆ 8 -THC pre-treatment also reduced OS-9 and MAN1B1.
We also found deregulation in the genes encoding for PDI (PDIA6, P4HB, TXNDC5) and ERO (ERO1A). Aβ 1-42 increased the genes encoding for ERO while reducing the expression of those encoding for PDI. In the oxidizing environment of the ER, unfolded proteins interact with PDI and undergo oxidative protein folding. In this way, misfolded substrate proteins can be reduced and refolded or isomerized to the appropriate native protein conformation. Misfolded proteins are reduced and isomerized by PDI and converted to their appropriate native conformation. Then, reduced PDI is reoxidized by ERO1. ∆ 8 -THC pre-treatment reduced ERO (ERO1B) while also increasing PDI (ERP29 and TXNDC5).
On the contrary, the genes involved in the processes of translocation, ubiquitination and degradation were mainly downregulated by Aβ 1-42 . Bap31, encoded by BCAP31 which we found downregulated, have several roles in ER homeostasis: membrane protein chaperone, quality control, and it is involved in ER stress and ERAD [56]. Its deficiency was associated with the formation of Aβ plaques in a murine AD model [57]. P97, encoded by VCP, also plays a critical role in protein dislocation in ERAD [46]; it is involved in aggregates clearance, and, indeed, its knockdown delayed the elimination of ubiquitinpositive aggregates [58]. We found the downregulation of VCP after Aβ 1-42 treatment.
The poly-ubiquitination of proteins is fundamental for their degradation by the proteasome, and different enzymes are required: an E1 activating enzyme activates ubiquitin in an ATP-dependent manner, an E2 ubiquitin-conjugating enzyme (Ubc), and an E3 ubiquitinprotein ligases that mediate the transfer of ubiquitin from the Ubc enzyme onto the target substrate. We found a downregulation in ubiquitin ligase complex subunits after treatment with Aβ 1-42 , such as UBE2G2 and others (RBX1, UBE2J1, UBQLN1, UBQLN2, UBQLN4, UBXN6, UBXN8). UBE2G2 was shown to be critically important for degradation through the ERAD of multiple substrates [59]. Additionally, FBXO6 was downregulated; it is a functional E3 ubiquitin ligase that plays a critical role in inhibiting ER stress-induced apoptosis [60]. ∆ 8 -THC pretreatment was able to reverse the alterations of the gene expression induced by Aβ 1-42 . In particular, the genes involved in protein targeting (EDEM3, MAN1B1) were downregulated, suggesting that there was less need for proteins involved in unfolded protein recognition. SEC61A1 and SEC61B were downregulated. The Sec61, which interacts with TRAP [61], encoded by SSR3, which was upregulated, mediates protein import into the ER and is also a candidate channel for the dislocation of ERAD substrates [62,63]. The expression of the genes involved in protein dislocation and ubiquitination increases (UBE2G2, SELENOS, TRAM1, UBE2J1, UBXN6, UBXN8, FBXO6). Then, if needed, misfolded/unfolded proteins can be dislocated to the cytosol and ubiquitinated.
Interestingly, ∆ 8 -THC reduced the expression of HERPUD1, encoding for HERP. HERP was reported to be involved in Aβ accumulation, including the formation of senile plaques [64]. ∆ 8 -THC also increased DERL1, encoding for Derlin-1, which plays a main role in the transport to the cytosol [65,66].
∆ 8 -THC also increased the expression of the shuttling factor RAD23A that delivers ubiquitin conjugates to the proteasome and activates its degradative capacity. ∆ 8 -THC also restored the expression of other genes that were involved in substrate delivery to the proteasome, such as ATXN3 [67].
Aβ 1-42 -treated cells also showed a downregulation of the genes encoding for 20S proteasome (PSMA5, PSMB2, PSMB3, PSMB4, PSMB5, PSMB7). The 20S proteasome was shown to be able to degrade misfolded, oxidized and intrinsically disordered proteins, but also Aβ, and to be the major degradation machinery under oxidizing conditions [68,69]. The 20S proteasome was reported to be inhibited in regions affected by Aβ, and Aβ aggregates were shown to inhibit proteasome activity in vitro [70]. It was shown that the Aβ precursor protein reduced the expression of the proteasome subunit α type-5 and β type-7, leading to cell death [71]. In line with the previous work, also in our work, these subunits were downregulated by Aβ 1-42 treatment. ∆ 8 -THC was able to partially upregulate the expression of proteasome subunits (PSMB5, PSMB6). In particular, we found the upregulation of PSMB5, whose overexpression was associated with increased resistance to Aβ 1-42 toxicity [72].
As we also said before, ER stress can trigger neuronal apoptosis. The Aβ 1-42 -induction of cell death in our study was demonstrated by the increase in BAX and the reduction in BCL2 gene expression. The transcriptomic results were also supported by Western blot analysis, which showed an increase in Bax protein levels and Bcl-2 reduction. ∆ 8 -THC pre-treatment reduced apoptosis induced by Aβ  , as demonstrated by the increase in Bcl-2 and the reduction in Bax levels.
RA-differentiated SH-SY5Y cells treated only with ∆ 8 -THC showed no ER stress; indeed, ATF6 and ERN1 were not differentially expressed compared to the control cells, and EIFAK3 was downregulated. Moreover, it increased some of the genes involved in the dislocation of misfolded proteins, such as DERL1, VCP, SSR3 and some proteasome subunits, including PSMB5, suggesting an efficient degradation of potentially unfolded proteins. Figure 5 reports the proteins encoded by DEGs and modulated in Aβ 1-42 and ∆ 8 -THC treated groups in the ER pathway.
deed, ATF6 and ERN1 were not differentially expressed compared to the control cells, and EIFAK3 was downregulated. Moreover, it increased some of the genes involved in the dislocation of misfolded proteins, such as DERL1, VCP, SSR3 and some proteasome subunits, including PSMB5, suggesting an efficient degradation of potentially unfolded proteins. Figure 5 reports the proteins encoded by DEGs and modulated in Aβ1-42 and Δ 8 -THC treated groups in the ER pathway.

Synthesis and Purification of Δ 8 -THC
To a stirred solution of CBD (200 mg, 0.636 mmol, 1eq) in DCM (5 mL), p-toluensulfonic acid (11 mg, 0.064 mmol, 0.1 eq) was added. The reaction was refluxed for 6 h, followed by TLC (Rf = 0.67, silica, petroleum ether-EtOAc 95:5) until the complete conversion of the starting material, which was then quenched with NaHCO3 s.s. and diluted with DCM. The combined organic phases were washed with brine, dried, and evaporated. The residue was purified by GCC on silica gel (pure petroleum ether to petroleum ether-EtOAc 9:1) to afford 182 mg (91%) of Δ 8 -THC as a brown oil.

Synthesis and Purification of ∆ 8 -THC
To a stirred solution of CBD (200 mg, 0.636 mmol, 1eq) in DCM (5 mL), p-toluensulfonic acid (11 mg, 0.064 mmol, 0.1 eq) was added. The reaction was refluxed for 6 h, followed by TLC (Rf = 0.67, silica, petroleum ether-EtOAc 95:5) until the complete conversion of the starting material, which was then quenched with NaHCO 3 s.s. and diluted with DCM. The combined organic phases were washed with brine, dried, and evaporated. The residue was purified by GCC on silica gel (pure petroleum ether to petroleum ether-EtOAc 9:1) to afford 182 mg (91%) of ∆ 8 -THC as a brown oil.

Cell Culture and Differentiation
The human neuroblastoma cell line SH-SY5Y was acquired from American Type Culture Collection (ATCC) (Manassas, VA, USA). Cells were grown in a monolayer at 37 °C

Cell Viability
Cell viability was evaluated with a Thiazolyl Blue Tetrazolium Bromide (MTT) assay. SH-SY5Y cells were cultured in 96-well plates, underwent RA differentiation, and were treated as reported in the previous paragraph. At the end of the treatment, the cells were incubated with a medium containing MTT (0.5 mg/mL; Sigma-Aldrich) at 37 • C for 4 h. The formed formazan crystals were dissolved in acidic isopropanol at 37 • C for 1 h, and the optical density was evaluated by the spectrophotometric measurement of absorbance using the microplate reader Victor NIVO TM (PerkinElmer, Waltham, MA, USA).

Extraction of Total RNA and cDNA Library Preparation
RNA extraction was carried out with a Maxwell ® RSC simplyRNA Cells Kit (Promega, Madison, WI, USA) according to the manufacturer's instruction. The preparation of the library was performed following the TruSeq RNA Exome protocol (Illumina, San Diego, CA, USA) as previously described [75].

RNA-Seq Data Analysis and Gene Evaluation
The raw data obtained from the NextSeq 550 Dx instrument of Illumina was evaluated using the fastqc tool version 0.11.4 from the Babraham Institute in Cambridge, UK. Adapters and low-quality bases were then eliminated through Trimmomatic [85] version 0.38 (Usadel Lab, Aachen, Germany). The cleaned reads were aligned to the human reference genome (GRCh38) using the STAR RNA-seq aligner [86] 2.7.3a (New York, NY, USA). The expression levels of the transcripts were computed using the htseq-count python package [87] version 0.6.1p1 (European Molecular Biology Laboratory (EMBL), Heidelberg, Germany). DEGs were identified using the DESeq2 library in R [88] version 3.6.3 (R Core Team). No cut-off was set on the fold change. Nevertheless, to drop false positive DEGs, the Benjamini-Hochberg procedure was used with a tight q-value of 0.01. The enrichment of the biological process terms of the gene ontology was also performed in R using the package biomaRt [89]

Protein Extraction and Western Blot Analysis
At the end of the treatment, SH-SY5Y were harvested with trypsin-Ethylenediaminetetraacetic acid (EDTA), and proteins were extracted using RIPA (Thermo Scientific™, Waltham, MA, USA) according to the manufacturer's instruction. Protein concentration was evaluated using the Bio-Rad Protein Assay (Bio-Rad Laboratories, Hercules, CA, USA) and bovine serum albumin (BSA) as standard. Twenty-five micrograms of proteins were separated on sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred onto a PVDF transfer membrane (Immobilon-P PVDF, Merck Millipore division of Merck KGaA, Darmstadt, Germany). Membranes were blocked for 1 h at room temperature, with PBS containing 5% non-fat dried milk. Then, membranes were incubated with primary antibodies overnight at 4 • C. The following primary antibodies were used: Bax (1:1000; Cell Signaling Technology, Danvers, MA, USA) and Bcl-2 (1:1000; Cell Signaling Technology, Danvers, MA, USA). The membranes were incubated with secondary antibodies and horseradish peroxidase (HRP)-conjugated anti-rabbit IgG (1:2000; Santa Cruz Biotechnology, Inc., Dallas, TX, USA) for 1 h at room temperature. To evaluate that blots were loaded with equal amounts of protein lysates, they were also incubated with an antibody for GAPDH HRP Conjugated (1:1000; Cell Signaling Technology). The relative expression of protein bands was visualized using an enhanced chemiluminescence system (Luminata Western HRP Substrates, Millipore Corporation, Billerica, MA, USA), and protein bands were obtained and quantified with a ChemiDoc™ MP System (Bio-Rad Laboratories S.r.l., Hercules, CA, USA) and analyzed with the software Image J 1.54d. The uncropped blots for Bax and Bcl-2 and relatives of GAPDH are available in the supplementary Figures S2 and S3, respectively.

Statistical Analysis
Statistical analysis of cell viability and Western blot was carried out using GraphPad Prism version 9.0 software (GraphPad Software, La Jolla, CA, USA). Multiple comparisons were performed using a one-way ANOVA test and the Bonferroni post hoc test. A p-value less than or equal to 0.05 was considered statistically significant. The results are expressed by the mean ± standard deviation (SD).

Conclusions
∆ 8 -THC reduced Aβ 1-42 -induced toxicity as a result of a reduction in ER stress. Indeed, ∆ 8 -THC restored proteostasis, increasing the expression of proteasome and ubiquitin subunits and reducing UPR, suggesting that misfolded/unfolded proteins were not accumulated but could be eliminated through the proteasome. As a consequence of the reduced ER stress, ∆ 8 -THC increased neuronal cell viability. The results suggested that ∆ 8 -THC may represent a novel neuroprotective agent in AD but also in other neurodegenerative diseases characterized by the accumulation of misfolded proteins.

Data Availability Statement:
The data presented in this study are openly available in the NCBI Sequence Read Archive at BioProject accession numbers PRJNA934843.