Alzheimer’s disease is the most common form of dementia, affecting over 55 million people worldwide (doubling every 25 years), but research is yet to develop preventive or restorative therapies; as such, in vitro models efficiently mimicking this neuropathology are extremely pertinent for current biomedical research, with obvious societal impact, and will continue to be for the foreseeable future. The present study explored and characterised different cell models that can used to study protein aggregation and protein aggregation-related diseases. Moreover, although, the importance of an in vitro AD model based on commercially available cell lines cannot be overstated, a more physiologically significant alternative model would also be of paramount value. One such tool is the use of fibroblasts from human donors, which can be obtained via minimally invasive biopsy, with each cell line reflecting the uniqueness of the donor themself. This work characterises alternative cellular models to mimic AD, one using the neuronal cell line SH-SY5Y, and the other using human fibroblasts from healthy donors.
Incubation of the neuronal cell line SH-SY5Y with aggregated Aβ1–42 greatly increases protein aggregation, relatively to control cells, even more so than proteasome inhibition (Fig. 1a), confirming it as a good model to study the protesostasis-related aspects in AD.
The different models feature distinct aggregates (Fig. 1a). Those induced by Aβ are typical of aggregates formed by the accumulation of peptides: this is the case of Aβ aggregation, but a similar process occurs with the accumulation of huntingtin fragments 47,48 or synphilin-1 49. The fact that the larger aggregates were obtained upon Aβ addition is consistent with literature, as it has been well reported that aggregated Aβ recruits further fragments to the deposits 50 possibly even including secreted APP 36. Given the described aggregate characteristics, it is reasonable to posit that Aβ approximates aggregate formation associated with AD.
The chaperone HSP-70 plays crucial roles in proteostasis, and is currently being explored as a good target to modulate the AD-phenotype, with studies suggesting shown that AD pathology can be supressed by increasing cellular levels of HSP-70 51–53. Our results show no differences in HSP-70 levels after treatment with the Aβ peptide, contrary to what was observed with MG-132, indicating that Aβ and MG-132 trigger distinct aggregate formation mechanisms.
We have previously described this model to increase tau phosphorylation in rat primary cortical neuronal cultures 37. Tau has 85 phosphorylatable residues (45 of which found in AD brains) 18; however, it has been demonstrated that residue Thr231 is the primary phosphorylation site for GSK-3β, critical for tau hyperphosphorylation, and phosphorylation at this residue can destabilize tau-binding to microtubules and promote tau aggregation 19. Although in this study only a slight increase in Thr231 phosphorylation was suggested in response to incubation with oligomerised Aβ1–42, it appeared to colocalize intensely with increased aggresomes, suggesting that phospho-tau is aggregating or being sequestrated in protein aggregates (Supplementary Fig. 2).
The Aβ1–42 peptide model also exhibits a decrease in extracellular APP and increase in intracellular APP, in both neuronal (Fig. 4) and non-neuronal (Fig. 5) cells lines, suggesting that APP metabolism is hindered, accordant to our previous work with primary neuronal cultures38.
High levels of α-synuclein in cerebrospinal fluid (CSF) have been reported in patients with AD and mild cognitive impairment. Consistently, our results suggest an increase in α-synuclein levels (Fig. 3) and aggregation (Supplementary Fig. 1) upon treatment with Aβ1–42 peptide, strengthening the relevance of this model to study different protein aggregation-related phenotypes.
With aging, mitochondrial function is compromised, impairing metabolism and increasing oxidative stress2. Mitochondria-produced ATP is important for several processes of protein degradation and homeostasis; being so, manipulation of mitochondria is a sensible strategy to develop protein-aggregation cell models26,54. Rotenone was used to induce oxidative stress in SH-SY5Y by inhibiting the mitochondrial electron chain transport, promoting electron accumulation and consequently increasing ROS levels31. Using a parallel approach, ATP production was disrupted by inhibiting ATP synthase with oligomycin55. Although no significant differences were observed in protein aggregation with both chemicals, a small increase in aggregation occurs with rotenone (Fig. 1), which induced smaller aggregate punctate formation, like MG-132 (Fig. 1). Recent studies in our laboratory described that rotenone-induced protein aggregation in human fibroblasts can be rescued by caloric restriction 27, and previous work has linked starvation with improved mitochondria activity, reducing oxidative stress56, which may also contribute to the observed results. Moreover, our results suggest that mitochondria manipulation induces protein aggregation through mechanisms distinct from aggregated Aβ1–42, as it fails to fully induce the AD-phenotype observed with that model.
Another relevant point is that rotenone-induced mitochondrial dysfunction can potentiate accumulation of misfolded proteins, like α-synuclein, being a commonly used model of PD28,29,31, and our results suggest this pesticide might be increasing the levels and aggregation of α-synuclein (Fig. 2 and Supplementary Fig. 1). While the physiological action of α-synuclein is important for mitochondrial homeostasis, its pathological aggregation can negatively impinge on mitochondrial function57,58: soluble α-synuclein is targeted to mitochondria59, and the interaction of oligomeric α-synuclein with mitochondria results in decreased respiration60. Taken together, results support the use of rotenone as a good model to study PD.
The molecular chaperone HSP-70 inhibits α-synuclein fibril formation and alters the characteristics of toxic α-synuclein aggregates61, with studies in animal models describing rotenone exposure to decrease HSP-70 expression 62; in various diseases, aggregation-induced toxicity is supressed by HSP-70 overexpression, reinforcing its important role in aggregation-related disease 63,64. However, no differences in HSP-70 levels were observed in SH-SY5Y cells after rotenone exposure, suggesting that other mechanisms of UPR can be altered and be responsible for the observed increase in α-synuclein aggregation.
Cellular homeostasis is maintained via ubiquitin-proteasome and lysosomal-autophagic systems to prevent protein aggregation 2. The aldehyde MG-132 binds to the active site of the proteasome, inhibiting its activity 65, and is frequently used as a positive control for protein aggregation66. As expected, protein aggregation increases with proteasome inhibition (Fig. 1), an increase that is accompanied by an in tau phosphorylation in residue Thr231. Tau hyperphosphorylation is a histological hallmark of many neurodegenerative disorders that can ultimately form tau aggregates, and proteasome inhibition has been reported to promote degradation and oligomeric accumulation of tau67. Aggregates produced by MG-132 are typical of aggregates containing ubiquitinated proteins, known as aggresome-like induced structures (ALIS) 68. These results are consistent with the state-of-the art, given the effect of MG-132 as a proteasome inhibitor.
On the other hand, work using neuroblastoma cell lines reports proteasome inhibition to result in tau proteolysis69, also consistent with the decrease in the levels of total tau observed in this work (Fig. 3). In parallel to decreased tau levels, APP levels (both intra- and extra-cellular) are also decreased by MG-132 (Fig. 4). A cellular safety mechanism to cope with increased protein aggregation is to reduce protein transduction, which can explain these results. Moreover, proteasome inhibition has been described to promote a reduction in mature APP/APLP1 (amyloid beta precursor-like protein 1) via autophagy induction 70.
In contrast to the other models explored in this study, HSP-70 expression was found to increase with proteasome inhibition, in agreement with other studies reporting MG-132 to result in accumulation of ubiquitinated proteins and increase in HSPs expression 71.
To conclude, this study establishes and characterizes different cellular models for protein aggregation-related diseases, which we believe are of great value for studies in related research fields. By adequately applying these protocols to model the more appropriate disease or biological process, several additional scientific questions and disease-related mechanisms can be explored.