Alzheimer’s disease (AD) is the most common cause of dementia worldwide (Anand et al., 2014). One of the hallmarks of AD is the formation of amyloid plaques in various parts of the brain, consisting mainly of beta-amyloid peptide (Aβ). Aβ is known to undergo various post-translational modifications that affect its aggregation, toxicity, enzymatic degradation, and interaction with protein partners (Barykin et al., 2017). Many studies note that the presence of Aβ₄₂ alone is not enough for the development of pathology: the appearance of modified forms that change the action of Aβ₄₂ serves as a trigger for the disease (Barykin et al., 2017). Some of these modifications include phosphorylation at serine 8 (pS8-Aβ₄₂) and isomerization of aspartic acid 7 (iso-Aβ₄₂). Aβ containing isomerized Asp7 was found in more than 50% of the Aβ molecules of amyloid plaques, as well as in soluble fraction (Mukherjee et al., 2021). This modification is more prone to aggregation than unmodified Aβ and has greater toxicity and resistance to degradation by enzymes (Kummer and Heneka, 2014; Gnoth et al., 2020). Another important modification of Aβ, phosphorylation of Ser8, influences the neurotoxicity of the peptide and has been found in the brains of AD patients and in the brains of transgenic animals modeling AD (Kummer and Heneka, 2014; Kumar et al., 2018). We have shown that iso-Aβ₄₂ is a more potent inhibitor of α7 nAChR than the unmodified peptide (Barykin et al., 2019), and the inhibition of Na⁺/K⁺-ATPase by beta-amyloid is completely eliminated by phosphorylation of the peptide at Ser8 (Barykin et al., 2018). Thus, pS8-Aβ₄₂ and iso-Aβ₄₂ represent important isoforms whose properties are significantly different from unmodified Aβ.

The difference in the pathogenicity of beta-amyloid isoforms for neuronal cells may be due to their different effects on the mechanical properties of cells. The mechanical properties of cells are closely related to many important biological functions, such as adhesion, division, motility, differentiation and deformation, and are controlled mainly by the cytoskeleton (Luo et al., 2016). Aβ has been shown to induce changes in the cytoskeleton and mechanical properties of neuronal cells (Ungureanu et al., 2016; Kolmogorov V. S. et al., 2023), but whether the modified forms have the same properties has not been previously studied. Establishing differences in the properties and effects of beta-amyloid isoforms may be important for a complete study of the mechanism of AD development.

In this study, we measured cell stiffness and ROS production in response to Aβ₄₂, pS8-Aβ₄₂ and iso-Aβ₄₂. Next, we aimed to discover which signaling cascades associated with the regulation of cell mechanical properties through influence on the actin cytoskeleton and ROS are activated by Aβ and its isoforms. Cofilin, which plays a key role in cytoskeletal dynamics, and its regulating kinases GSK3β, ERK1/2, p38 and LIMK1 were chosen as target proteins for studying the influence of Aβ isoforms (Figure 1).

The mechanical properties of neuronal cells play a significant role in a wide range of biological functions, such as cellular homeostasis, proliferation, chemical and electrical signaling (Kolmogorov V. et al., 2023). These properties are determined mainly by cytoskeletal elements: actin cytoskeleton, microtubules, intermediate filaments (Luo et al., 2016). Oligomeric Aβ can alter cytoskeletal dynamics, which leads to changes in cell stiffness (Gao et al., 2019), and also stimulates ROS production in neuronal cells (Wang et al., 2010). In this work, the SICM method was used to measure the stiffness of SH-SY5Y cells in the presence of Aβ isoforms. Previously, using this method, it was established on SH-SY5Y cells that beta-amyloid leads to an increase in the Young’s modulus of the entire cell (Kolmogorov V. S. et al., 2023). We showed that Aβ₄₂, pS8-Aβ₄₂, iso-Aβ₄₂ cause a significant increase in cell membrane stiffness, as well as an increase in the level of ROS (Figures 2, 3), which correlates with previously published data. The effects of beta-amyloid isoforms on these parameters differed: Aβ₄₂ led to the greatest increase in stiffness after 4 h of incubation, and pS8-Aβ₄₂ led to the greatest increase in stiffness and ROS production in cells after 24 h. At the same time, it is interesting that for Aβ₄₂ and pS8-Aβ₄₂, in contrast to iso-Aβ₄₂, there is a tendency for the ROS level to increase over time.

The actin cytoskeleton is known to play a key role in synaptic transmission and plasticity (Pelucchi et al., 2020). Many studies have shown that disruption of the actin cytoskeleton using agents such as cytochalasin D, latrunculin A and latrunculin B results in a decrease in the Young’s modulus of cells (Roduit et al., 2009; Mihai et al., 2012; Pogoda et al., 2012; Louise et al., 2014; Ramos et al., 2014). It is worth noting that there is still no information on how much microtubules affect cell stiffness (Luo et al., 2016), and the possibility of intermediate filaments to influence stiffness independently of the rearrangement of actin filaments remains questionable (Gruenbaum and Aebi, 2014). Thus, the organization of actin filaments is considered to be the most important factor determining cell stiffness (Luo et al., 2016), so in our work we focused specifically on this element of the cytoskeleton.

Several actin-binding proteins are known to be altered in AD brains and animal models of AD (Pelucchi et al., 2020). Some of these important proteins are actin depolymerization factor (ADF) and cofilin. ADF/cofilin regulate actin filament dynamics (Heredia et al., 2006). ADF/cofilin promote actin depolymerization and create a new pool of G-actin monomers available for the formation of new filaments, thus increasing the rate of actin filament turnover in cells (Sarmiere and Bamburg, 2004; Bernstein and Bamburg, 2010; Rust, 2015). Since mammalian neurons contain approximately 5–10 times more cofilin than ADF (Minamide et al., 2000; Garvalov et al., 2007), we focused specifically on cofilin. Cofilin is inactivated by phosphorylation of Ser3 by LIM kinase 1 (LIMK1) and activated by dephosphorylation of Ser3 by Slingshot family protein phosphatases (SSH) (Pelucchi et al., 2020). We compared the effects of synthetic peptides Aβ₄₂, pS8-Aβ₄₂, and iso-Aβ₄₂ on cofilin and its regulating kinases. We found that iso-Aβ₄₂ activated cofilin by dephosphorylation, whereas Aβ₄₂ and pS8-Aβ₄₂ did not cause such changes. Aβ₄₀ has previously been shown to increase the levels of Ser3-phosphorylated ADF/cofilin and Thr508-phosphorylated LIMK1 (P-LIMK1), which is accompanied by neuritic degeneration and neuronal cell death (Heredia et al., 2006). However, it is worth noting that this study used fibrillar Aβ₄₀ and higher concentrations (20 μM), while soluble forms of Aβ₄₀ did not change cofilin phosphorylation. Thus, it is possible that soluble Aβ₄₀ and Aβ₄₂ do not affect cofilin phosphorylation, in contrast to post-translationally modified forms of Aβ. At the same time, Kim et al. reported a decrease in cofilin phosphorylation in the brain of AD patients and in the forebrain of APP/PS1 mice, and also showed that the addition of Aβ₄₂ oligomers to cortical neuronal cultures causes cofilin activation (Kim et al., 2013). These results are consistent with the activation of cofilin by iso-Aβ₄₂ established in our work: isomerized Asp7 is contained in more than 50% of Aβ molecules of amyloid plaques and is also enriched in the soluble fraction (Bugrova et al., 2021; Mukherjee et al., 2021), which explains the activation of cofilin in the brain of patients with AD and in vivo models of AD. The activation of cofilin only by iso-Aβ₄₂ may be one of the reasons for the greater toxicity of this isoform observed in many studies (Mitkevich et al., 2013). The researchers also note that the state of cofilin phosphorylation depends on both age and the stage of AD pathology (Barone et al., 2014). Interestingly, not only cofilin phosphorylation but also the cofilin to actin ratio influences actin filament stabilization. As the cofilin/actin ratio increases, cofilin triggers actin assembly and stabilizes the filaments (Andrianantoandro and Pollard, 2006; Bamburg and Bloom, 2009). We found that all beta-amyloid isoforms increased the cofilin/actin ratio after 4 h of incubation with the peptides, indicating polymerization of actin filaments and consistent with the increase in cell stiffness measured by SICM.

We also found differences in the effects of Aβ₄₂, pS8-Aβ₄₂ and iso-Aβ₄₂ on the activity of cofilin-regulating kinases after 30 min of incubation: inhibition of GSK3β (Aβ₄₂ and iso-Aβ₄₂), inactivation of p38 (Aβ₄₂ and iso-Aβ₄₂). This leads to actin depolymerization, but the signaling pathways that cause this effect differ. Interestingly, pS8-Aβ₄₂ appears to have a different signaling cascade via Erk activation, leading to actin polymerization. It is worth noting that this Aβ isoform had a distinctive pattern of effects on stiffness: after 4 h, pS8-Aβ₄₂ caused a significantly smaller increase in the Young’s modulus of cells than Aβ₄₂ and iso-Aβ₄₂, but after 24 h, pS8-Aβ₄₂ led to greater stiffness than the other isoforms. One of the reasons for these differences in Aβ isoforms may be the activation of different signaling cascades. It has been previously shown that some pathogenic properties of Aβ are neutralized by its phosphorylation (Jamasbi et al., 2017; Barykin et al., 2018). The activation of a different signaling pathway from other isoforms may be responsible for altered pathogenic properties of pS8-Aβ₄₂. After 4 h, a change in signaling protein activation occurs, leading to actin polymerization and corresponding to an increase in cell stiffness measured by SICM. This discrepancy, at first glance, is explained by the fact that the rupture and depolymerization of actin filaments promotes an increase in the concentration of G-actin and the formation of free barbed ends of the filaments, which is necessary for the rapid reorganization of the actin cytoskeleton and further polymerization of actin (Mizuno, 2013; Pelucchi et al., 2020). It should be noted that the role of LIMK in cofilin regulation is ambiguous. Thus, it has been shown that Aβ_1–40 and Aβ_25–35 fibrils induce LIMK activation, which leads to cofilin inactivation (Heredia et al., 2006). However, Aβ_1–42 was also found to activate LIMK1, which was paradoxically associated with increased cofilin activation, suggesting other pathways of cofilin regulation by Aβ (e.g., SSH1) (Wang et al., 2020; Kang and Woo, 2019). It has also been shown that cofilin activation is increased in AD with a simultaneous lack of changes in LIMK activation (Kim et al., 2013), which is consistent with our results.

It is known that ROS are not only damaging agents, but also mediators involved in cellular signaling and regulation (Corcoran and Cotter, 2013). ROS can affect cell stiffness in several ways (Supplementary Figure S6). In current study we focused on signaling cascades regulating actin cytoskeleton that can be induced by ROS. Thus, the activity of MAP kinases such as p38 and ERK can be regulated by ROS (Son et al., 2013; Zhang et al., 2016). Activation of p38 by ROS can affect the cytoskeleton both through cofilin (Figure 1) and through phosphorylation of the heat shock protein HSP27, which regulates microfilament dynamics, the redox state of actin, and some actin regulatory proteins (Guay et al., 1997, p. 27). At the same time, ERK, after stimulation by ROS, is able to activate actin regulatory complexes, which leads to actin polymerization (Taulet et al., 2012). Interestingly, pS8-Aβ₄₂ caused the maximum increase in ROS observed in SH-SY5Y cells (Figure 3), while only pS8-Aβ₄₂ led to ERK activation in our experiment (Figure 8). Hence, significant ERK activation by pS8-Aβ₄₂ could be in part mediated by stimulated production of ROS. The researchers also note that oxidative stress activates GSK3β (Zhang et al., 2016), which plays an important role in both actin filament regulation and microtubule regulation. However, we did not observe these effects. It is known that ROS, through the oxidation of 14–3-3zeta, activate cofilin phosphatase, which leads to dephosphorylation and activation of cofilin (Kim et al., 2009). In addition to actin, ROS can also affect cell stiffness through other cytoskeletal elements. Thus, oxidative stress suppresses microtubule-associated proteins and affects tubulin through post-translational modifications. Neurofilaments become phosphorylated during oxidative stress, leading to the formation of protein aggregates (Gardiner et al., 2013). ROS are known to activate the MAP kinase JNK via several pathways (Zhang et al., 2016). We found that Aβ isoforms activate JNK (Supplementary Figure S3), which corresponds to increased ROS levels in SH-SY5Y cells (Figure 3). Signaling pathways such as NF-κB and PI3K-Akt also depend on ROS (Zhang et al., 2016), however, we did not detect activation of NF-κB factor and Akt kinase (Supplementary Figures S4, S5). Thus, beta-amyloid peptides used in our investigation do not activate the NF-κB and PI3K-Akt pathways in SH-SY5Y cells.

Reactive oxygen species are able to influence the reorganization of the actin cytoskeleton not only indirectly (via redox-sensitive enzymes), but also by direct oxidative modifications of actin. The researchers note that actin filaments can use oxidative stress for their reorganization in a context-dependent way, as oxidative stress can either increase cellular actin aggregation (Gardiner et al., 2013) or reduce the rate of filament formation via S-glutathionylation of actin (Haseena et al., 2022). In a mouse model of AD, a role of Aβ-induced oxidative stress in actin glutathionylation and F-actin reduction has been demonstrated (Kommaddi et al., 2019). It is also noted that deglutathionylation of G-actin leads to a 6-fold increase in the rate of polymerization (Wang et al., 2001). Thus, the effect of ROS on actin polymerization and, accordingly, cell stiffness is ambiguous and may depend on the cell type and additional regulatory mechanisms. Although depolymerization of actin filaments leads to a decrease in the Young’s modulus of cells, polymerization of actin filaments is not the only factor determining cell stiffness, and it is necessary to take the spatial organization of actin filaments into account (Luo et al., 2016).

In this work, it is shown that the effects of Aβ₄₂, pS8-Aβ₄₂ and iso-Aβ₄₂ on signaling cascades associated with the regulation of mechanical properties of cells are different. This leads to the fact that, depending on the post-translational modification that the Aβ molecule acquires, beta-amyloid increases the Young’s modulus of the SH-SY5Y cell membrane and the level of ROS to varying degrees. Based on the obtained data, we proposed mechanisms mediating the effect of beta-amyloid peptide and its modified forms on the mechanical properties of neuronal cells. Thus, we showed that Aβ and its isoforms differently affect the activity of proteins that regulate the actin cytoskeleton, such as cofilin, GSK3β, LIMK, ERK and p38. We found that short-term incubation with beta-amyloid peptides activates signaling pathways that stimulate actin depolymerization, with the exception of ERK for pS8-Aβ₄₂, while longer incubation results in activation of signaling cascades that lead to actin polymerization. In this case, pS8-Aβ₄₂ has the most stable effect on cell stiffness. These results may explain the reasons for the different pathogenicity of Aβ₄₂, pS8-Aβ₄₂ and iso-Aβ₄₂ for neuronal cells, which is important for a complete understanding of AD pathology.