The American Alzheimer’s Association has estimated that ca. 55 million people worldwide are living with Alzheimer’s disease (AD) and other neurodegenerative disease leading to dementia (Alzheimer’s Association, 2023). AD is a devastating degenerative brain disease and the most common form of dementia. The pathogenic mechanism of AD is not fully understood; however, the aggregation of two different proteins in the parenchymal brain constitute the main hallmarks of the disease: the amyloid-β peptide, which forms extracellular amyloid plaques, and the tubulin-associated unit (tau) isoforms, which form intracellular hyperphosphorylated neurofibrillary tangles (Vermunt et al., 2019; Bistaffa et al., 2020; Pons and Rivest, 2022). AD is typically diagnosed in vivo when irreparable brain damage has occurred, while the definite diagnosis of AD can only be made post-mortem, upon the detection of the aforementioned aggregates through brain autopsy (Bistaffa et al., 2020). Therefore, one of the primary goals of the research in this field is to develop a sensitive, reproducible and cost-effective approach for the identification of diagnostic biomarkers, particularly in the early phases of AD, when a diagnosis based on cognitive symptoms is more uncertain and therapeutic intervention could be more efficacious (Jack et al., 2018; Bellomo et al., 2021; Bellomo et al., 2024).

The presence of amyloid-β in human brains and cerebrospinal fluid (CSF) throughout life is physiological (Broersen et al., 2010), and although its exact function is still unknown, the amyloid-β peptide is produced when β- and γ-secretase sequentially cleave the amyloid precursor protein (APP) (Müller et al., 2017; Steiner et al., 2018). APP has been found to play a role in the brain’s neuronal growth promoting the migration of neurons during the early stages of brain development (O’Brien and Wong, 2011). Furthermore, APP is involved in a variety of biological processes, including synapse maintenance, plasticity, transcriptional regulation and neuroprotection (Zheng and Koo, 2011). β-secretase initially cleaves APP in its extracellular domain, producing the C-terminal APP fragment (C99), which is then cleaved by γ-secretase at different sites releasing in the extracellular space amyloid-β species ranging from 37 to 43 amino acids. Longer amyloid-β variants, such as Aβ_1-42, exhibit a high propensity for aggregation and eventually undergo to fibrillation process (Steiner et al., 2018). The non-amyloidogenic cleavage of APP by α-secretase prevents the formation of the abovementioned species and thus the fibrillation process. Mutations in the human APP gene, in proximity to the γ-secretase cleavage site, could be responsible for the formation of the amyloid-β species with higher propensity to aggregate (Chen et al., 2017). Furthermore, age-dependent loss of regulatory mechanisms, caused by long-term inflammatory conditions, results in the dysregulation of cellular systems involved in the clearance and degradation of misfolded or damaged neuronal proteins (Höhn et al., 2020). The failure of proteostasis thus promotes the accumulation of aberrant protein aggregates that may lead to the onset of AD (Krstic and Knuesel, 2013; Bigi et al., 2024a). In the last decades, small oligomers of Aβ_1-42, formed early during the aggregation process or released from mature fibrils, have acquired increasing importance as primary toxic species in AD pathogenesis (Kayed et al., 2004; Benilova et al., 2012; Bigi et al., 2022; Limbocker et al., 2023). Moreover, amyloid-β aggregates were identified in the CSF (Kasai et al., 2013; Savage et al., 2014; De et al., 2019; Sun et al., 2021; Nirmalraj et al., 2023; Bigi et al., 2024b).

Atomic force microscopy (AFM), due to its capability for the nano-structural analysis of individual macromolecules, has extensively demonstrated to be a powerful technique for the study of the amyloid-β aggregation as well as for the morphological characterization of the amyloidogenic aggregates, such as oligomers, protofibrils and fibrils (Harper et al., 1997a; Harper et al., 1997b; Blackley et al., 1999; 2000; Kowalewski and Holtzman, 1999; Kowalewski and Holtzman, 1999; Dahlgren et al., 2002; Arimon et al., 2005; Bartolini et al., 2011; Moores et al., 2011; Jiang et al., 2012; Drolle et al., 2014; Breydo et al., 2016; Stylianou et al., 2019; Nirmalraj et al., 2020). The combination of AFM with circular dichroism spectroscopy (CD) and the measurement of the ζ-potential allowed to identify two different aggregation pathways, the amorphous and the fibrous, monitoring variations in the secondary structures and correlating each pathway with the colloidal stability of the aggregation intermediates, which is essential to the fibrillation process (Jiang et al., 2012). Moreover, the aforementioned study monitored the changes in the surface charges of the amyloid-β molecules due to Cu²⁺ binding, demonstrating that the metal ion enhances the amorphous aggregate formation. The binding of Cu²⁺ to Aβ_1-42 has been also exploited as a probe for estimating the intramolecular distances in the oligomers by double electron-electron resonance (DEER), in combination with AFM and surface enhanced Raman spectroscopy (SERS) (Banchelli et al., 2021). SERS is a powerful optical technique which allows to obtain the chemo-structural characterization of molecular species, by providing a label-free detection at sub-micromolar concentrations together with a spectral fingerprint information (D’Andrea et al., 2018; D’Andrea et al., 2023; Banchelli et al., 2019; Banchelli et al., 2020; Polykretis et al., 2022).

In this work we have examined the Aβ_1-42 fibrillogenesis timeline using AFM, Thioflavin T (ThT) fluorescence assay and SERS in order to characterize the aggregation intermediates from a morphological and spectroscopic point of view. In particular, we have monitored the progressive variations in their secondary structure during all phases of fibrillation, from the first minutes until the formation of mature fibrils. This approach allowed to identify protofibrillar species at early-stages of aggregation, which form peculiar macromolecular networks over time.

In the past few years extensive effort has been made to identify the structural determinants of the amyloid-β aggregates that are responsible for their ability to induce neurodegeneration (Hardy and Higgins, 1992; Haass and Selkoe, 2007; Benilova et al., 2012; Chiti and Dobson, 2017; Bigi et al., 2022; Limbocker et al., 2023). In this study, we employed AFM and SERS to examine the Aβ_1-42 fibrillogenesis, studying the aggregation intermediates morphologically and monitoring spectroscopically the gradual changes in their secondary structures, throughout the course of the aggregation process. We identified oligomers and small protofibrils since the first incubation time-points. This finding is in accordance with the species observed by Blackley et al. by in-situ AFM during the first 135 min of incubation (Blackley et al., 1999; Blackley et al., 2000). Furthermore, we observed that such protofibrils, interact with each other forming large macromolecular networks (Figures 1C, 3A). Interestingly, at 24 h of incubation the mature amyloid-β fibrils were frequently found lying above such structures (as shown in Figure 3A panel on the bottom left) and this could indicate that these networks of protofibrils play a role in the development of mature fibrils. The physiological and physicochemical aspects driving their formation, such as the types of interactions that take place between the protofibrils, should be further investigated, in order to elucidate their role in the aggregation pathway. Our study supports the idea of the existence of different aggregation pathways, as observed in previous studies (Jiang et al., 2012; Lipiec et al., 2018), since we detected both fibrillar structures and amorphous aggregates during all the incubation time-points. Ultimately, we were able to monitor the gradual change in the secondary structure of the aggregation species by fitting the amide I band of the SERS spectra. We tracked the change from a conformation in which the random coil had a prominent contribution, in the early stages of incubation, to β-sheet-rich structures, which become the main contributors at long incubation times. SERS results show that the morphological differences between the first rigid filaments, formed between 240’ and 480’ (Figure 2) and the earlier flexible protofibrils, formed at 60’ and 120’ (Figure 1), reflect a critical conformational change (Figure 6) and lead to the formation of mature fibrils at 24 h (Figure 3). Remarkably, the exponential (or elongation) phase begins after 8 h of incubation at 25°C (Figure 4), concurrently with the structural transition observed at 480’ by SERS, when the β-sheet conformation becomes the main contributor (Figure 6). This trend is in accordance with previous findings reporting that the amyloid-β aggregation pathway is characterized by the transition of Aβ peptides from their soluble forms into disease-associated β-sheet-rich conformers (Janek et al., 2001; Chiti and Dobson, 2017). The contribution of the α-helix initially increases, peaking at 120’, concomitantly to the decrease of the random coil conformation, and then gradually goes down. This pattern could be explained by the structural rearrangements taking place during the different stages of the amyloid-β aggregation. The Aβ_1-42 monomer contains two α-helices: S8-V24 and 28 K28-V38 (Janek et al., 2001; Crescenzi et al., 2002; Santoro et al., 2021). The α-helical structures are likely more exposed during the first 120’ of incubation, when the sample contains small-size early aggregation species, and exhibit a stronger SERS signal, with respect to the later stages of incubation, when the transition into larger β-sheet-rich species occurs. Such results align with the Aβ_1-42 aggregation kinetics described in previous circular dichroism investigations (Bartolini et al., 2007; Jiang et al., 2012; De Simone et al., 2019); slight variations may arise due to different protein concentrations and aggregation protocols, but overall trends are in agreement.

In this study, we provide the significance of investigating the early phases of fibrillogenesis, to better understand the molecular pathophysiology of AD and identify potential pharmaceutical targets that could prevent or slow down the aggregation process. Our findings indicate that there is a peculiar step in the fibrillogenesis timeline when protofibrils serve as a template for the formation of larger size fibrils, which is characterized by a transition in the conformation of the secondary structure. In fact, when AFM topography starts to reveal the presence of such larger size fibrils, the organized β-sheet-rich structures spectroscopically prevail over the random coil conformation. The critical role of protofibrils in vitro strongly suggests that they might be crucial in the formation of amyloid fibrils in the brains of AD patients as well. Thus, the development of new therapeutic molecules able to inhibit the protofibril to fibril transition in vitro might lead the way for the development of therapies against AD. Finally, we highlighted how a direct, label-free and fast optical technique such as SERS, can be exploited for chemo-structural investigation of the Aβ_1-42 aggregation process, as well as for other misfolded proteins, using minimal volumes of sample.