The deposits and inclusions in neurodegenerative disorders such as NFTs, senile plaques, Lewy bodies… consist of several proteins and non-proteinaceous components (carbohydrates, nucleic acids, metals, lipids, lipid rafts and cholesterol) that could be linked to the amyloid polymorphisms observed in diseases (Kollmer et al., 2016; Stewart et al., 2016; Shahmoradian et al., 2019; Li and Liu, 2021). The proteome analysis of amyloid deposits has revealed hundreds of proteins (Drummond et al., 2017; Lutz and Peng, 2018). Some of them act as critical regulators in protein misfolding diseases exemplified by heat shock proteins (HSPs) and their co-chaperones, 14-3-3 proteins (Xu et al., 2013; Jia et al., 2014), S100B calcium-binding protein (Moreira et al., 2021), and the peptidyl-prolyl isomerases (PPIases) FKBPs and Pin1 (Hamdane et al., 2002; Landrieu et al., 2006a; Balastik et al., 2007; Lippens et al., 2007; Chambraud et al., 2010; Giustiniani et al., 2012; Giustiniani et al., 2014; Kamah et al., 2016; Chen et al., 2018; Wang et al., 2020a). Notably, Pin1 and 14-3-3 proteins interact with phosphorylated forms of tau making the link with amyloid PTMs (Lu et al., 1999; Zhou et al., 2000; Smet et al., 2004; Lim and Ping Lu, 2005; Smet et al., 2005; Pastorino et al., 2006; Landrieu et al., 2011; Kondo et al., 2017; Neves et al., 2021). A role of cofactors has been shed into light in the aggregation of tau protein to specifically address the challenges of forming amyloid fibrils from full-length tau in vitro (Fichou et al., 2018; Fichou et al., 2019) in contrast to Aβ or α-synuclein that readily form amyloid fibrils in a wider range of conditions. Glycosaminoglycans, lipid membranes and metal ions are key cofactors that were pointed out in the amyloidogenic process. They have been found to modulate aggregation rates and are associated with amyloid deposits within the brain. Interactions of amyloidogenic species with cofactors may represent an orthogonal strategy of intervention to aggregation inhibitors in neurodegenerative disorders.

The polymorphism of amyloid structures from the same protein reflects distinct environments leading ultimately to different diseases. In this respect, PTMs and non-amino acid components associated with the fibrils have focused particular attention (Figure 2). Arising from the most recent cryo-EM structures of human prion PrP, wild-type α-synuclein from multiple system atrophy (MSA), tau from corticobasal degeneration (CBD) - all of which were from brains of patients- and phospho-Tyr39 (pY39) α-synuclein from semisynthesis (Arakhamia et al., 2020; Wang et al., 2020b; Schweighauser et al., 2020; Zhang et al., 2020; Zhao et al., 2020), it has been proposed to categorize PTMs based on their location with respect to the fibril core (Li and Liu, 2021). PTMs in the interior of the core are likely involved in the initial step of fibril assembly while PTMs on the exterior may act rather in the polymorph selection either by driving the folding of protofilaments or stabilizing the protofilament interface. The role in fibril assembly of PTMs outside the core, within the “fuzzy coat”, still remains poorly defined although they are known to regulate the protein functions, interactions and aggregation properties by modulating the rate of fibrillar assembly, toxicity and phase separation. Importantly, they remain accessible even within the amyloid fibril and may still be targeted by posttranslational modifying and proteolytic enzymes (Wegmann et al., 2013; Ulamec et al., 2020).

Hyperphosphorylation, a common feature of tau proteins in NFTs and inclusions from diverse tauopathies, is mainly found within the N- and C-terminal regions flanking the amyloid core (Morishima-Kawashima et al., 1995a; Hanger et al., 1998; Alonso et al., 2001; Hanger et al., 2007; Hanger et al., 2009; Šimić et al., 2016) that greatly inhibit the formation of filaments (Abraha et al., 2000; Lövestam et al., 2021). Tau also exhibits a large diversity of PTMs such as lysine ubiquitination, SUMOylation, acetylation and methylation (Morris et al., 2015; Wesseling et al., 2020). These modifications can be specifically linked to the disease stage and mediate the structural diversity of tau strains (Arakhamia et al., 2020). Interestingly, the seeding activity of the hyperphosphorylated, oligomeric tau species was found to be heterogeneous from one patient with pure, typical AD to another while enhanced seeding activity and worse clinical outcomes both correlate with specific PTM sites (Dujardin et al., 2020). Whereas phosphorylation of α-synuclein at serine 129 (pS129) is a dominant characteristic of PD inclusions such as Lewy bodies, its role in aggregation and toxicity of α-synuclein has not yet been clearly established (Anderson et al., 2006; Oueslati, 2016; Ghanem et al., 2022). TDP-43, the main component of intracellular ubiquitin inclusion bodies found as a hallmark of ALS-FTLD (Frontotemporal Lobar Degeneration), is hyperphosphorylated and polyubiquitinated whereas these PTMs were not detected in normal brain (Dong and Chen, 2018).

Commonly associated to phosphoproteins, the O-β-linked N-acetylglucosaminylation (O-GlcNAc) is another PTM of serine/threonine residues that corresponds to the addition of a single sugar moiety that is regulated in a dynamic fashion by the antagonist action of two enzymes, the O-GlcNAc transferase (OGT) and O-GlcNAc hydrolase (OGA) (Iyer and Hart, 2003). Protein O-GlcNAcylation is extremely sensitive to glucose uptake and metabolism that may be altered in aging brain. Additionally, the O-GlcNAc modification has been reported for neuronal proteins such as APP, tau and α-synuclein highlighting a potential role in neurodegenerative diseases (Lazarus et al., 2009; Ma et al., 2017). O-GlcNAcylation of amyloid-forming proteins has been shown to regulate aggregation (Yuzwa et al., 2014a) and to some extent, phosphorylation of tau (Liu et al., 2004; Gong et al., 2006; Liu et al., 2009; Smet-Nocca et al., 2011; Bourré et al., 2018; Cantrelle et al., 2021). The treatment of transgenic mice with Thiamet-G, an OGA inhibitor, results in increased brain O-GlcNAc levels, and alleviate tau pathology and associated neurodegeneration offering an alternative therapeutic strategy to kinase and aggregation inhibitors in tauopathies (Yuzwa et al., 2008; Yuzwa et al., 2012; Yuzwa et al., 2014b; Graham et al., 2014; Hastings et al., 2017; Lee et al., 2021).

In addition to PTMs involving the covalent linkage of proteinaceous or small chemical entities, truncated forms of amyloid proteins are also frequently found associated to pathological transformation. The N-terminal region of Htt is the site of HD-associated pathogenic changes through an elongation of the CAG repeat of htt gene encoding an expanded polyglutamine repeat. The truncated N-terminal proteoforms are more toxic than full-length Htt and form intranuclear inclusions that disrupt synaptic and axonal functions (Sun et al., 2002). At the basis of the amyloid cascade hypothesis in AD, the Aβ peptide is the product of the sequential cleavage of the transmembrane APP protein by secretases that generates peptides of different length, the most common forms being Aβ40 and Aβ42. With a proportion significantly increased in AD brain, this latter form is the most neurotoxic and readily form oligomers and fibrils in a wide range of conditions (Nirmalraj et al., 2020; Yang et al., 2022). A large panel of tau fragments resulting from cleavage at N- or C-terminal regions or both are found in fibrillar structures, cerebrospinal and interstitial fluids, and plasma of patients with different tauopathies (Quinn et al., 2018; Boyarko and Hook, 2021). The predominant role of proteolysis in various neurodegenerative disorders deserves particular attention for characterizing the fragments, their toxicity related to aggregation and transcellular spreading, and their role in the selection of strain polymorphs. This approach could afford new biomarkers and disease-modifying therapies by modulating the fragment generation and associated toxicity (Rodriguez Camargo et al., 2021).

The wide range of disease-associated modifications represents several challenges: 1) discriminating between physiological and disease-associated changes, 2) characterizing PTMs in terms of site-specific identification, quantification, and crosstalk between PTMs, and their relevance to disease, 3) characterizing PTM-induced conformational changes, 4) identifying enzymes responsible of the installation/removal of specific PTMs and defining the pathway of their (dys)regulation, 5) determining the functional role of specific PTMs in physiology and pathology, and 6) identifying and deciphering the role of amyloid interacting entities or cofactors (Kametani et al., 2020). The role of proteins, non-proteinaceous entities and PTMs in the aggregation process, toxicity and spreading of various species that form during the fibrillization course still deserves further investigations (Figure 2). They may have prominent implications in modulating nucleation, aggregation rate, selection of fibril polymorph, seeding capacity, and amyloid toxicity. Additionally, interactions of amyloid-prone proteins with cofactors and posttranslationally modifying enzymes could be valuable targets for therapeutic intervention (Dujardin et al., 2020).

Either a three-fragment ligation or chemoenzymatic semisynthesis approach was employed to achieve tyrosine phosphorylation of α-synuclein at Y39 to evaluate its impact on aggregation properties and toxicity (Pan et al., 2021; Pan et al., 2020; Dikiy et al., 2016). Phosphorylation of Y39 was made possible by the chemoenzymatic strategy because the three other Tyr residues of α-synuclein are all located in the C-terminus that was recombinantly expressed in bacteria while the N-terminal part was enzymatically phosphorylated before ligation. Furthermore, desulfurization following the ligation steps restores native residues at junction sites considering the absence of cysteine in the native sequence of α-synuclein. It is noteworthy that recombinant expression of all fragments further assembled by ligation allowed the uniform ¹⁵N isotopic labeling of the site-specific pY39 α-synuclein for NMR studies (Vogl et al., 2021). It has been shown using these strategies that pY39 accelerates aggregation kinetics (Pan et al., 2020; Dikiy et al., 2016) and alters fibril morphology of α-synuclein with pY39 attracting the entire N-terminus within the fibril core through an array of electrostatic interactions giving rise to the largest fibril core (residues 1–100) ever seen for the α-synuclein amyloid fibrils (Figure 2) (Zhao et al., 2020). The homogenously phosphorylated α-synuclein at S129, one of the major pathological marks of Lewy bodies in PD, forms a different amyloid fold and has different propagation properties related to the wild-type protein, indicating the formation of a distinct strain associated to higher toxicity (Ma et al., 2016). Be it either localized at pY39 or pS129, single phosphorylation of α-synuclein provided by NCL highlights a capacity of site-specific phosphorylation to modulate the fold of the amyloid structure and pathological strain properties. In contrast, phosphorylation at Y125 does not alter the structure and morphology of the α-synuclein fibrils (Hejjaoui et al., 2012).

As α-synuclein has multiple O-GlcNAcylation sites, the site-specific O-GlcNAcylation provided by EPL underscored its inhibitory role in fibril assembly on either T72, T75, T81 or the three together, or S87, with the strongest effect observed for T75, all these sites being embedded within the fibril core (Marotta et al., 2015; Lewis et al., 2017; Li et al., 2018b; Zhang et al., 2019b; Levine et al., 2019; Tavassoly et al., 2021). Additionally, the combination of three T72/T75/T81 GlcNAc O-glycosylation reduces the aggregation of wild-type α-synuclein or aggregation-prone mutant (A53T) that causes early-onset, familial PD. O-GlcNAcylation alters the structure of aggregates in a site-specific manner, reduces the cytotoxicity of extracellular fibrils and impaired the calpain-mediated proteolysis of α-synuclein. Interestingly, the Pratt’s group has highlighted that O-GlcNAc at T72 is unique in its aggregation inhibitory properties as compared to other single sugars including O-GalNAc suggesting an effect that extend beyond its mere bulky size and polar properties, owing likely to the chirality of specific asymmetric carbons that would require further investigations (Galesic et al., 2021).

Finally, the investigation of site-specific ubiquitination of α-synuclein involved a strategy of disulfide-directed ubiquitination, as implemented by Pratt and co-workers, requiring a lysine-to-cysteine point mutation for installation of ubiquitin through a disulfide bond instead of the native isopeptide bond. This strategy involved the recombinant production of a ubiquitin-intein fusion protein followed by reaction with cysteamine that simultaneously affords intein cleavage and thiol functionalization of ubiquitin C-terminus (Figures 4A,B). The subsequent thiol activation as a disulfide or a redox non-labile function by reaction with dibromoacetone provided ubiquitinated α-synuclein by reaction with a free thiol function positioned at diverse biologically relevant sites (K6, K10, K12, K21, K23, K32, K34, K46, and K96) (Meier et al., 2012; Abeywardana et al., 2013; Moon et al., 2020; Lewis et al., 2016). Whatever the ubiquitination or polyubiquitination site, an inhibition of α-synuclein aggregation was observed (Meier et al., 2012; Haj-Yahya et al., 2013). However, the disulfide bond is not chemically stable and other approaches were used to establish stable isopeptide linkage of ubiquitin. An alternative strategy formerly developed by Lashuel and co-workers although less flexible affords a native isopeptide linkage between K6 of α-synuclein and ubiquitin (Figures 4A,C). The synthesis of a thiol protected δ-mercaptolysine derivative is used in SPPS of a N-terminal fragment, then EPL with a recombinant C-terminal fragment provides the modified K6 within full-length α-synuclein. Subsequent NCL between the thiol function of δ-mercaptolysine and ubiquitin C-terminal thioester forms a native isopeptide bond between ubiquitin and the targeted lysine after subsequent desulfurization of the thiol handle of δ-mercaptolysine (Hejjaoui et al., 2011).

Aβ peptides from AD brain contain a variety of peptide lengths and post-translationally modified forms with phosphorylation, isomerization, and pyroglutamate, that modulate aggregation and propagation properties, and toxicity. In AD, modifications of the N-terminus in several Aβ subtypes were shown to both accelerate fibrillation and stabilize fibril structures. Phosphorylation of S8 is a modification of Aβ featured in late-stage of AD in the dispersed, membrane- and plaque-associated fractions (Kumar et al., 2011; Rijal Upadhaya et al., 2014). Homogenous phosphorylation of S8 of Aβ(1–40) peptide, afforded by SPPS, induces a higher cross-seeding efficiency when compared to unmodified Aβ(1–40) and a modification of fibril structure shown by ssNMR that highlights a tight N-terminus association with the amyloid core (Hu et al., 2019). In contrast, selective enzymatic modification of S26 stabilizes oligomeric forms but inhibits aggregation. Hence, it has been proposed that an ordered N-terminal region may be favorable to pS8-Aβ(1–40) strain to propagate to multiple Aβ subtypes by exerting a dominant role in fibril morphology. Furthermore, ordered N-terminal conformations in Aβ fibril structures may be biologically relevant as illustrated by brain-extracted Aβ40 and Aβ42 fibril structures. In contrast, Y10 O-glycosylation of Aβ with a (Galβ1-3GalNAc) disaccharide or sialylated trisaccharide (NeuAcα2-3Galβ1-3GalNAc) was shown to destabilize the amyloid structure to form a new fibril polymorph with a less stable protofilament interface rendering fibrils more prone to degradation in agreement with short tyrosine O-glycosylated forms found in CSF of AD patients (Liu et al., 2021).

Recently, the O-GlcNAc glycosylation of small heat shock proteins (sHSPs) by EPL in the Pratt’s group has extended targeting modulators of amyloid aggregation to site-specific PTMs of their cofactors. O-GlcNAc modification was found to selectively improve the anti-amyloid activity of HSP27 and other sHSPs in aggregation of both α-synuclein and Aβ(1–42), by competing with intramolecular interactions. The subsequent conformational rearrangement upon HSP27 O-GlcNAcylation most likely favors the formation of larger multimers with increased activity (Balana et al., 2021). Besides highlighting a protective role of O-GlcNAc in amyloid aggregation, this work opens new avenues for orthogonal strategies to inhibit the formation of amyloid fibril by targeting their cofactors.

The long sequence of tau protein combined to a high proportion of polar residues (Ser, Thr, Lys, Arg) and Pro in the proline-rich domain coincides with a large number of PTMs such as phosphorylation, acetylation, ubiquitination, SUMOylation, methylation and O-GlcNAcylation. Hyperphosphorylation was shown to have a dominant role in tau fibril formation and loss-of-function in MT stabilization (Alonso et al., 1994; Morishima-Kawashima et al., 1995b; Alonso et al., 2001; Tepper et al., 2014), but other PTMs may have important effects in modulating tau functions and amyloid assembly. Site-specific PTMs may exert antagonist roles as well with some phosphorylation sites promoting aggregation while others have an inhibitory effect (Schneider et al., 1999; Liu et al., 2007; Despres et al., 2017; Brotzakis et al., 2021). Tuning on and off specific phospho-epitopes was shown to modulate tau function on tubulin polymerization or the Pin1-mediated regulation of tau dephosphorylation by PP2A (Amniai et al., 2009; Landrieu et al., 2011). The use of enzymes for introducing a limited pool of PTMs generally requires multiple mutations into an amino acid that cannot be modified as exemplified by the restriction of phosphorylation patterns to single phospho-epitopes, such as AT8 and PHF-1, for functional investigations (Amniai et al., 2009; Landrieu et al., 2011; Despres et al., 2017; Cantrelle et al., 2021). Alternatively, mimetics of PTM-amino acid may be inserted at single or multiple place mimicking a permanently modified state (Eidenmüller et al., 2001; Min et al., 2015). First attempts of full-length tau EPL semisynthesis were made by Hackenberger and co-workers, for the introduction of phosphorylation on S404 in the C-terminal region by ligation of a C-terminal phosphorylated peptide (from residue 390–441) to a N-terminal recombinantly expressed fragment (from residue 1–389) fused to intein (Broncel et al., 2012). Further improvement of the ligation strategy was required to afford the triple phosphorylation of PHF-1 epitope (pS396/pS400/pS404) or S400 O-GlcNAcylation owing to the length of the synthetic peptide (52 residues) (Reimann et al., 2015; Schwagerus et al., 2016). Hence, the synthesis of the C-terminal peptide involved NCL of two fragments followed by desulfurization, then EPL with the recombinant N-terminal fragment provides the full-length protein. A traceless purification strategy was also implemented through the use of a photocleavable biotin handle to purify the ligation product (Figure 5).

The presence of two native cysteine in tau primary sequence prevents from desulfurization of the EPL product. To circumvent this issue, the Lashuel’s group has exploited the native cysteine residues (C291 and C322) in a three-fragment ligation strategy for the synthesis of the tau K18 fragment (residues 243–372) corresponding to the microtubule-binding repeats (MTBR) containing four repeat/inter-repeat sequences, a region overlapping the core of amyloid fibrils from diverse tauopathies. This approach involves the native cysteine at the NCL junction sites preventing any introduction of non-native cysteine. The total chemical synthesis of K18 allowed the installation of phosphorylation at single (pS356) or multiple sites (pS262/pS356 and pS258/pS262/pS356) known to modulate tau function in microtubule polymerization/binding and amyloid assembly (Schneider et al., 1999). The site-specific phosphorylation of K18 proved to inhibit its aggregation in vitro, its seeding property in biosensor cells, and its ability to promote microtubule polymerization (Haj ‐ Yahya et al., 2020). The same strategy including first NCL of a C-terminal synthetic fragment was used in combination with an expressed N-terminal tau fragment (residues 2–245) for EPL allowing the semisynthesis of full-length tau after ligation of five fragments with either pY310, Ac-K280 or pS396/pS404 modification (Haj-Yahya and Lashuel, 2018). To obtain wild-type tau, two desulfurization steps were performed restoring native alanine at ligation sites before the incorporation of both fragments bearing native cysteine residues. The site-specific acetylation of K280 reproduces with increased outcome the effect of K280Q acetylation mimicking mutant on tau aggregation by forming rather oligomers and short fibrils instead of the long filaments observed for unmodified tau consistently with the increased seeding efficiency and toxicity of this mutant in AD models of tau pathology (Min et al., 2010; Cohen et al., 2011; Gorsky et al., 2016).

PHF-tau isolated from AD brains is conjugated to ubiquitin at multiple lysine residues within the MTBR, such as monoubiquitination at K254, K257, K311 and K317, and K48-linked polyubiquitination at K254, K311 and K353. Hence, the chemoselective disulfide coupling reaction implemented for ubiquitination of α-synuclein was similarly employed for the site-specific monoubiquitination of the K18 tau fragment at K254, K311 or K353 in which both native cysteine were mutated into alanine to prevent unwanted conjugation of ubiquitin. In contrast to multiple enzyme-derived ubiquitination that prevents the formation of both prefibrillar oligomers and amyloid fibrils, the semisynthetic ubiquitinated tau conjugates form oligomers and filaments to different extent with ubiquitination at K311 position being the strongest inhibitor (Munari et al., 2020).

Mutant Huntingtin as a primary cause of Huntington’s disease is characterized by an abnormal expansion of polyglutamine (poly Q) tract at the N terminus which has been described as pathological when exceeding 40 glutamine repeats instead of the normal 9 to 35. The length of the expanded polyQ tract was suggested to be proportional to disease severity. The first 17 N-terminal residues of Htt act in modulating Htt structure, interactions and aggregation. Hence, polyQ expansion was shown to cause misfolding of the N-terminus and could play a key role in aggregation and toxicity. The polyQ expansion in mutant Htt exon1 causes conformational changes and increases phosphorylation at multiple sites across the entire protein (Jung et al., 2020). It was suggested that phosphorylation of Htt N-terminus could reverse the conformational rigidity related to polyQ expansion and improve the toxic properties. Phosphorylation of the N-terminal fragment of Htt at T3 homogeneously obtained by EPL, either in wild-type (23Q) or mutant Htt protein (43Q) was shown to stabilize the α-helical conformation of the N-terminal 17 amino acids and significantly inhibit aggregation while K6 acetylation reverses the inhibitory effect of pT3 without exhibiting any intrinsic inhibitory effect by itself (Ansaloni et al., 2014; Chiki et al., 2017). More recently, a new chemoenzymatic semisynthesis of Htt N-terminal fragment (1–170) that forms nuclear and cytoplasmic inclusions in cell and animal models of HD enables the introduction of phosphorylation in exon3 (Kolla et al., 2021). Phosphorylation of T107 was shown to have opposite roles on Htt(1–171)-43Q aggregation depending on the phosphorylation status of S116 highlighting the benefits of site-specific PTM installation through semisynthesis in deciphering the PTM code that regulates the aggregation properties of amyloidogenic proteins.

The semisynthesis of TDP-43 prion-like domain bearing a site-specific phosphorylation at S404 required the use of denaturing conditions to solubilize protein fragments expressed in bacterial expression systems. Indeed, the strategy of EPL through purification of full-length TDP-43 or prion-like domain as protein thioesters by the traditional fusion intein-CBD was unsuccessful to produce soluble TDP-43 variants in contrast to α-synuclein, Htt or tau proteins. Hence, the selected approach made use of a polyhistidine tag attached to a C-terminal cysteine residue which is submitted to S-cyanylated cysteine-promoted C-terminal hydrazinolysis to remove the tag. The tag removal then leaves a reactive C-terminal hydrazide that can be converted into a thioester for further NCL with the C-terminal pS404 fragment afforded by SPPS. Using this procedure, phosphorylation of S404 of TDP-43 prion-like domain was shown to accelerate the amyloid aggregation of the TDP-43 prion-like domain and worsen cytotoxicity (Li et al., 2020a).

Although the semisynthesis of post-translationally modified proteins has taken advantage of the NCL/EPL strategy, its major drawback is related to the need for a combined expertise in both peptide synthesis and chemical ligation reactions, and when applicable, the refolding of protein fragments of interest may be limiting. Expansion of the genetic code using amber stop codons and an orthogonal tRNA/tRNA synthase pair is an alternative strategy to introduce site-specific PTMs such as phosphoserine, acetyl-lysine, mono-/di-methyl-lysine, and UAAs (Wang et al., 2006; Tarrant and Cole, 2009; Park et al., 2011; Ge and Woo, 2021). However, this route has not been extensively applied so far to amyloid proteins. Finally, a more flexible method analogous to NCL implemented the sortase-mediated ligation (SML), or sortagging, which benefits from the transpeptidase activity of bacterial Sortase A (SrtA) enzymes that recognize with a high substrate specificity LPXT motifs and conjugate them to oligoglycine units through a native peptide bond (Dai et al., 2019; Li et al., 2020b). This versatile ligation approach demonstrates broad applications in vivo and in vitro for protein engineering, for instance by ligating a peptide substrate bearing PTMs, UAAs or other labeling probes (obtained from SPPS) into sortagging reactions. The directed evolution of SrtA to recognize the LMVGG sequence of Aβ (residues 34–38) enabled labeling of endogenous Aβ in human CSF for sensitive detection and conjugating a hydrophilic peptide to Aβ42 that blocks amyloid aggregation (Podracky et al., 2021). Dual-color fluorogenic aggregation sensors were introduced by SML to monitor protein co-aggregation in transthyretin amyloidosis by a thermal shift assay (Bai et al., 2021).

The site-specific chemical installation of PTMs made use of the reactivity of cysteine thiol group and requires first the site-directed mutagenesis of the PTM site to introduce a non-native cysteine residue, and eventually replacing native cysteine(s) by serine or alanine residues. An acetyl-lysine analog can be introduced through Lys-to-Cys mutation at a specific site and subsequent thiol alkylation with methylthiocarbonylaziridine leading to a thioether bond bearing a terminal S-methyl thiocarbamate group as acetyl mimic (Huang et al., 2010). Other chemical approaches for the site-specific ubiquitination of proteins, comprise the formation of thioether bonds (Hemantha et al., 2014) or triazole moiety through Copper(I)-catalyzed azide-alkyne cycloaddition (CuAAC) (Rösner et al., 2015) as stable analogs of the ubiquitin isopeptide bond. Mono- and di-methylated (symmetric or asymmetric) arginine-containing proteins can be prepared by reaction of cysteine thiol with α,β-unsaturated amidines leading to the corresponding methylated arginine analogs (Le et al., 2013). The chemical installation of mono-, di- and tri-methyl-lysine analogs can be provided by thiol alkylation with the respective N-methylated (2-chloroethyl)-ammonium halide (Simon et al., 2007). However, none of these reactions were applied so far to amyloid proteins to our knowledge.

Besides NCL, an increasing number of chemical methods has been developed to encode or decode PTMs in various proteins, as extensively described in excellent reviews (Chuh et al., 2016; Ge and Woo, 2021; Yang et al., 2018). A versatile approach is the chemical editing of amino acids into protein after their synthesis, thus independently of the ribosome or any enzymatic machinery. Dehydroalanine (Dha)/dehydrobutyrine (Dhb) are versatile chemical precursors to a range of PTMs including phosphorylation, glycosylation, methylation, acetylation, lipidation, and their analogs (Figure 6A). Dha/Dhb precursors can be introduced chemically or genetically into proteins (Jones, 2020; Yang et al., 2019). These reactive Ser/Thr analogs allow following olefin addition chemistry introducing a wide variety of PTMs in a strategy named “posttranslational chemical mutagenesis” as well as UAAs or molecular probes for purification or detection purposes. A simple way to obtain Dha site-specifically is through β-elimination of the cysteine thiol function that requires 1) the site-directed mutagenesis of the PTM target site to install a cysteine precursor and 2) the Cys-to-Dha conversion through a bis-alkylation elimination reaction (Figure 6A) (Chalker et al., 2011). Although the Cys-to-Dha conversion was used to map cysteine accessibility in a globular protein (Bertoldo et al., 2017), all cysteines are equally reactive in IDPs and, whenever needed, native cysteines must be mutated. An alternative strategy is the genetic introduction of Dha precursors reactive to β-elimination, like phosphoserine, through genetic code expansion using amber codon suppression (Yang et al., 2016).

Subsequent to Dha formation, the installation of PTMs or PTM analogs is provided by Michael addition (Figure 6A). Thiol derivative precursors of PTMs are the most convenient way to introduce PTM but the formation of the thio-ether bond induces Cα racemization (and also at Cβ in the case of Dhb derivatives) which seems not to be an issue in most amyloid case studies. This strategy was employed to introduce PTMs in full-length tau protein like pseudo-phosphorylation at pS262, pS356, pS199 by reaction with thiophosphate (to generate a phosphocysteine) and lysine K311 acetylation (K311-Ac) and dimethylation (K311-diMe) by reaction with N-acetylcysteamine and captamine (to generate thio-ether mimetics of PTM-lysine), respectively (Figure 6B) (Brotzakis et al., 2021; Lindstedt et al., 2021). While K311-diMe and K311-Ac have little effect on the formation of microtubules, pseudo-phosphorylation of pS262 and pS356 in the MTBR as well as pS199 in the proline-rich region were shown to inhibit the polymerization activity of tau confirming the role of pS262/pS356 that was previously investigated by in vitro kinase phosphorylation and highlighting a novel functional role for pS199 that is located far outside the MTBR, in the proline-rich domain (PRD). This is not surprising however, since other AD-specific phosphorylation epitopes embedded in the PRD were shown to modulate microtubule assembly (Amniai et al., 2009). Alternative elegant and versatile strategies to create C-C (sp3-sp3) bonds instead of unnatural carbon-heteroatom bonds were described, albeit not applied so far to amyloid proteins. They made use of mild, carbon-centered free radical chemistry that are compatible with Dha precursor reactivity and aqueous conditions suitable for protein chemical reactions. Most importantly, free radicals are unreactive with most of functionalities found in biomolecules. The reaction of Dha with a wide range of free radical precursor halides generates a stabilized Cα free radical that is further quenched after formation of the C-C bond allowing the installation of side chain chemical diversity (unnatural, fluorinated, PTM or isotopically labeled amino acids) at a specific position with a high site- and regio-selectivity (creation of Cβ-Cγ bonds), albeit with Cα racemization (Wright et al., 2016; Aguilar Troyano et al., 2021).

Historical fluorescent probes for the detection of amyloids (e.g., ThS, ThT, ANS, Bis-ANS, Congo Red, Nile Red…) (Figure 8A) (Naiki et al., 1989; Biancalana and Koide, 2010; Krebs et al., 2005; Groenning, 2010; Amdursky et al., 2012) were recently complemented by a wide range of aggregation-induced emission (AIE) molecules (Tang et al., 2022; Tang et al., 2021) including hexaphenylsilole (HPS), tetraphenylethylene (TPE) and tetraphenylbutadiene that expand the detection of aggregates formed along the amyloidogenic pathway such as oligomers that are poorly detected by ThT, Congo Red and their derivatives. AIE molecules usually possess twisted structures in solution limiting the fluorescence yield by non-radiative transition (Kumar et al., 2017; Aliyan et al., 2019). Restricted intramolecular motions upon binding to amyloid β-sheets alleviates this phenomenon due to electrostatic and/or hydrophobic interactions, structural complementarity and/or chemical reactions (Figure 8B). For instance, increased fluorescence signal of new AIE probes with the size increase of Aβ aggregates enables a sensitive detection of Aβ aggregation and inhibition (Gour et al., 2019). Development of AIE probes includes the coverage of a large wavelength spectrum from the visible to the near-infrared (NIR) (Figure 8C). NIR probes are more convenient for super-resolution imaging of amyloids in cells and tissues by increasing the spatiotemporal resolution and photostability while reducing tissue damages. Probes for excited state intramolecular proton transfer (ESIPT) or a series of boron dipyrromethene (BODIPY) derivatives were designed with AIE properties for the follow-up of amyloid aggregation (Figure 8C) (Hu et al., 2011; Liu et al., 2019). A fluorescent BODIPY-based binuclear Zn(II) complex was rationally designed as a molecular probe with phospho-selective binding properties for peptides presenting phosphorylated groups at the relative i and i+4 positions. This molecule allowed for a selective detection of NFTs of hyperphosphorylated tau proteins in AD brain as it discriminated between NFTs and SPs of Aβ peptides in histological imaging of hippocampus sections from AD patients (Ojida et al., 2009). The luminescent amyloid-binding conjugated poly- and oligothiophenes (LCPs and LCOs) is a class of amyloid dyes of fibrillar aggregates for emission fluorescence staining or imaging (Figure 8C) (Åslund et al., 2009; Brelstaff et al., 2015; Klingstedt et al., 2011). LCOs are uniquely capable to discriminate different molecular architectures. The aggregate-specific emission is achieved due to conformational restriction of the thiophene backbone upon interaction with specific protein aggregates. Using the LCOs fluorescence signature, fibril polymorphism can be monitored both in vivo and in vitro. LCOs will be useful to link protein conformational features with disease phenotypes for a variety of neurodegenerative proteinopathies (Magnusson et al., 2014). Another series of molecular rotor fluorophores named AggFluor derived from the chromophore core of green fluorescent proteins (GFPs) was rationally designed with a gradient of viscosity sensitivity over a wide range (Figure 8D). AggFluor probes were capable of differentiating between soluble oligomers and insoluble aggregates through distinct turn-on fluorescence. This method was extended to dual-color imaging of aggregation and its modulation thereof with small-molecule proteostasis regulators in living cells (Wolstenholme et al., 2020).

Probing of amyloids with gold nanoparticles or nanorods, or through immunogold labeling as gold nanoparticle conjugates to a specific antibody allowed mapping amyloid fibrils in tissue sections as well as identifying proteins and some of their molecular features. The presence of PTMs and cofactors, or the presence/accessibility of a particular sequence can be characterized in this manner (Fitzpatrick et al., 2017; Falcon et al., 2018b; Goedert et al., 1996; Al ‐ Hilaly et al., 2020). However, these methods do not provide detailed information on the fibril morphology and polymorphism owing to either the bulkiness of the particle itself or to the distance between the particles and coated fibrils due to the size of the antibody and/or nanoparticle attachment linker. Functionalization of gold nanoparticles of 3 nm-diameter with negatively charged ligands (e.g. 11-mercapto-1-undecanesulfonate: 1-octanethiol mixture) enables to detect amyloid polymorphism over a variety of amyloid sequences under hydrated conditions by cryo-EM imaging (Figure 8E) (Cendrowska et al., 2020). The functionalized particles act as contrast agents to rapidly stain fibrillar structures. Fibrils are stained either at the fibril edges as exemplified with Aβ40 and tau R2 peptides, or on the whole fibril surface as for proteins tau and Htt exon 1 with a 43 polyQ segment, probably through interactions with the fibril fuzzy coat. This nanoparticle labeling facilitates the characterization of fibril morphology thereby revealing distinct polymorphs. Moreover, differential fibril decoration also reflects distinct morphologies. The functionalized gold nanoparticles have highlighted a higher homogeneity of periodicity length distribution for in vivo related to synthetic fibrils in agreement with cryo-EM structures.

Site-specific labeling of amyloid proteins allows to track conformation dynamics, interactions, modifications and seeding aspects by both in vitro approaches based on biophysical experiments and cell-based assays through microscopy imaging. Moreover, protein fluorescent labeling can be helpful in the screening of compounds for anti-aggregation or disaggregation activity by avoiding displacement of external fluorescent probes by competitive binding. The labels can be introduced with approaches similar to introduction of PTMs as detailed in paragraph 2 by incorporation of the labeled amino acid, or an amino acid amenable to bioorthogonal click chemistry (Arsić et al., 2022; Shimogawa and Petersson, 2021), by UAA mutagenesis, or based on suppression of stop codons. Another strategy of label incorporation is the conjugation of SPPS (peptide modified with a label) combined with NCL (see paragraph 2.1). In addition, the labels can be introduced using genetically engineered reactive tags for N- or C- terminal labeling, such as SNAP-tag (Keppler et al., 2003) CLIP-tag (Gautier et al., 2008) and Halo-tag (Liu et al., 2018) that react with derivatives of O-benzylguanine bearing a chemical probe to label the protein via a benzyl linker (Figure 8D). SNAP and CLIP or Halo tags can be simultaneously and differentially labeled with fluorescent reporters in living cells (Gautier et al., 2008; Jung et al., 2019). Compared with fluorescent proteins as GFP or mCherry, these tags are small and can be modified with more stable and brighter fluorophores.

The easiest way to introduce a tag is probably by using Cys residues as attachment points for maleimide or iodoacetamide derivatives, or other thiol reactive species (Gunnoo and Madder, 2016). The Cys residues can be native protein residues, or introduced at a unique position by site-specific mutagenesis, with replacement of the natural Cys residues by an unreactive amino acid. This has been conveniently used for α-synuclein labeling with an environment-sensitive fluorescent reporter to probe binding to lipid membranes and fibrillization by discriminating between unstructured, membrane‐bound and fibrillar states (Kucherak et al., 20182018). However, for the introduction of two different probes, a selective (de)protection of one of the two cysteine is required involving necessarily SPPS of a fragment at least of the protein of interest. Otherwise, a combination of different methods should be used. Cysteine conjugation was combined with amber codon suppression, transferase mediated N-terminal modification and NCL to produce α-synuclein variants bearing single or double site-specific fluorescent labels to probe conformational changes upon fibril formation and cellular uptake of fibrils (Haney et al., 2016; Haney et al., 2017).

Fluorescent probes differ whether investigating amyloidogenic processes in vitro or in living cells, and upon the fluorescence experiment employed being ensemble or single-molecule Förster resonance energy transfer (FRET), fluorescence polarization (FP), fluorescence correlated spectroscopy (FCS) or ESIPT. In any cases, careful considerations must be taken regarding the choice of labeling sites to avoid probe-induced perturbations of the protein conformation and fibrillization process even though small-molecule fluorophores are less invasive than their proteinaceous counterparts (e.g., GFP and its derivatives). Fluorophore are sensitive to change in their environments (e.g., exposure to the solvent, pH variation, ion concentration) as probed by FP and FCS. Introduction of two fluorophore labels gives access to FRET measurements based on distance-dependent fluorophore interactions. The distance between the fluorescent probes and dynamic fluctuation of these distances can be evaluated during fibril formation, upon interaction with small molecules (for example aggregation inhibitors) or due to a PTM. Single molecule FRET is of high interest when heterogeneity is considered, for example to characterize a population of oligomers. The use of fluorophore labels combined with microscopy imaging of live-cells is of importance to monitor cellular uptake, seeding and proteolysis.

The measurement of distance between FRET pairs in tau mutants carrying a tryptophan and a AEDANS-labeled cysteine, both residues being introduced by site-directed mutagenesis highlighted a paperclip-like conformation in native, soluble tau protein where both N- and C-termini fold over the MTBR (Figure 9B). Conformational changes of the paperclip preferential dynamic fold was observed upon phosphorylation and denaturation (Jeganathan et al., 2006; Jeganathan et al., 2012). The global fold was further confirmed by paramagnetic relaxation enhancement (PRE) NMR signals (see here below). The synthesis and FRET analysis of a site-specific, dual-labeled Htt exon1 indicated a progressive compaction of the protein upon increasing polyQ length in contrast to the pathological threshold length associated to HD (Warner et al., 2017).

Different conformational ensembles of α-synuclein were investigated inside cells using FRET pairs of fluorophores. FRET measurements allow to distinguish the unstructured α-synuclein and the α-helical conformation when α-synuclein is bound to the membrane, as both conformations are characterized by differences in the distance between both labels, the distance between the labeled amino-acid position being larger in the unstructured form (Fakhree et al., 2018). Introduction of labels can be coupled to introduction of modifications, using combination of the strategies here above briefly described using strategies addressed in paragraph 2. The semisynthesis of pY39 α-synuclein allowed to include a pair of FRET fluorescent labels using 1) coupling to cysteine and 2) UAA (propargyl-tyrosine, PpY) incorporation by amber codon suppression followed by click chemistry to reveal that phosphorylation of Y39 primarily acts on aggregation with only small interference with monomer conformation (Pan et al., 2020).

Processing of APP to generate the Aβ peptide was elegantly investigated in cells using a dual labeling scheme (van Husen et al., 2019). The APP C-terminal tail was labeled using a SNAP-tag while an UAA (trans-cyclooctene lysine) was introduced at a specific location in APP, corresponding to the Aβ peptide, by stop codon suppression. The side-chain of this UAA was next modified to attach a fluorescent dye (6-methyl-tetrazine-BODIPY-FL) by click chemistry reaction, while the unnatural residue was functionalized using TMR-Star. This double-labeling scheme allows to follow APP processing into a C-terminal fragment and Aβ peptide, and imaging the trafficking of the APP cleavage products in live cells.

Some in vivo experiments take advantage of the specific pH that characterizes some subcellular microenvironments to track the processing of amyloid proteins once they enter the cells. For example, CPX azide derivative was attached on α-synuclein using the UAA PpY functionalized by an azide-alkyne cycloaddition in a click chemistry approach. The labeled α-synuclein incorporated in fibrils was then tracked over time after it had entered neurons. Aggregates penetrating endosomes were highlighted by the green to red emission shift of CPX in that specific compartment (Jun et al., 2019).

Site-directed spin labeling, coupled to nuclear magnetic and/or electron paramagnetic resonance (EPR) measurements, are also of interest to study conformational fluctuations of amyloid proteins. A commonly employed paramagnetic label is the nitroxide moiety, which harbors an unpaired electron. This label is small and expected to cause no conformational perturbation of the functionalized protein. The paramagnetic effect of the electron is indirectly detected by NMR upon perturbation of the recorded signals due to enhanced transverse relaxation rate induced by proximity to the label (Clore et al., 2007). This effect has a r⁻⁶ dependence on the electron-proton distance (r) and thus allows the detection of long-range interactions in proteins (Figure 9A). To determine an ensemble of conformations consistent with PRE measurements, NMR signal intensity ratio (with/without nitroxide effect) can be converted into distance restraints. Direct measurements by EPR spectroscopy give information on the dynamics of the probe environment and is a convenient way to track oligomer formation (Zurlo et al., 2019). Double nitroxide labeling have interesting applications in both cases. Pulsed double electron–electron resonance (DEER) EPR on frozen samples can be used to derive a distribution of distances between the paramagnetic probes. For NMR applications, paramagnetic relaxation interference (PRI) in the presence of two nitroxide probes has been proposed to detect concerted conformational fluctuation in disordered proteins (Kurzbach et al., 2016). PRI being sensitive to correlated motions in the protein can help to detect a sparsely populated state in the ensemble, which nevertheless can be significant on the aggregation pathway. The Overhauser dynamic nuclear polarization (DNP) made use of site-specific nitroxide spin label attachment to monitor electron-spin interactions with water molecules and detect localized perturbations of hydration during tau protein aggregation related to the magnetic dipolar nature of interactions between the nitroxide unpaired electron and water protons that are mostly localized within 5Å. This allows discriminating hydration changes upon the formation of organized protein-protein interfaces associated to fibrils from non-specific protein-protein interactions (Pavlova et al., 2009).

Kinetics of α-synuclein aggregation was followed by continuous wave (CW) EPR using nitroxide spin-labeling to monitor oligomer formation/disappearance and how the aggregation pathway develops. The CW-EPR method uses the rotational diffusion time of the spin-labeled protein, visualized as EPR lineshapes, to cover the nanosecond to second time scales. The label dynamics is influenced by the local structure and macromolecular interactions at proximity of the probe. The MTSL label ((1-oxyl-2,2,5,5-tetramethylpyrroline-3-methyl)-methanethiosulfonate) was attached on α-synuclein C56, introduced by site-specific mutagenesis. The monomer, oligomer and fibril concentrations were evaluated at different time points during kinetics experiment, based on mobility differences of the species (Zurlo et al., 2019). Pulsed EPR measurements of various amyloid fibrils were also performed to gain structural information (Bedrood et al., 2012; Siddiqua et al., 2012; Pornsuwan et al., 2013).

Interestingly, EPR is also amenable to in vivo experiments (Cattani et al., 2017). Intracellular CW-EPR combined to a systematic spin-labeling scan of micro-injected recombinant α-synuclein shows that the majority of α-synuclein remains in monomeric, intrinsically disordered state in the Xenopus laevis oocyte cells, even in the case of disease variants A30P or A53T that are correlated with PD familial forms. These latter variants show an increased rate of oligomerization and decreased membrane binding (Flagmeier et al., 2016). The spin label for this in-cell study is 3-maleimido-PROXYL, which is more stable in the reducing environment of the cell than MTSL.

PRE of NMR signals were used to obtain structural information on α-synuclein. Three MTSL nitroxide labels were attached to single cysteine mutants. Measurements showed that α-synuclein adopts in its native state an ensemble of conformations that is stabilized by long-range interactions. These interactions are characterized by contacts between the central and C-terminal domains that shield the aggregation-prone non-amyloid-β component (NAC) (Bertoncini et al., 2005a; Dedmon et al., 2005). PRE of NMR signals additionally show that the familial PD mutations, A30P and A53T, perturb the ensemble of α-synuclein conformations pointing toward a reduced shielding of the hydrophobic NAC region in the A30P and A53T mutants of α-synuclein (Bertoncini et al., 2005b).

PRE of NMR signals were also used to build tau conformational ensemble. The paramagnetic probe MTSL was attached on the native cysteine residues, or on five single-cysteine mutants located in the various tau domains (Figure 9A). The intricate network of transient long-range interactions confirms the “hairpin model” of tau dynamic conformation. Interactions of the spin-label with distinct area of the tau sequence indeed show that the C-terminal domain transiently interacts with both the PRD and the N-terminus. The conformational dynamics of tau wild-type or harboring the P301L familial FTDP mutation were additionally compared using a combination of PRE and PRI experiments (Kawasaki and Tate, 2020). PRI data demonstrate alteration of the short- and long-range correlated motions in P301L tau mutant, promoting transient exposure of the PHF6 motif that plays a role as nucleus of tau aggregation (Kawasaki and Tate, 2020).

Brain imaging is a major area for which chemical biology has made a crucial contribution to support research on disease mechanisms in central nervous system proteinopathies and to deliver promising compounds for clinical use by providing tracers for in vivo imaging using positron emission tomography (PET) (Ni and Nitsch, 2022). This topic can be here only briefly covered given its scope. The challenge in this field is to provide specific and sensitive tracers reaching the brain, based on ¹⁸F, ¹¹C or ³H radio-compounds. Ideally, the tracers should allow early and differential detection of diagnostic biomarkers before the appearance of clinical symptoms to facilitate access to treatment ahead of the irreversible damages due to neuronal and synaptic connectivity loss.

Imaging of AD brain deposits started with the radio-compound [¹¹C]Pittsburgh compound B (Klunk et al., 2004) that derives from ThT to detect β-amyloid deposits by PET imaging (Figure 8B). Amyloid PET is based on β-sheet structure detection and compounds are mainly benzothiazole and benzoxazole derivatives. Among these tracers, three have already been approved for clinical use: florbetapir (Clark, 2011), flutemetamol (Curtis et al., 2015) and florbetaben (Figure 10A) (Sabri et al., 2015). Six binding sites on Aβ fibrils are proposed for the tracers, based on molecular modeling studies. Binding sites can be divided between core sites (buried in the fold with low solvent accessibility), and surface binding sites, the latter providing less specificity (Murugan et al., 2016). These in silico studies suggest that amyloid tracers of different structures could detect different Aβ fibrils.

Tau-specific ligands have also been developed for clinical use, and enable in vivo PET imaging of tau deposition (Leuzy et al., 2019; Saint-Aubert et al., 2017). Monitoring tau deposits is of interest for diagnosis as cortical retention of tau tracers better correlates with cognitive decline than Aβ deposits (Pontecorvo et al., 2017). These small molecules are however specific for the β-sheet structure adopted in tau fibers, and not to tau protein per se. This structure is found in other proteinopathies that might co-exist, which can complicate the image interpretation. Tau tracers should not only be able to cross the blood brain barrier, but additionally penetrate the intracellular compartment where tau fibrils also reside. Tau tracers belong to three major different chemical families, namely, pyridinyl-butadienyl-benzothiazole derivative (¹¹C-PBB3) derived from the same family as Pittsburgh compound B, benzimidazole-pyrimidine derivatives (¹⁸F-T807/AV 1451, and ¹⁸F-T808) as well as different arylquinoline derivatives (¹⁸F-THK5105, ¹⁸F-THK523, ¹⁸F-THK5117, and ¹⁸F-THK5351) (Figure 10B). These tracers bind to tau with affinities in the nanomolar-picomolar range. These first-generation tracers are however reported to bind to a number of off-targets such as monoamine oxidase-B in the basal ganglia (Lemoine et al., 2017). It leads to optimization of the tracer binding properties and delivered a second-generation of tracers with better specificity and a broader dynamic range. Based on in vitro binding assays, three different high affinity binding sites have been proposed in tau fibrils (Figure 10B). Computer-assisted docking of the tracers based on the cryo-EM structures of tau fibrils predict four different binding sites for the various tracers (Murugan et al., 2018; Kuang et al., 2020; Murugan et al., 2021).

The tau tracer AV-1451 (Johnson et al., 2016), also known as flortaucipir, specifically binds to tau fibril from AD and non-AD tauopathies (Lowe et al., 2016). The clinical validity of flortaucipir has been demonstrated as diagnostic biomarker and it is approved for imaging tauopathy in AD patients. A study based on PET imaging with [¹⁸F]AV-1451 confirms that tau pathology can propagate though a neuron-to-neuron transfer (Cope et al., 2018). The different types of tau deposits in the different diseases, highlighted at the atomic level by the cryo-EM structures (Lippens and Gigant, 2019), made it quite challenging to develop tau PET tracers. Given the heterogeneity in tau strains, tracers could be developed that recognize a sub-set of such strains. This strain heterogeneity, not only between pathologies but also between individual patients, is highly relevant because it is proposed to be linked to clinical heterogeneity in patients with typical AD (Dujardin et al., 2020). These in vivo diagnostics tool compounds will have huge impact on future assembly of better characterized cohorts and provide an adequate monitoring in clinical assays of the treatment effects.

Manipulating amino acid chirality into amyloid sequences has been implemented through SPPS to reveal amyloidogenic pathways and critical hot spots in amyloid assembly (Foley and Raskatov, 2021). In a strategy termed “chiral editing”, d-amino acid (D-AA) enantiomers that are site-specifically introduced either at selected positions of interest or randomly by D-AA scanning proved to be useful tools to highlight mechanistic details of Aβ aggregation and toxicity, as well as key residues in this process. Mirror peptides of Aβ42 and Aβ40 that incorporate D-AA along the entire sequence were shown to accelerate Aβ fibrillization into nontoxic fibrils in a racemic mixture with L-Aβ peptide (Dutta et al., 2017). The chiral inactivation of Aβ generates distinct fibrillar structures likely through the formation of rippled instead of pleated cross-β sheets (Dutta et al., 2019a). D-Aβ42 oligomers were also shown to have a reduced, if any, cytotoxicity and cellular internalization (Dutta et al., 2019b). Stereoselective interactions with chiral components of the phospholipid membrane could be responsible for this differential effect. Overall, the advantage of D-AA and mirror peptides is that chiral mutations do not change the physicochemical properties of the peptides (side chain chemical groups, size, polarity, charge…). It was shown that D-peptides targeting the mirror VQIINK tau hexapeptide (PHF6*) fibrils inhibit full-length tau aggregation while exhibiting protease stability and reduced immunogenicity (Figure 11A) (Dammers et al., 2016; Malhis et al., 2021). Based on these properties, D-peptides may be useful as therapeutic and diagnostic agents in amyloid-related diseases (Abdulbagi et al., 2021).

It is well documented that some amyloid-forming sequences spontaneously undergo into conformational changes and aggregate during SPPS, purification and other processing steps. Keeping amyloids under control with molecular switches is an efficient strategy to decipher their mechanism of aggregation and toxicity by controlling the onset of aggregation at early stages of initiation/nucleation, and identifying aggregation hot spots in amyloid sequences (Butterfield et al., 2012). A molecular switch is defined as a reversible modification at a selected location within the peptide sequence that tunes on and off the amyloid aggregation properties. O→N (or S→N) acyl shift of isopeptide bonds in so-called “click peptides” or “switch peptides” are usual molecular switches that are under control of a protecting group (Figure 11B). The native peptide bond is restored by O→N or S→N intramolecular rearrangement once the protection is released. The introduction of O-acyl or S-acyl isopeptide bonds efficiently disrupt misfolding and oligomerization of amyloid-forming sequences. In some instances, the release of an inhibitory element elicits amyloid assembly. A panel of triggering factors (pH change, enzymatic reactions, reducing agents, photoreactions) can be used for this purpose allowing a rapid switch from the inert into aggregation-prone sequence as conveniently illustrated for small, synthetic Aβ, IAPP or PrP peptides (Taniguchi et al., 2009; Taniguchi et al., 2006; Bosques and Imperiali, 2003). This provided information into the fine molecular mechanisms of Aβ42 (dis)aggregation by decoupling aggregation-prone sequence from a nucleating or structural element to prevent aggregation or, reversely, by favoring disassembly of amyloid fibrils into a soluble conformation (Figure 11B) (Dos Santos et al., 2005; Mimna et al., 2007). These methods also found applications in the screening of anti-aggregation compounds (Sohma et al., 2011).