1. Introduction
Amyloid formation is a widespread phenomenon based on the generic property of the polypeptide chain to self-assemble into a cross-β-sheet containing oligomers and fibrils [
1,
2]. Their growth and accumulation are manifested in numerous amyloid-related diseases [
3,
4,
5], including neurodegenerative diseases, such as Alzheimer’s and Parkinson’s. Despite the key clinical significance of amyloid formation, the mechanisms of its inhibition, reversal, and modification remain elusive. The cross-β-sheet structure at the core of amyloid fibrils is stabilized by the numerous hydrogen bonds of the polypeptide backbone. In addition, π-π stacking of aromatic residues can also contribute to amyloid self-assembly and stability [
6]. Small phenolic compounds alone or conjugated with various groups were found to be effective in targeting the monomeric polypeptides as well as their aggregates. They have shown potential activity in animal models of Parkinson’s disease, and some have already entered clinical trials [
7,
8,
9,
10].
Here, we consider the effect of cyclic compounds and their conjugates on the amyloid formation of pro-inflammatory S100A9 protein, which was found to be a common denominator in Alzheimer’s and Parkinson’s diseases as well as in traumatic brain injury, which is considered to be a pre-cursor state for neurodegenerative ailments [
11,
12,
13]. Indeed, amyloid formation is commonly associated with neuroinflammation, and pro-inflammatory S100A9 protein acts both as an alarmin, inducing the production of pro-inflammatory cytokines, and as a highly amyloidogenic protein, which self-assembles into amyloids under physiological conditions [
14,
15]. By combining in vitro, ex vivo, and in vivo studies, we have demonstrated that S100A9 may drive the amyloid-neuroinflammatory cascade in neurodegeneration both in humans and in a mice model [
11,
12,
13,
16,
17]. Moreover, S100A9 co-aggregates and forms joint complexes with major amyloid polypeptides, such as amyloid β (Aβ) peptide in Alzheimer’s disease [
11,
18] and with α-synuclein in Parkinson’s [
12]. This would imply that upon altering the aggregation pathways and final structures of one component of the amyloid cascade, other components and the whole cascade could also be modified.
Furthermore, growing evidence demonstrates that apart from disease-associated amyloids, there are also functional amyloids, which play useful roles in numerous biological processes in many organisms, ranging from mammals and insects to fungi and bacteria [
19,
20]. For example, peptide and protein hormones can be stored in an inert amyloid state in secretory granules of the endocrine system [
21], transcription and translation can be regulated by prion proteins in yeast [
22], spidroins enhance spider web tensile strength, and bacterial functional amyloids, such as curli and FapC, participate in the formation of bacterial biofilms [
19]. In recent years, the possibility of cross-seeding of disease-associated amyloids by functional bacterial amyloids has received increasing attention as this mechanism can be linked to the gut–brain axis of neurodegeneration disease development upon changes of the gut microbiome [
23,
24]. It has been shown that the aggregation and toxicity of Aβ, the pathogenic peptide associated with Alzheimer’s disease, can be seeded by FapC amyloid fragments of
Pseudomonas aeruginosa, which colonizes the gut microbiome through infections [
24]. Thus, the link between the gut microbiome, their metabolites, and neurological disorders, such as Alzheimer’s and Parkinson’s [
23,
24], suggests an additional pathway that attunes amyloid formation of disease-related proteins via modulation of amyloid self-assembly of other amyloid counterparts and their amyloid seeds.
To target S100A9 and its amyloids, we developed conjugated molecules by combining cyclic compounds, which may potentially target aromatic residues and hydrophobic regions within the amyloid structures, with charged di-peptides, in order to make the hydrophobic moiety more soluble. One group is based on L-DOPA—L-3,4-dihydroxyphenylalanine (DOPA), which is an amino acid that is made and used as part of the normal biology in humans as well as in animals and plants. DOPA is produced from the amino acid L-tyrosine by the enzyme tyrosine hydroxylase and can act as an L-tyrosine mimetic to be incorporated into proteins in place of L-tyrosine, generating protease-resistant and aggregation-prone proteins [
25]. Sufferers from Parkinson’s disease are exposed to DOPA over many years as a therapeutic agent for their medical condition as it serves as a pre-cursor for the catecholamine neurotransmitter dopamine. Dopamine is produced in the brain and plays several important biological roles, acting as both a neurotransmitter and hormone [
25,
26,
27]. Dopamine production and homeostasis in neurons is regulated by α-synuclein through the interaction with tyrosine hydroxylase. It has been reported that dopamine can form adducts with α-synuclein in vitro, which stabilize the α-synuclein protofibrils and inhibit its fibril formation, while certain dopamine derivatives alleviated α-synuclein-engendered defects in a Parkinson’s animal model [
28,
29,
30,
31,
32]. Naphthoquinone–DOPA hybrids inhibit α-synuclein and tau aggregation, disrupt preformed fibrils [
33,
34], and attenuate aggregate-induced toxicity [
33]. DOPA and DOPA-conjugated naphtalenediimides also modulate Aβ toxicity [
35]. Moreover, recent studies have reported the self-assembly of DOPA-containing building blocks into nanometric fibers [
36,
37], characterized by a unique functionality [
38]. In addition, a short synthetic pentapeptide, containing two DOPA moieties, D-DOPA-N-K-DOPA, retained the ability to spontaneously self-assemble into amyloid-like fibrillar assemblies in water, and those assemblies displayed structural properties characteristic of amyloids [
39]. Here, we linked DOPA to negatively charged aspartic acid residue (D) and positively charged histidine residue (H) to produce the following soluble compounds: DOPA-D-H, DOPA-D-H-DOPA, and DOPA-H-H-DOPA (
Figure 1).
In addition, DOPA was linked to saturated macrocyclic polyamine cyclen—1,4,7,10-tetraazacyclododecane (cyclen), which belongs to a class of nitrogen-containing heterocyclic compounds, and DOPA-cyclen and H-E-cyclen were used here to compare their effect on S100A9 fibrillation vs. DOPA-based conjugates. Macrocyclic polyamines, including cyclen, have been very broadly used in medicinal chemistry as therapeutic or diagnostic agents [
40,
41] as well as to target amyloid formation via chelating metal ions [
42,
43,
44,
45]. Cyclen was also linked to recognition of the amino acid sequence KLVFF to target and inhibit the aggregation of Aβ peptide [
46]. Cyclen increases the solubility of DOPA and provides additional modes of interactions with S100A9. Here, we applied a range of experimental techniques, such as thioflavin-T (ThT)-based amyloid kinetic monitoring, atomic force microscopy (AFM) analysis, and S100A9 titration with the compounds of interest, combining them with computational analysis performed by ligand docking and molecular dynamic (MD) simulation, to draw a broad picture of the inter-molecular interactions between the selected compounds and S100A9, and their effect on S100A9 amyloid self-assembly.
3. Discussion
DOPA and cyclen-based compounds have been studied intensively in research targeting amyloid formation of α-synuclein, Aβ, and tau, polypeptides involved in neurodegenerative diseases, and these compounds have shown some potency in amyloid inhibition [
28,
29,
30,
31,
32,
33,
34,
35]. DOPA is a precursor of the neurotransmitter dopamine and Parkinson’s sufferers have received DOPA as disease-modifying treatment for years [
25,
26]. Therefore, it was important to examine the effect of DOPA and cyclen-based compounds modified by conjugated amino acid residues on the amyloid self-assembly of the pro-inflammatory protein S100A9, which has been shown to be a central component of the amyloid neuro-inflammatory cascade in Alzheimer’s, Parkinson’s, and traumatic brain injury [
11,
12,
13]. The kinetic analysis of S100A9 amyloid’s self-assembly in the presence of each DOPA and cyclen-based compound taken at both molar ratios of protein to ligand of 1:1 and 1:10 did not show any significant effects on the rate of amyloid formation (
Figure 2). This contrasts with previous findings of the inhibiting effects of DOPA and cyclen-based ligands on the amyloid fibrillation of α-synuclein, Aβ, and tau [
28,
29,
30,
31,
32,
33,
34,
35], indicating that there is no single silver bullet able to inhibit and reverse all the diversity of amyloids. Interestingly, by comparison to Aβ and α-synuclein, S100A9 undergoes amyloid aggregation without a pronounced lag phase, corresponding to a nucleation or oligomerization process, and cannot be seeded by preformed homo or hetero-amyloids [
12,
15,
18,
54]. The kinetics of S100A9 amyloid formation are described well by the generic Finke–Watzky autocatalytic model, in which initial protein misfolding and β-sheet formation are defined as the ‘nucleation’ step, spontaneously taking place within individual S100A9 molecules at a higher rate than the subsequent amyloid assembly. Therefore, amyloid self-assembly, described as an autocatalytic process, will proceed if misfolded amyloid-prone S100A9 is populated on a macroscopic time scale [
15]. This suggests that dynamic binding of DOPA and cyclen-based derivatives of S100A9 homodimer did not promote S100A9 misfolding, which would consequently affect its amyloid kinetics; however, this effect was not observed here.
Importantly, the AFM-cross-section analysis of co-aggregates of S100A9 with the above compounds, apart from H-E-cyclen, clearly demonstrated that the amyloid fibrils became two to three times thicker compared to S100A9 fibrillated alone (
Figure 3). This indicates that the ligand bindings and encapsulation into amyloid fibrils altered the protein conformation and therefore the packing of proteinaceous material at the fibrillar interface, leading to their thickening. Amyloid formation is generally characterized by significant polymorphism, which may be relevant to a specific type of disease and disease development [
60,
61]. Thicker fibrils may also slow down the tissue propagation of amyloids as it was suggested previously, as the clumping and thickening of amyloid co-aggregates of S100A9 with NCAM constructs has been observed [
54]. Thus, analysis of the amyloid fibril morphology and structural studies of amyloid aggregates may provide additional insight into the potential development of the amyloid disease pathology and amyloid propagation [
62], which were shown for Aβ peptide [
61] and prions [
63,
64].
All DOPA and cyclen-based ligands were shown to bind to S100A9 homodimer upon titration as monitored by intrinsic fluorescence (
Figure 4). Their complex formations were characterized by similar apparent dissociation constants in a sub-micromolar range as determined by fitting the titration curves to a single site binding model. Molecular docking and MD simulation provided additional and important insight into the nature of the binding sites and mechanisms of ligand binding, enabling us to differentiate between the ligand binding modes for each ligand. Indeed, S100A9 is a small protein with no deep cavities on its surface, which makes it a challenging target for drug design [
48]. All five studied ligands were docked on the protein surface in the relatively shallow diagonal groove and some additional low-affinity binding sites at the convex interface between the 2
d and 3
d helices on each monomer (
Figure 5). All ligands are characterized by high H-bond donor and acceptor capacities (
Figure 1) and are able to make three to six H-bonds with the protein surface at the initial docking (
Figure 6). RMSD analysis showed that some complexes with DOPA-H-H-DOPA and DOPA-D-H-DOPA were characterized by less than 5 Å values, indicating their limited mobility and good fit to the binding site. However, other bound ligands were more mobile or even dissociated from S100A9 homodimer if RMSD increased beyond 25 Å. Overall, the binding of various ligands to the major diagonal groove resulted in rather similar apparent K
d values as determined in the fluorescence titration experiments.
It is important to note that S100A9 may exist as a homo- and heterodimer, forming complex with S100A8 [
14]. The thermal stability of S100A9 homodimer is somewhat lower than S100A8/S100A9 heterodimer as reported previously [
65], which may make it more amyloid prone by facilitating its misfolding [
15]. Previously, in studies of ex vivo brain tissues in Alzheimer’s and Parkinson’s disease, mild cognitive impairment, and traumatic brain injury, we observed the extra and intracellular amyloid deposits of S100A9, but not S100A8 [
11,
12,
13,
66]. Moreover, in mice models of Alzheimer’s disease and traumatic brain injury, we also detected depositions of S100A9, but not S100A8 [
16,
17]. Furthermore, S100A9 forms amyloid complexes with Aβ, by templating on the Aβ fibrillar surface, which may also contribute to the development of joint amyloid deposits in Alzheimer’s disease [
11,
18], though these polypeptides do not form a mixed cross-β-sheet [
67]. By contrast, in the aging prostate, we observed the co-localization of amyloid deposits of both S100A9 and S100A8 [
68]. Even though amyloid development in the brain tissues could be the primary target for DOPA-based compounds, and therefore, here, we studied their interactions with S100A9, it remains an open possibility that in some other functional or pathological conditions in other organ and tissues than the brain, binding of DOPA and cyclen-based compounds to S100A9 may affect its interaction and co-aggregation with S100A8.
It is important to note that DOPA and cyclen-based ligands did not interact with the region on the S100A9 surface, including Lys 50 to Lys 54, which has been shown previously to be critical for S100A9 amyloid self-assembly and by blocking this specific amino acid sequence by polyoxoniobates, we were able effectively hinder S100A9 amyloid growth [
53]. However, as the S100A9 amyloid surface is important for its quaternary complex formation and amyloid co-aggregation with Aβ peptide [
18] and potentially with other proteins and disease-related and functional amyloids, the amyloid morphology-modifying effect of DOPA and cyclen-based compounds co-aggregated together with S100A9 into fibrils should be taken into account in future studies of protein hetero-aggregation.