CPA and DPCPX bind to Alpha-Synuclein in vitro.
Initially, both CPA and DPCPX were tested by nanopore analysis. Once a stable pore was created, a final concentration of 1 µM α-Syn was inserted on the cis side of the perfusion cup. At first, a few bumping events at 30 pA and a higher number of translocation events around 85 pA were observed, whereas intercalation events (I) were rarely encountered, as shown in Fig. 1b.2. Thus, these observations confirm similar recordings of stable blockade and bumping currents of α-Syn, as previously reported [33]. Then, 10 µM final drug concentration was inserted on the cis side and possible changes of blockade events could be observed.
First, we tested the potential binding of adenosine with α-Syn in our nanopore setup. As shown in Fig. 2b, adenosine appears to have weak binding affinity to α-Syn. The majority of the events of the − 86 pA blockade current are related to α-Syn translocation; however, there are fewer events observed in the translocation peak in the α-Syn and adenosine histogram (52%) in comparison to α-Syn alone (66%). Interestingly, CPA and DPCPX appear to bind to α-Syn as well, as shown in Fig. 2c and d. For the first time, we observed a decrease in the blockade current of the translocation peak from − 85 pA for α-Syn alone to -89 pA for the α-Syn and CPA complex (Fig. 2c), as the percentage of events decreased from 66–40% for CPA. Taken together the observed effects of CPA and, to a lesser extent, adenosine on the α-Syn translocation events are clear signs of binding. Conversely, DPCPX caused a small increase of the blockade current to -84 pA which was accompanied by a decrease of the number of events in the translocation peak (Fig. 2d). The blockade times of translocation and bumping peaks of α-Syn with and without adenosine, CPA and DPCPX were calculated. Representative exponential time graphs are shown in Fig. 3. The times of translocation and bumping events for α-Syn alone are well established [33]. The times of translocation events (α-Syn alone, 0.52 ms) decrease when α-Syn was combined with adenosine (0.46 ms), CPA (0.47 ms) or DPCPX (0.42 ms), which further indicates a potential binding of these drugs to the protein. On the other hand, we observed increased bumping times when α-Syn was incubated with adenosine, CPA or DPCPX (see Fig. 3e-h). A full summary of the blockade populations and times is shown in Table 1.
Caffeine, a nonselective inhibitor of all the adenosine receptors (A1, A2A, A2B and A3) [44], has been previously studied with nanopore analysis and it showed binding to the N- and C-terminus forming a neuroprotective loop conformation [40]. It is known that caffeine competitively antagonizes adenosine’s effect [45, 46]. Although the blockade populations of adenosine and CPA show some similarities to the histogram of caffeine binding to α-Syn [47], however, both adenosine and CPA decrease the blockade current of the translocation events of α-Syn to -86 and − 89 pA, respectively (see Fig. 2b and c). Previous results using 5 µM caffeine showed that the translocation population decreased to 44% from 81% of α-Syn alone, whereas the bumping population significantly increased to 38% compared to 9% of α-Syn alone [47]. This is similar to the decrease of the translocation population (37%) and increase of the bumping population (46%) observed with DPCPX + α-Syn (see Fig. 2d). Therefore, similar to caffeine, DPCPX potentially binds to the N- and C-termini of α-Syn forming a loop conformation. Further probing of the exact binding of both CPA and DPCPX was conducted using separate domains of α-Syn.
Alpha-Synuclein domain investigations of CPA and DPCPX
The two domains of α-Syn, namely the N- and C-termini, and the ΔNAC construct consisting of α-Syn with deleted non-Amyloidβ-component region (see Fig. 1d and e) were tested against CPA or DPCPX. The behaviour of each domain is different in a standard nanopore analysis at a direct current voltage of 100 mV (Fig. 4a, d, g). The blockade current histogram of the N-terminus has a single Gaussian peak at -30 pA due to bumping events. The N-terminus is positively charged (+ 4). Consequently, it will be difficult for this N-terminal fragment to translocate through the pore under the applied positive transmembrane voltage. Conversely, the C-terminus contains a total of 12 negative charges which permits translocation through the pore. The blockade current histogram has a large and wide translocation peak at -69 pA and a fairly small bumping peak at -30 pA. The ΔNAC has two peaks, a large peak at -86 pA due to translocation and a smaller one at -27 pA due to bumping.
In Fig. 4 are shown the blockade current histograms of each α-Syn domain in presence of CPA (Fig. 4b, e, h) and DPCPX (Fig. 4c, f, i). With the addition of CPA, the N-terminus proportion of bumping events decreased significantly from 70–46% (Fig. 4a vs. b). Conversely, the widespread block of events between − 50 and − 100 pA in the N-terminus control developed into a well-defined broad translocation peak at -26 pA with a population of 35%. The changes observed in the ΔNAC after the addition of CPA are remarkable and demonstrate clear signs of binding (Fig. 4d vs. e). The broad translocation peak at -86 pA has been reduced into a small cluster of events, whereas the small bumping peak significantly increased in population from 19–66% and has shifted to -36 pA from − 27pA. Interestingly, the C-terminal domain histogram profiles in absence and presence of CPA did not show significant differences (Fig. 4g vs. h), which indicates that CPA does not interact with the C-terminus.
In contrast, DPCPX produced different profile histograms of each of the α-Syn domains when compared to both the control and CPA. The N-terminus histogram profile shows that DPCPX caused a decrease in the proportion of bumping events from 70–41%, and DPCPX also revealed two additional peaks, namely the translocation peak at -72 pA and an intercalation peak at -51 pA (Fig. 4a vs. c). The intercalation peak at -51 pA has a low population of events (21%) whereas the translocation peak is broader and has a similar proportion to the translocation peak of CPA. The ΔNAC translocation peak disappears with the addition of DPCPX; instead, two peaks with similar proportion of events are observed at -24 and − 39 pA, representing the bumping and intercalation peaks, respectively (Fig. 4d vs. f). Lastly, the C-terminus histogram of DPCPX indicates a decrease of the bumping events from 20–7%, an emergence of an intercalation peak at -31 pA, and a significant decrease of the translocation peak from 77–38% (Fig. 4g vs. i). For convenience, all the blockade intensities and populations events are shown in Table 2.
Molecular dynamics simulations reveal CPA, adenosine and 2-aminoindan bind to the N-terminus of Alpha-Synuclein.
In order to confirm our biophysical results obtained from nanopore analysis, we further characterized the α-Syn-drug complexes by performing molecular docking attempts of the three A1R ligands as well as 1- and 2-aminoindan with α-Syn. Based on the nanopore analysis, a simulation of adenosine, A1R agonist CPA and methamphetamine analog 2-aminoindan was conducted (Fig. 5). As α-Syn crystal structure is still unknown, the 1XQ8 micelle-bound human α-Syn structure was used [42]. As shown in Fig. 5, adenosine, CPA and 2-aminoindan all bind to the same region of the N-terminus of α-Syn (blue alpha-helix region). As seen in Fig. 5a, adenosine forms hydrogen bonds with the lysine 32 (K32) and lysine 43 (K43) in the positively charged cleft of the α-Syn N-terminus. Adenosine forms additional hydrogen bonds with negatively charged glutamic acid 35 (E35) and interacts with the aromatic hydrophobic tyrosine 39 (Y39) and valine 40 (V40). CPA is a chemical derivative of adenosine and shows greater selectivity as A1R agonist, and thus it is expected to interact with α-Syn similarly to adenosine. As shown in Fig. 5b, CPA interacts with similar positively charged cleft containing K32 and K43 and forms additional bonds with E35. However, CPA further interacts with the polar uncharged threonine 33 (T33) and with the hydrophobic alanine 29 (A29) and V40. Lastly, 2-aminoindan is also shown to form hydrogen bonds with the negatively charged E35 (Fig. 5c), which is in common with adenosine and CPA. Moreover, as with adenosine and CPA docking simulations, 2-aminoindan also interacts with V40 and K43.
DPCPX and 1-aminoindan show interactions with Alpha-Synuclein N- and C-termini from molecular dynamics simulations.
In contrast with adenosine, CPA and 2-aminoindan, the A1R antagonist DPCPX and Rasagiline metabolite 1-aminoindan appear to possess a more intricate binding interactions with α-Syn. For these simulations both the 1XQ8 micelle-bound human α-Syn and the homology modeled structures were used. As seen in the first panel of Fig. 6, DPCPX binds to the α-Syn N-terminus (amino acids K10, A11 and V13) and the end of the NAC region (amino acids K80, T81, A85, S87 and isoleucine 88 (I88)) using the homology modeled structure (Fig. 6a), but it binds only to the distal region of the α-Syn C-terminus when the 1XQ8 structure is used (Fig. 6b). Similarly, 1-aminoindan binds to the N-terminus (amino acids phenylalanine 4 (F4) and leucine 8 (L8)) and NAC region (amino acids T81, A85 and I88) with the homology modeled structure (Fig. 6c), but it appears to bind only to the α-Syn C-terminus using the 1XQ8 model (Fig. 6d). As the full crystal structure of α-Syn is not yet available, we suggest that using the two conformations of the 1XQ8 structure and the homology modeled structure of α-Syn in our molecular docking attempts should give a more complete information of the binding interactions of DPCPX and 1-aminoindan with α-Syn. Taken together, the molecular docking studies appear to confirm the results obtained from nanopore analyses, namely that adenosine, CPA and 2-aminoindan only interacted with the α-Syn N-terminus; this is expected to promote α-Syn conformation that promotes α-Syn aggregation. In contrast, DPCPX and 1-aminoindan showed binding to both N- and C-terminal regions of α-Syn; this binding pattern is expected to promote an α-Syn conformation that prevents α-Syn aggregation.
Adenosine A1 receptor agonist and drugs that bind to α-Syn N-terminus increased α-Syn expression and aggregation in substantia nigral neurons.
To investigate whether drug binding to the N- and/or C-terminus of α-Syn can affect the levels of α-Syn expression and aggregation in vivo, we administered the drugs individually or in combination with the A1R agonist CPA. Our results show that 7-day systemic administration of CPA, alone or in combination with 2-aminoindan, leads to increased expression and aggregation of α-Syn in the SN pars compacta (Figs. 7,8). Figure 7a shows a low magnification of the whole SN stained with 3,3’Diaminobenzidine (DAB)-tyrosine hydroxylase (TH), indicating the locations of the dopaminergic neurons in the SN [48] especially in the pars compacta region (square box) used for subsequent quantification of α-Syn and aggregation levels (Fig. 7b, Fig. 8) and neurodegeneration levels (Fig. 9). We focused on the SN pars compacta region of the midbrain which is linked to PD, as the pathology of this disease is characterized by significant loss of dopaminergic neurons in this region [49, 50]. Subsequently, fixed coronal slices of 40 µm of SN were probed for TH (a marker for dopaminergic neurons), α-Syn and Thioflavin-S (Thio-S), a fluorescent stain for α-Syn aggregates and amyloids. The slices were also labelled with DAPI (fluorescent stain that binds to adenine-thymine regions in DNA) to detect cell nuclei. Confocal images were taken with 63X oil immersion objective lens at high magnification (Zeiss microscopy). Representative images of the pars compacta region labelled with DAPI (first column), TH (second column), and α-Syn (third column) show that α-Syn is localized in the somas (cytosol, nuclei) and presumably the dendrites of dopaminergic neurons (Fig. 7b). Interestingly, α-Syn expression was increased by CPA in absence or presence of DPCPX, 1-aminoindan, or 2-aminoindan (Fig. 7b, Fig. 8a-b) compared to the control (1% DMSO in 9% saline). The 2-aminoindan + CPA treatment induced the highest levels of the α-Syn protein expression (Fig. 8a,b). In contrast, treatments of animals with DPCPX, 1-aminoindan or 2-aminoindan alone did not significantly increase α-Syn protein expression.
To determine whether these changes in α-Syn protein expression correlated with the levels of α-Syn aggregation, the SN pars compacta was co-labelled with α-Syn marker and Thio-S. As shown in Fig. 8a-c, treatments with CPA, DPCPX + CPA, 1-aminoindan + CPA, and 2-aminoindan + CPA all increased Thio-S levels. Treatment with 2-aminoindan alone also enhanced Thio-S labelling, and co-administration of 2-aminoindan with CPA caused a significant further elevation of Thio-S compared to 2-aminoindan treatment. In contrast, treatments with DPCPX and 1-aminoindan alone did not significantly increase Thio-S labelling and also did not significantly attenuate the CPA-induced increase in Thio-S levels. The colocalization of Thio-S signal with α-Syn was increased in all the treatments compared to control (about 2-fold increase in Pearson correlation coefficients, Fig. 8c). Together, these results suggest that 7-day chronic treatments with compounds that bind to the N-terminus of α-Syn (e.g., the A1R agonist CPA, 2-aminoindan) significantly increased α-Syn accumulation and aggregation.
CPA and 2-aminoindan increase neurodegeneration of SN pars compacta dopaminergic neurons.
Having shown that CPA and 2-aminoindan alone or in combination can increase α-Syn aggregation, we then determined whether these treatments could lead to neuronal damage. Hence, we used the FluoroJade C (FJC) as a common fluorescent marker for neurodegeneration in the CNS [31]. FJC staining was performed in nigral slices from − 5.30 to -5.60 bregma intervals. Figure 9a panels showing high magnification images of FJC staining in the pars compacta region of SN indicate that CPA alone, 2-aminoindan alone, and 2-aminoindan + CPA co-administration all increased the levels of FJC fluorescence (as summarized in Fig. 9b). In contrast, DPCPX or 1-aminoindan alone did not significantly increase FJC staining, but both drugs were effective in attenuating CPA-induced increase in neurodegeneration (summarized in Fig. 9b). Taken together with the above results from nanopore analysis, molecular docking and Thio-S labelling, these results suggest that compounds that bind to both N- and C-termini of α-Syn (e.g., DPCPX and 1-aminoindan) may be effective in attenuating the neurotoxic effects of compounds that bind to and promote α-Syn accumulation and misfolding (e.g., CPA, adenosine, and 2-aminoindan).