Cysteine-based Mimic of Arginylation Reproduces Neuroprotective Effects of the Authentic Post-translational Modification on α-Synuclein

Abstract

Arginylation is an under-studied post-translational modification (PTM) involving the transfer of arginine to aspartate or glutamate sidechains in a protein. Among the targets of this PTM is α-synuclein (αS), a neuronal protein involved in regulating synaptic vesicles. The aggregation of αS is implicated in neurodegenerative diseases, particularly in Parkinson’s disease, and arginylation has been found to protect against this pathological process. Arginylated αS has been studied through semi-synthesis involving multi-part native chemical ligation (NCL), but this can be very labor-intensive with low yields. Here, we present a facile way to introduce a mimic of the arginylation modification into a protein of interest, compatible with orthogonal installation of labels such as fluorophores. We synthesize bromoacetyl arginine and react it with recombinant, site-specific cysteine mutants of αS. We validate the mimic by testing the vesicle binding affinity of mimic-arginylated αS as well as its aggregation kinetics and monomer incorporation into fibrils, and comparing these results to those of authentically arginylated αS produced through NCL. In cultured neurons, we compare the fibril seeding capabilities of pre-formed fibrils carrying a small percentage of arginylated αS. We find that, consistent with authentically arginylated αS, mimic arginylated αS does not perturb the protein’s native function but alters aggregation kinetics and monomer incorporation. Both mimic and authentically modified αS suppress aggregation in neuronal cells. Our results provide further insight into the neuroprotective effects of αS arginylation, and our alternative strategy to generate arginylated αS enables the study of this PTM in proteins not accessible through NCL.

Graphical Abstract

INTRODUCTION

α-Synuclein (αS) is a neuronal protein involved in regulating synaptic vesicles, and its pathological aggregation underlies a class of neurodegenerative diseases including Parkinson’s disease (PD). Among the many post-translational modifications (PTMs) that are thought to affect the function and disease-relevant properties of αS, arginylation is a rather unusual one. Arginylation is the post-translational attachment of the amino acid arginine to the N-terminus or to a glutamate or aspartate sidechain via an isopeptide bond, a modification carried out by the enzyme Arg-tRNA-protein-transferase (ATE1).¹ The PTM was initially believed to occur only on the protein N-terminus,^2-3 but was later also identified on side chain carboxylates.⁴ It is this latter type of site that ATE1 recognizes on αS, catalyzing the transfer of arginine onto glutamates 46 and 83.⁵

Investigations into the effects of arginylation in αS have uncovered the ability of the PTM to reduce fibril seeding in cells: αS aggregation in neurons seeded by pre-formed fibrils of αS is reduced in the presence of arginylation.⁵ Abolishing the arginylation modification by mutation of glutamates 46 and/or 83 to alanine leads to increased αS accumulation in cells, and lack of arginylation leads to αS aggregation in the mouse brain, further supporting the ability of arginylation to reduce αS-induced neurotoxicity. We previously studied the arginylation modification in chemical detail using semi-synthesis to introduce arginylated glutamate (E^Arg) site-specifically at one or both sites in αS.⁶ When assayed for their ability to bind lipid vesicles, αS constructs with all combinations of arginylation were found to have unaltered vesicle affinity compared to unmodified control αS, suggesting that arginylation does not perturb its native function. In contrast, for pathological aggregation, a dose-dependent effect was identified, in which arginylated αS displayed reduced aggregation rates and monomer incorporation into fibrils as a function of increasing percentage of αS-E^Arg in the monomer starting material. This inhibition of aggregation is potentially neuroprotective. Indeed, recent studies using the semi-synthetic E^Arg proteins to quantify the levels of αS arginylation in PD patient brains show it to be significantly reduced compared to healthy controls,⁷ motivating us to seek methods that allow its study in vivo.

Our previous semi-syntheses involved optimizing production of the protected E^Arg amino acid, its incorporation into αS fragments by solid phase peptide synthesis (SPPS), expression and fluorescent labeling of recombinant αS fragments with unnatural amino acid click chemistry handles, and multi-part native chemical ligation (NCL) to afford αS with multiple modifications (Supporting Information, SI, Scheme S2). Although such semi-syntheses of arginylated-αS enabled the first biochemical studies of this PTM with site-specific installation of the exact chemical group of interest, they are not amenable to production of sufficient quantities of protein systematic in vitro and in vivo studies. We therefore asked whether we could install a mimic of Glu arginylation that is synthetically facile and deviates minimally in structure from authentic E^Arg.

The simplest examples of using PTM mimics include mutating a serine, threonine, or tyrosine to glutamate as a substitute for phosphorylation.⁸ The mutation of lysine to glutamine⁹ or glutamate¹⁰ has also been used as a surrogate for acetyl lysine. Although these substitutions are easy to perform, it is questionable whether the mimic has the same properties as the PTM of interest. Indeed, while glutamine has been found to closely mimic acetyl lysine with regards to nucleosome array self-association,⁹ glutamate does not reproduce the effects of phosphotyrosine in αS aggregation because of differences in both sterics and electronics between the two amino acids.¹¹ Mimics of PTMs obtained with the aid of small molecules include ubiquitination¹² or SUMOylation¹³ linked via a disulfide to a cysteine mutation introduced at the site of interest, a structure that resembles the native lysine linkage. Other surrogates for PTMs that involve some synthesis also have the advantage of affording a product with minimal differences in chemical structure to the native PTM. For example, not only ubiquitination and SUMOylation, but also lysine acylation analogs can be introduced using a small molecule auxiliary to afford an acyl thialysine linkage.¹⁴

Given the versatility of the amino acid cysteine, we took advantage of its ability to undergo reactions as a thiol nucleophile to generate an arginylation mimic nearly identical to the authentic modification. We developed the small molecule bromoacetyl arginine, which, after conjugation with a cysteine introduced by site-directed mutagenesis in the protein of interest, affords an arginylated glutamate mimic with the difference of a single S addition in the sidechain. While one might assume that such a modification is non-perturbing, recent analyses of a CH₂-to-S substitution for introducing Lys PTM mimics at Cys has shown that this is not always the case.^15-16 Therefore, to validate use of our arginylation mimic, we explored the effects of the arginylation mimic on protein function and aggregation. We assayed the effects of mimic arginylation on the lipid binding affinity of αS, an in vitro model for the function of the protein in modulating vesicle trafficking. We also tested the mimic’s effects on pathological processes through in vitro measurements of aggregation kinetics and monomer incorporation into fibrils. We then compared the aggregation seeding capabilities of mimic versus authentically arginylated αS pre-formed fibrils (PFFs) in vitro as well as in cultured neurons. Finally, to interpret the seeding effects, we have carried out preliminary structural characterization of the fibril seeds.

RESULTS AND DISCUSSION

Synthesis of Arginylation Mimics.

To access the arginylation mimics, we first synthesized the small molecule bromoacetyl Arg (Scheme 1). The synthesis was performed on protected arginine tert-butyl ester [Arg(Pbf)-OtBu, 1] on a 0.1 mmol scale. 2-Bromoacetic acid was activated by reaction with isobutylchloroformate (IBCF) and N-methyl morpholiner (NMM) to form a mixed anhydride to which Arg(Pbf)-OtBu was added. After purification, the intermediate product 2 was obtained in 61% yield. The ester 2 was deprotected using trifluoroacetic acid (TFA) and the crude product was purified by reverse phase high performance liquid chromatography (RP-HPLC), affording bromoacetyl Arg (3) as a white powder in 42% yield after lyophilization.

Scheme 1. Synthesis of bromoacetyl Arg (3) and C^EArg αS.

Top: Synthesis of bromoacetyl Arg (3). Bottom: Reaction of αS cysteine mutant with bromoacetyl arginine yields arginylation mimic at desired site(s). The strategy allows for orthogonal installation of fluorophores such as Atto488-azide by copper-catalyzed azide-alkyne cycloaddition.

For reaction with 3, we expressed αS single cysteine mutants E₄₆C and E₈₃C, as well as the double cysteine mutant, as intein fusions to aid in handling (SI, Scheme S1).¹⁷ We also expressed each of these mutants with a propargyltyrosine (π) at position 114 using amber codon suppression, allowing orthogonal fluorescent labeling of each construct using methods previously established in our laboratory.^{6, 18-21} The fluorophore Atto488 azide (Atto488-N₃) was appended via copper-catalyzed click chemistry (Scheme 1). Expression yields for protein constructs were 3-7 mg per 500 mL of culture. Bromoacetylation reactions were done on a 175-200 nmol scale, and conversion was quantitative for all constructs. The mimic arginylated protein products were purified by RP-HPLC and exchanged into buffer using spin concentrators, resulting in an isolated yield of 55% for αS mimic arginylated at site 46 (αS-C^EArg₄₆), 34% for that at 83 (αS-C^EArg₈₃), and 42% for the doubly modified construct (αS-C^EArg₄₆C^EArg₈₃). For proteins with π, click labeling reactions were quantitative, giving αS-C^EArg₄₆π⁴⁸⁸₁₁₄, αS-C^EArg₈₃π⁴⁸⁸₁₁₄, and αS-C^EArg₄₆C^EArg₈₃π⁴⁸⁸₁₁₄ in isolated yields of 29%, 65%, and 25%, respectively. In general, yields of the C^EArg mimic proteins were twice those of authentic E^Arg proteins made through NCL, with 4-8 fewer steps required (SI, Table S1).

Comparison of Effects Lipid Vesicle Binding.

To test the effect of the mimic arginylation on the function of αS, we measured the lipid vesicle binding affinity by fluorescence correlation spectroscopy (FCS). In FCS, a fluorescently labeled species is observed diffusing through the focal volume, giving rise to fluctuations in fluorescence intensity over time. This signal is compared to itself over time by autocorrelation, and the data are fit to obtain information on the properties of the sample. Negatively charged lipid vesicles, for which αS has high affinity,²² were used in these assays as in our previous studies of the authentically arginylated αS.⁶ Vesicles containing 50:50 1-palmitoyl-2-oleoyl-sn-glycero-3-phospho-L-serine/1-palmitoyl-2-oleoyl-glycero-3-phosphocholine (POPS/PC) were prepared synthetically. Protein fluorescently labeled at position 114, but without any PTM (αS-π⁴⁸⁸₁₁₄), was used as the wild type (WT) control. The diffusion times of the protein and of the lipid vesicles were first measured independently. In the αS lipid-binding assay, fluorescent protein mimic arginylated at one site, (αS-C^EArg₄₆π⁴⁸⁸₁₁₄ or αS-C^EArg₈₃π⁴⁸⁸₁₁₄) or at both sites (αS-C^EArg₄₆C^EArg₈₃π⁴⁸⁸₁₁₄) was added to a range of concentrations of lipid vesicles (0.005-0.5 mM total lipid). The fraction of protein bound at each lipid concentration was obtained from the fits and used to generate a binding curve for each protein construct (Figure 1).

Figure 1.

Lipid binding affinity of single arginylation mimic αS-C^EArg₄₆ and αS-C^EArg₈₃, and double arginylation mimic αS-C^EArg₄₆C^EArg₈₃. *The affinities of authentically arginylated αS-E^Arg₄₆ and αS-E^Arg₈₃, and αS-E^Arg₄₆E^Arg₈₃ as well as familial PD mutant E₄₆K, measured previously,⁶ are shown for comparison.

We found that the mimic arginylated proteins bound to POPS/POPC lipid vesicles with affinities similar to that of unmodified (WT) αS. In all cases, the POPS/POPC lipid vesicle affinities obtained for proteins with C^EArg mimic arginylation are similar to those obtained for proteins with authentic E^Arg made through NCL (Figure 1). Both authentic and mimic arginylation at E₈₃ resulted in a small decrease in affinity compared to WT, unlike the PD mutant E₄₆K, which increases affinity. Thus, mimic arginylation, like the authentic arginylation, should not significantly perturb the native function of αS. Importantly, these data demonstrate that the new method of arginylation provides αS constructs that reproduce the effects of those observed with the previous more labor-intensive synthesis.

Comparison of Effects on In Vitro Aggregation.

As mimic arginylation had the same effects on αS function as the authentic modification, we wished to know whether the mimic could recapitulate the neuroprotective effects of the authentic arginylation in reducing aggregation. We prepared samples of αS monomer, mixing WT αS with 5% or 10% arginylated αS. These percentages were chosen because our analysis of αS arginylation levels in PD patient brains and healthy controls indicated that were in the 5-25% range.⁷ Thus, aggregations of 5% and 10% mixtures of either authentic E^Arg αS or mimic C^EArg αS were determined to be pathophysiologically relevant. Aggregation was carried out in vitro by shaking at 37 °C. The presence of fibrils was detected using Congo Red, the absorbance of which was used as an indicator of the extent of fibril formation at each time point (Figure 2, individual curves and 5% mixture data are shown in SI, Figure S11). The resulting kinetic curves showed that mimic arginylation at either glutamate 46 or glutamate 83 slowed aggregation, consistent with authentic arginylation at these sites. Intriguingly, the dose-dependence of the effect seems to be stronger for position 46; for position 83, both 5% and 10% arginylation led to a similar ~1.5-fold decrease in aggregation rate. However, 1% arginylation at glutamate 83 was not sufficient to affect aggregation rate (SI, Figure S11). Mimic arginylation at both sites reduced the aggregation kinetics even further than at glutamate 46 or 83 alone at a comparable percentage, again consistent with the previously reported authentic arginylation data (Table 1).

Figure 2.

Aggregation kinetics and total monomer incorporation for mimic arginylated αS. WT monomer starting material was mixed with different percentages of mimic arginylated αS, aggregated by shaking at 37 °C, and monitored by Congo Red. Left: Aggregation curves showing that mimic arginylation at either glutamate 46 or 83 slows aggregation of αS and that the doubly mimic arginylated αS slows aggregation further. Aggregation time shown relative to matched WT control for each trial. Center: Relative time to half completion of aggregation (T_1/2), normalized to that of WT. Results shown as mean with standard deviation (n=3) **p < 0.01; ***p < 0.001; ns, not significant. Right: Aggregation end point samples were centrifuged to pellet fibrils for quantification by gel band density, from which percentage incorporation of each αS variant was calculated and normalized to that of WT. Results shown as mean with standard deviation (n≥3) **p < 0.01; ns, not significant. For both T_1/2 and incorporation percentage, mimic arginylated effects are similar to effects for the corresponding authentic arginylated construct (see Table 1 and SI, Figures S11-S12).

Table 1.

Aggregation kinetics and incorporation percentage

	Mimic CEArg	Authentic EArg
Arginylation Sitea,b	T1/2c	T1/2c
46 (5%)	0.89 ± 0.02	1.28 ± 0.10
46 (10%)	2.20 ± 0.03	1.69 ± 0.04
83 (5%)	1.46 ± 0.05	1.46 ± 0.11
83 (10%)	1.37 ± 0.10	1.62 ± 0.11
46 and 83 (5%)	1.89 ± 0.09	1.78 ± 0.12
46 and 83 (10%)	2.49 ± 0.04	2.80 ± 0.23
	Inc. %d	Inc. %d
46 (5%)	102 ± 8	97 ± 10
46 (10%)	100 ± 9	93 ± 3
83 (5%)	97 ± 6	81 ± 3
83 (10%)	70 ± 13	63 ± 7
46 and 83 (5%)	65 ± 1	62 ± 11
46 and 83 (10%)	65 ± 3	62 ± 4

^aPercent mixture with WT monomer indicated in parentheses.

^bThe values for authentically arginylated αS-E^Arg₄₆ and αS-E^Arg₈₃, and αS-E^Arg₄₆E^Arg₈₃, measured previously,⁶ are shown for comparison (10% αS-E^Arg₄₆ data were not performed previously).

^cT_1/2, determined relative to WT control run in parallel.

^dIncorporation percentage, determined relative to WT control run in parallel.

In addition to comparing the mimic C^EArg and authentic EArg modifications with regards to their influence on the kinetics of fibril formation, we examined the aggregation end point samples for total monomer incorporation. Samples from the final time point (48 h) of aggregation were pelleted to isolate fibrils from any remaining monomer. The fibrils were dissociated and run on a polyacrylamide (SDS-PAGE) gel, and band intensities were quantified to assess the amount of protein present (Figure 2, primary SDS-PAGE data are shown in SI, Figure S12). Similar to results seen with the authentic modification, although aggregation was slowed at 10% dosing, αS mimic arginylated at E₄₆ did not display any significant inhibitory effect on monomer incorporation at either 5% or 10% dosing. Mimic arginylation at E₈₃ reduced the extent of fibril incorporation, although this was only significant when αS-C^EArg₈₃ was present at 10%, showing that the mimic was not quite as potent as authentic arginylation in this case. The doubly mimic arginylated αS-C^EArg₄₆C^EArg₈₃ also displayed decreased fibril incorporation. In this case, both 5% and 10% αS-C^EArg₄₆C^EArg₈₃ led to similar decreases in incorporation, again consistent with effects observed for authentic arginylation. Thus, with one exception, the C^EArg mimic constructs quantitatively reproduced the effects observed with authentic E^Arg NCL constructs in fibril incorporation assays (Table 1).

In addition to further validating use of the arginylation mimic, several aspects of the aggregation data are notable, particularly since we had not performed aggregation with 10% E₄₆ arginylation in our previous report.⁶ First, although 5% arginylation at position 46 has no effect on aggregation, 10% dosing leads to a significant slowing of aggregation, but without altering incorporation (also consistent with a high final Congo Red ratio, Figure 2). On the other hand, either 5% or 10% arginylation at position 83 causes ~1.5-fold slowing of aggregation, but the effect on incorporation percentage is larger for 10%, regardless of whether mimic or authentic arginylation is used. Finally, although neither 46 nor 83 arginylation alters incorporation percentage at 5% dosing, 5% of the doubly arginylated construct causes a ~35% decrease in monomer incorporation. Collectively, we take these data to indicate that arginylation at position 46 or 83 exerts effects through a distinct mechanism, where position 46 may affect monomer more than fibril (leading to slower rates, but WT-levels of incorporation) and position 83 may affect fibril more than monomer. We were pleased to see that these relatively nuanced effects were reproduced by the C^EArg mimic, supporting its use in future detailed mechanistic studies.

Comparison of Effects on Neuronal Pathology Seeding.

Given the ability of the mimic arginylation to recapitulate the properties of the bona fide modification in these in vitro studies of αS function and disease-like behavior, we wanted to test the validity of the mimic in cellular contexts. To determine whether mimic arginylated αS fibrils display similar aggregation seeding capability as authentically arginylated αS in cultured neurons, we prepared PFFs consisting of WT αS mixed with 5% modified αS, for each authentic or mimic arginylated construct. We chose to prepare PFFs with only 5% arginylation because this is consistent with our observations of the levels of arginylation in αS from PD patient samples.⁷

Before testing seeding in mouse neurons, we examined the effects of PFF arginylation on in vitro WT αS aggregation reactions performed with PFF seeds that were 10% of total monomer concentration in the reactions (i.e., 10% PFF seeds, 90% WT αS monomer). Aggregation was carried out by shaking at 37 °C and the reactions were analyzed to determine T_1/2 and total monomer incorporation (Figure 3).

Figure 3.

Aggregation kinetics and total monomer incorporation for seeded aggregation assays. WT monomer starting material was mixed with various PFF seeds, and aggregated by shaking at 37 °C, and monitored by Thioflavin T. Left: Relative time to half completion of aggregation (T_1/2). Results shown as mean with standard deviation (n=6) **p < 0.01; ***p < 0.001; ns, not significant. Right: Aggregation end point samples were quantified as in Figure 2. Results shown as mean with standard deviation (n=6) *p < 0.05; **p < 0.01; ***p < 0.001.

Unlike the results seen with de novo aggregation where arginylation was inhibitory to varying degrees, seeding by αS PFFs significantly increased the rate and extent of fibril incorporation compared to WT PFFs for all arginylation constructs, with similar effects observed between pairs of authentic and mimic arginylated PFF seeds (SI, Figures S13-S14). The difference between the seeded and de novo aggregation data suggests that arginylation has different effects on the nucleation and elongation phases of aggregation, further motivating cell-based PFF seeding studies to observe which effect is dominant. Other than a small difference between the αS-E^Arg₄₆ and αS-C^EArg₄₆ seeding effects on aggregation rate, these results supported the validity of using the arginylation mimic (SI, Table S2), so all mimic constructs were carried forward to studies in neurons.

To test PFF seeding in a physiological context, mouse primary cortical neurons were grown for 2 weeks on coated coverslips and treated with 1 μg of each type of PFF per well (WT or 5% of the arginylation constructs), according to established protocols.^23-25 After 1 week, αS aggregates were detected with an antibody that detects phosphorylated serine 129 (pS₁₂₉), a marker for pathology. The total pS₁₂₉ signal (Figure 4 and SI, Figure S15) and the area of aggregates (SI, Figure S16) were quantified for neurons seeded with each type of PFF.

Figure 4.

Seeding of aggregation by authentic and mimic arginylated αS PFFs in mouse primary cortical neurons. Left: Representative images of neuron cultures treated with unmodified or arginylated αS PFFs as indicated, stained with pS₁₂₉ antibody (red) to reveal αS intracellular aggregates. Nuclei are stained with DAPI (blue). Scale bar, 10 μm. Right: Quantification of total pS₁₂₉ signal from neurons seeded by different αS PFFs. Results shown as mean with standard error: ns, not significant; **p < 0.01; ***p < 0.001, ****p < 0.0001.

In comparison to WT control PFFs, which produced pS₁₂₉ signal of 3.29 x 10⁵ ± 8.8 x 10⁴ AU SEM (arbitrary units, standard error of the mean), PFFs of 5% αS-E^Arg₄₆ and αS-C^EArg₄₆ decreased the total pS₁₂₉ signal to 2.17 x 10⁵ ± 9.1 x 10⁴ AU SEM and 2.12 x 10⁵ ± 5.6 x 10⁴ AU SEM, respectively. Similarly, seeding with PFFs of 5% αS-E^Arg₈₃ and αS-C^EArg₈₃ resulted in reduced total pS₁₂₉ signals of 1.98 x 10⁵ ± 5.4 x 10⁴ AU SEM and 2.47 x 10⁵ ± 6.0 x 10⁴ AU SEM, respectively with the αS-E^Arg₈₃ values just outside of significance. PFFs of αS arginylated at both sites also reduced total pS₁₂₉ signal to 2.10 x 10⁵ ± 5.3 x 10⁴ AU SEM for authentic αS-E^Arg₄₆E^Arg₈₃ and 2.14 x 10⁵ ± 6.9 x 10⁴ AU SEM for mimic αS-C^EArg₄₆C^EArg₈₃, showing that double arginylation did not confer additional inhibitory effects and that the mimic recapitulated the results of the authentic double modification.

A comparison of mean aggregate area (SI, Figure S16) produced a similar pattern of results to that seen with total pS₁₂₉ signal with all arginylated constructs reducing aggregate area. Notably, the effect of αS-E^Arg₈₃ was significant for the area measurements (p <0.1) and nearly identical to the αS-C^EArg₈₃ effect. WT PFFs produced an aggregate area of 340 ± 49 μm² SEM; PFFs of αS arginylated at E₈₃ resulted in an aggregate area of 236 ± 49 μm² SEM, and the mimic arginylation produced a very similar area decrease, to 246 ± 40 μm² SEM. The total pS129 signal and aggregate area data indicate that arginylation has significant inhibitory effects on aggregate seeding in neurons, and that the mimic once again reproduces the effects of authentic arginylation, as it does for the in vitro fibrillization assays.

The neuronal seeding results not only provide additional validation of the use of mimic arginylation, but also demonstrate that the effects of arginylation differ from those seen in vitro. In comparison to de novo aggregation, even 5% dosing arginylation at position 46 reduces seeding capacity at a level comparable to position 83 or double arginylation. For all sites, whereas in vitro PFF seeding accelerated aggregation more dramatically for arginylated PFFs than for WT PFFs, in neurons, arginylated PFFs had reduced seeding capacity. That the effects of arginylation are different in neurons could be attributable to several factors, including variations in seeding mechanism and the presence of additional PTMs in the αS pool in neurons that alter fibril incorporation. Given the successful validation of the C^EArg mimic in the studies presented here, we are excited to use this more facile modification to pursue such hypotheses.

Structural Comparisons of αS PFFs.

We wished to know whether distinct αS fibril structures led to the effects of arginylation on aggregation observed in vitro and in neurons. To examine whether arginylation alters αS fibril morphology, we performed transmission electron microscopy (TEM) imaging. There were no gross morphological differences observed in 5% modified PFFs relative to WT PFFs, as we have previously determined for authentic arginylation,⁷ and we here determined for the three mimic arginylation constructs (SI, Figure S17).

To provide a more detailed analysis of fibril conformation, we performed the common Proteinase K digestion assays on 5% mimic arginylated PFFs.²⁶ These assays required significant amounts of protein, so we performed digestion experiments with mimics only. PFFs were mixed with Proteinase K and incubated at 37 °C for 10, 20 or 30 min. The digestion reactions were resolved by SDS-PAGE, and each band intensity was quantified and normalized to that of uncleaved αS.

No major difference of band patterns was observed in the 10-15 kDa range, although differences in kinetics were observed (Figure 5, and SI, Table S3 and Figure S18). Arginylation at site 46 accelerated protease cleavage significantly, and most full-length protein was digested before the 10 min timepoint. Arginylation at site 83 did not affect digestion kinetics, except that the relative intensity of band 2 from was lower. While band 1 was not visible for this construct, this might be due to a slightly lower overall concentration of the PFFs. Arginylation at both sites accelerated digestion, although not as much as C^EArg₄₆ alone. The differences in proteinase K digestion indicate looser fibril packing for C^EArg₄₆ constructs, which may underlie the differences in seeding kinetics that we observed for the C^EArg₄₆ and E^Arg₄₆ constructs in comparison to the other arginylated constructs. For more detailed structural analyses, techniques such as solid state NMR or cryo-electron microscopy will be useful, and the more abundant arginylation mimic constructs will help to enable such studies.

Figure 5.

Proteinase K digestion of PFFs with 5% arginylation mimic. Each type of PFF was mixed with Proteinase K and incubated at 37 °C for 10, 20, or 30 min. Digested reactions were resolved on 12% SDS-PAGE gels stained with Coomassie Brilliant Blue dye. Major bands are labeled on the right (FL: Full-length αS). Molecular weight (MW) marker masses are in kD. Each band intensity was normalized to the intensity of the uncleaved PFFs. Results shown as mean with standard error of the mean from 3 independent digestion experiments.

CONCLUSION

In summary, we developed a synthetically facile way to access proteins bearing a mimic of the arginylation PTM, and demonstrated that this PTM protects against αS fibril formation both in vitro and in cellular contexts. To access the arginylation mimics, we synthesized the small molecule bromoacetyl arginine (3) and recombinantly produced αS cysteine mutants at specific sites of interest: 46, 83, or both. We reacted the αS-C₄₆, αS-C₈₃, or αS-C₄₆C₈₃ constructs with bromoacetyl arginine to produce a modification that very closely resembles glutamate arginylation, deviating only by a sulfur addition in the sidechain. We also demonstrated that the cysteine-small molecule reaction is compatible with the orthogonal incorporation of fluorophores via copper-catalyzed click chemistry. The yields of proteins made with mimic arginylation are twice those made with authentic arginyaltion through NCL, and more importantly, the number of steps is vastly reduced so that modified protein can be produced in a few days and reactions can be easily scaled up to produce material for more extensive in vitro, cellular, or structural studies.

We examined the native function of mimic arginylated αS by assaying its lipid vesicle binding affinity with FCS and probed the effects of mimic arginylation on the aggregation kinetics and fibril incorporation of αS. Our results showed that mimic arginylation recapitulates the effects of authentic arginylation in that it does not alter the vesicle affinity of αS, but does protect against fibril formation, with the modification at either or both sites slowing aggregation and/or decreasing the percentage of monomer found in the fibril end products. The nuanced differences in the effects of arginylation at position 46 or 83 could be further investigated using NMR or Förster resonance energy transfer (FRET) studies of the effects of arginylation on monomer conformation, as we have performed previously for WT αS and αS with other PTMs.^{20, 27} Differences in fibril conformation revealed by our Proteinase K experiments could be studied through cryo-electron microscopy, as Liu, Li, and coworkers have done for αS with Tyr₃₉ phosphorylation,²⁸ through FRET, as we have done for WT αS,¹⁹ or through solid state NMR.²⁹ Since our mimic arginylation results were consistent with those obtained with authentically arginylated αS afforded by semi-synthesis, we can attempt these structural and biophysical studies with more easily accessible mimic arginylated constructs. In cultured neurons, mimic and authentically arginylated αS PFFs displayed the same decrease in total pS₁₂₉ signal and area of seeded aggregates. Again, more in-depth cell-based investigations of seeding mechanism, such as using fluorescently-labeled constructs to evaluate effects of arginylation on fibril uptake,^24-25 can be performed using the Cys-based mimic.

Our findings provide further evidence for the protective nature of arginylation in αS and enable production of large quantities of arginylated αS for solid state NMR, more extensive mechanistic studies, or in vivo studies such as injection into mouse brain, a common model for Parkinson’s disease pathology (Note: Although only ~100 μg of protein is required for a cohort of mice, large quantities must be aggregated and then sterile-filtered).³⁰ While NCL has proven to be very valuable in studying the effects of PTMs in neurodegenerative disease,³¹ semi-synthetic PFFs have seen limited use in neuronal studies, in part due to the difficulties in preparation of modified protein, particularly with additional probes such as fluorophores.^32-33 Beyond the use of the C^EArg mimic with αS, our data provide an important direct evaluation of the mimicry of an authentic PTM by a Cys-based mimic for the protein semi-synthesis community. These types of study are relatively rare because of the synthetic challenges involved and sometimes reveal that the mimicry is not ideal.^{15-16, 34-38} Finally, the use of our arginylation mimic strategy should enable the exploration of this unusual PTM in other proteins identified in proteomics, including structural components of the cytoskeleton and key metabolic enzymes.^39-40