Terminal Alkynes as Raman Probes of α-Synuclein in Solution and in Cells

Abstract

Conformational changes of α-synuclein (α-syn) are central to its biological function and Parkinson’s disease pathology. Here, terminal alkynes (homopropargylglycine) were employed as environmentally sensitive Raman probes at residues 1, 5, 116, and 127, to characterize soluble (disordered), micelle-bound (α-helical), and fibrillar (β-sheet) α-syn. Along with the full-length protein, a disease-related C-terminal truncation (1–115) was also studied. For the first time, β-sheet α-syn amyloid structure was detected by the amide-I band in N27 dopaminergic rat cells, where a reciprocal relationship between levels of fibrils and lipids was seen. Site-specific spectral features of the terminal alkynes also revealed heterogeneity of the cellular environment. This work shows the versatility of Raman microspectroscopy and the power of unnatural amino acids in providing structural and residue-level insights in solution and in cells.

Small but mighty.

Small, terminal alkynes were biosynthetically incorporated into α-synuclein as vibrational reporters. These bio-orthogonal probes are very powerful in providing site-specific information on protein conformational changes in cells, such as protein-lipid interactions.

α-Synuclein (α-syn) is a small, conformationally dynamic protein implicated in the pathogenesis of Parkinson’s disease (PD). Natively, α-syn is intrinsically disordered and abundantly expressed in neurons.^[1] In a diseased state, α-syn accumulates in the cytosol and aggregates into amyloid fibrils, constituting the main proteinaceous component of Lewy bodies (LBs), cellular inclusions that are a histopathological hallmark of PD.^[2] Importantly, α-syn has a genetic connection with PD. Duplication or triplication of the SNCA gene, which encodes for α-syn, as well as missense mutations all lead to familial forms of PD.^[3] The primary sequence of α-syn can be described by three partially-overlapping regions (Figure 1A): the membrane binding region (residues 1–100) with 7 imperfect amphipathic repeats that form an α-helix in the presence of lipids, the β-sheet-containing amyloid core (residues 37–99), and the acidic C-terminal region (residues 100–140), which remains disordered and solvent-exposed in both membrane-bound and fibrillar states.^[4] Interestingly, all PD-related mutations (A18T, A29S, A30P, E46K, H50Q, G51D, A53E, and A53T) are found in the lipid-binding N-terminal region of α-syn.^[5] Collectively, these observations show an intimate relationship between α-syn and lipid interactions.

Figure 1.

Biosynthetic incorporation of HPG in α-syn. (A) Schematic of α-syn sequence, highlighting membrane binding, amyloid core, and acidic regions. Stars indicate HPG substitution sites (1, 5, 116, and 127). Blue boxes represent amphipathic repeats (consensus sequence KXKEGV). (B) Molecular structures for HPG and Met. (C) Representative Raman spectra of HPG-α-syn-treated (red) and untreated cells (black) in “cellular-quiet” region. Inset: Bright field images of HPG-α-syn-treated cell (right) and untreated cell (left). Raman data were collected at spatial locations along the arrows. Scale bars represent 10 μm. (D) Raman spectra of alkyne stretch of HPG in various solvents.

In this work, we have used an unnatural amino acid, homopropargylglycine, as a vibrational probe (C≡C stretch) and Raman microspectroscopy to structurally characterize α-syn in solution and in the presence of lipids, as well as translating these measurements into cells. The use of unnatural amino acids as vibrational probes is an area of active research.^[6] Raman microspectroscopy is a powerful vibrational technique that couples the chemical specificity of Raman spectroscopy with the spatial resolution of a microscope. Raman spectroscopy is commonly used to measure protein amide bands, which arise from coupled vibrational modes of the polypeptide backbone and directly report on protein secondary structure.^[7] Secondary structures of α-helix, β-sheet, or random coil all show characteristic peak maximum and shape, making protein conformational changes observable.^[8] Additionally, Raman spectra can easily be collected in water, an integral solvent for biological systems.^{[7a, 7b]} However, its utility in cellular studies has yet to be fully realized with respect to direct determination of protein structure due to the presence of overlapping biomolecular vibrations in similar spectral ranges.

To overcome this challenge, we have biosynthetically incorporated homopropargylglycine (HPG), an unnatural amino acid analogue of methionine (Met) ^[9] (Figure 1B) into α-syn at the four native Met sites (residues 1, 5, 116, and 127) through expression in a Met-auxotroph E. coli strain and defined growth conditions (See Supporting Information). This site-specific incorporation of a C≡C bond, a biorthogonal moiety, yields a stretching frequency that sits in the “cellularly quiet” region of the Raman spectrum (~1800–2400 cm^−1]), free of background from endogenous cellular components (Figure 1C) and is sensitive to local environment (Figure 1D and Table S1) as previously reported for a related compound, propargyl alcohol.^[10] Thus, this Raman probe yields a unique peak for α-syn, discriminating it from those of endogenous biomolecules. Previously, HPG has been used as a metabolic label for global protein synthesis in cell culture^[11] and fluorophore-labelling,^[12] but to the best of our knowledge, this is the first use of HPG to study protein-lipid interactions and amyloid formation.

To evaluate respective HPG signatures located in the N- vs. C-terminus, we compared the full-length (FL) α-syn with a C-terminally truncated variant, 1–115. The truncated protein contains only N-terminal HPG sites, as residues 116–140 are removed (abbreviated hereafter as 115t). C-terminal truncations of α-syn, including 115t, are biologically relevant because they are commonly found in LBs.^[13] Both HPG-containing proteins exhibited similar aggregation kinetics and formed fibrils as assessed by Thioflavin T emission and transmission electron microscopy, respectively (Figures S1 and S2). Similar to the wild-type protein,^[14] the HPG-containing α-syn also adopted α-helical conformation in the presence of zwitterionic lysophosphatidylcholine (LPC, Figure 2A) micelles (Figure S3). LPC is a relevant model system because it is found in synaptic vesicles,^[15] where α-syn is thought to perform its biological function,^[16] and resembles the highly curved membranes that α-syn prefers.^{[14, 17]} These results show the small alkyne probes are not perturbative and can be used as reporters of α-syn amyloid formation and lipid interactions.

Figure 2.

Raman spectroscopic comparison of full-length and C-terminally truncated α-syn in vitro. (A) Molecular structure of LPC. Representative Raman spectra of LPC in the amide-I (B) and C≡C (C) stretching regions. Averaged Raman spectra of amide-I (D) and C≡C (E) stretching regions of FL HPG-α-syn in solution (1 mM, purple, N = 4), in the presence of 100 mM LPC (pink, N = 5), and in the fibrillar form (10 μM, yellow, N = 2). Averaged Raman spectra of amide-I (F) and C≡C (G) stretching regions of 115t HPG-α-syn in solution (1 mM, blue, N = 5), in the presence of 100 mM LysoPC (red, N = 6), and in the fibrillar form (10 μM, green, N = 2). For ease of comparison, spectra are scaled and offest with the region between 1800 and 2040 cm⁻¹ omitted. Representative full spectra are shown in Figures S1 and S2. Shading in panels D–G represents 1 standard deviation of the mean. Dashed lines in panels D–G mark peak center positions.

Raman characterization of the two proteins in solution, in the presence of LPC micelles and in an aggregated fibrillar state (in Figure 2) was made on a home-built Raman microscope as previously described.^[7b] The saturated LPC has no Raman signals in the protein amide-I or alkyne stretching regions (Figure 2B and 2C). As expected for disordered polypeptides, soluble FL (Figure 2D, purple) and 115t HPG-α-syn (Figure 2F, blue) exhibit broad amide-I bands.^{[7b, 8]} However, distinctive maxima of 1673 cm⁻¹ and 1666 cm⁻¹ were observed for FL and 115t, respectively, suggesting that the removal of the C-terminal residues influences the conformational space that the protein samples. In the presence of LPC micelles, a dramatic spectral shift to lower energy is seen for both proteins with a clear peak arising around 1650 cm⁻¹ (Figure 2D, pink, Figure 2E, red), indicative of α-helical secondary structure^[8] upon micelle binding. Interestingly, a larger peak intensity ratio of the protein amide-I band (1650 cm⁻¹) to the LPC peak (1725 cm⁻¹) is observed for 115t, thus indicating a higher helical content compared to that of FL HPG-α-syn, which was also seen by circular dichroism spectroscopy (Figure S3). Thus, the C-terminal residues appear to impede α-helical formation in the FL protein, potentially by imposing a steric effect, making it more difficult to wrap around the micelle.^[18] In their aggregated, fibrillar (Figure 2D, yellow, Figure 2F, green) states, characteristic peaks for β-sheet ^{[7b, 8]} associated with amyloid formation appear at 1666 cm⁻¹.

The HPG alkyne-stretching peak is observable for both α-syn variants under all conditions examined (Figures 2E and 2G), which is remarkable given that there are only four and two alkynes per protein for FL and 115t, respectively. Relative peak intensities of amide-I and C≡C can be viewed in Figure S2. For soluble FL protein, the C≡C peak is centered at 2112 cm⁻¹ (full width at half maximum (FWHM) of 18 cm⁻¹), with a shoulder at ~2092 cm⁻¹. Upon lipid binding, a subtle shift to lower wavenumbers (2110 cm⁻¹; FWHM = 17 cm⁻¹) is seen with the same broad shoulder, indicating a more hydrophobic environment (Figure 1D), consistent with a lipid-bound state. In the fibrillar form, the lower energy shoulder disappears and a peak remains at 2113 cm⁻¹ (FWHM of 17 cm⁻¹). This is intriguing because it reveals that the environment HPG moieties experience has changed, despite the fact that the N- and C- termini are expected to remain unstructured, outside the amyloid core. Consistent with this observation, the greatest variation of peak position is between fibrillar FL (2113 cm⁻¹) and the HPG model in water (2116 cm⁻¹). Of note, for soluble and LPC-bound 115t, the lower energy shoulder is missing, suggestive that this band likely arises from the C-terminal alkynes at positions 116 and 127. The main HPG peak is measured at 2114, 2112, and 2114 cm⁻¹ with a FWHM of 18 cm⁻¹ for soluble, lipid-bound, and fibrillar 115t. Also, a small peak is observed ~2072 cm^−1] for 115t fibrils, suggesting there is greater change of the N-terminal HPG sidechains in the fibrillar state.

We moved next to interrogating α-syn conformation inside cellular environments. In these experiments, cultured N27 rat dopaminergic cells were treated with fibrils formed in vitro from FL and 115t HPG-α-syn. To provide spectral contrast in the amide-I region, the proteins were uniformly isotopically-labelled with ¹³C,^[19] with the exception of the HPG side chains. Figures 3A and 3B show brightfield images of N27 cells after a 24-h treatment with FL and 115t ¹³C-HPG-α-syn fibrils, respectively. Upon fibril treatment, the membranes are noticeably perturbed with blebs compared to untreated cells (Figure 1C, left inset).

Figure 3.

Raman microspectroscopy of cultured N27 rat dopaminergeric cells after 24-h treatment of fibrils formed in vitro. Brightfield images of N27 cell treated with 20 μM FL ¹³C-HPG-α-syn (A, expanded image of right inset from Fig. 1) and with 25 μM 115t ¹³C-HPG-α-syn (B). Scale bars represent 10 μm. (C) Averaged Raman spectra of HPG region for untreated cell (light gray, N = 6), FL-treated cell (red, N = 12, individual spectrum are shown in Fig. 1C), 115t-treated cell (purple, N = 6), compared to in vitro FL fibrils (gray, N = 3), and in vitro 115t fibrils (black, N = 3) in the presence of LPC. (D) Spatially-resolved Raman spectra collected from untreated (light gray), FL-treated (red-gradient spectra), and 115t-treated (purple-gradient spectra) cells compared to in vitro FL (gray) and 115t (black) fibrils in the presence of LPC micelles. Spectra are offset for ease of comparison. Color of spectrum corresponds to spatial position along gradient arrows in (A and B). Additional cellular Raman data shown in Figure S4.

Similar to in vitro measurements, the HPG signature is clearly visible in fibril-treated cells (Figure 3C). The position and width of this peak, however, is noticeably different in cultured cells for FL compared to 115t. When averaged from different spatial locations across the cell, the FL HPG peak is shifted and broader than that of 115t (2113 cm⁻¹; FWHM = 22 cm⁻¹ vs. 2117 cm⁻¹; FWHM = 13 cm⁻¹), which is reasonable as it has four HPG moieties. We interpret this as an indication of the C-terminal alkynes experiencing different local environments than the N-terminal ones. The spectral features of 115t in cells are also narrower and shifted compared to in vitro 115t fibrils with an additional second peak at 2140 cm⁻¹, suggesting an influence of the cellular environment or interactions. To test if the membranes could be involved, LPC was added to fibrils formed in vitro. However, the fibril spectrum remained similar in the presence of LPC (2114 cm⁻¹; FWHM = 20 cm⁻¹). In contrast, the peak for FL HPG-α-syn in cells is very similar to in vitro fibrils in the presence of LPC (2113 cm⁻¹; FWHM = 22 cm⁻¹), and slightly broader than that of fibrils alone (2113 cm⁻¹; FWHM = 17 cm⁻¹).

These results indicate changes in the local environments of HPG sidechains in cells. Considering the differences seen for FL and 115t, we suggest that the cellular interactions of the N- and C-terminal regions of α-syn are distinct. While these interactions cannot be directly evaluated, a simple unfolding of the fibrillar material is ruled out because the amide-I region of both FL and 115t fibrils maintain β-sheet character (1623 cm⁻¹) after being taken up by cells (Figure 3D). The site-specific terminal alkyne, therefore, demonstrates an ability to report on the heterogeneity of the cellular environment, even when the bulk measurement of secondary structure indicates little change in β-sheet conformation.

Interestingly, we observed variations in the β-sheet-to-lipid peak ratios (β-sheet (C=O) and lipid (C=C) peaks^[20] at 1623 and 1658 cm⁻¹, respectively) as a function of spatial locations across each cell. We see a similar inverse correlation between the intensities of these two peaks for both FL and 115t HPG-α-syn-treated cells. We interpret this relationship as first, an indication of amyloid and lipid co-localization and second, an influence on local lipid concentration by the fibrils. In particular, for FL-treated cells, the average intensity of the lipid peak at 1658 cm⁻¹ is consistently stronger than that observed in 115t-treated cells, suggesting there is more lipid accumulation by the FL fibrils. This observation is intriguing and mirrors recent ultrastructural cryoelectron microscopic images which revealed LBs to be very crowded environments composed of both lipid membrane and amyloid materials.^[21]

Collectively, our data demonstrate the utility and power of bio-orthogonal labelling strategies for studying α-syn conformation and interactions in solution and in cells. The environmental sensitivity of the terminal alkyne HPG is extremely valuable as a site-specific Raman probe. By coupling uniform ¹³C-isotopic labelling of the peptide backbone (C=O) and site-specific unnatural amino acid substitution (C≡C), we showed that β-sheet secondary structure of amyloid fibrils and region-specific differences in α-syn interactions can be observed in cellular environments, a feat not previously achieved. From this work, an unexpected effect of the C-terminal region is revealed. In the absence of C-terminal residues, the HPG peak in 115t is distinct from the full length in cells but not those measured in vitro. Together with the observed lower lipid peak intensity for 115t, we conclude that the C-terminal region of α-syn plays a more active role in cellular interactions than previously anticipated as this region is often overlooked. Clearly, more work is needed to establish whether these observations involve lipid membranes and are related to mechanisms of cytotoxicity exerted by amyloid fibrils. Nevertheless, this successful demonstration of incorporating bio-orthogonal vibrational probes into proteins for Raman studies is broadly applicable and this strategy can be utilized to gain insights into site-specific conformational changes in other amyloidogenic peptides and proteins.

Experimental Section

Experimental details are available in the online supporting information.