pH-Dependent Fibril Maturation of a Pmel17 Repeat Domain Isoform Revealed by Tryptophan Fluorescence

Abstract

The pre-melanosomal protein (Pmel17) aggregates within melanosomes to form functional amyloid fibrils that facilitate melanin polymerization. The repeat domain (RPT) of Pmel17 fibrillates under strict acidic melanosomal pH. Alternative splicing results in a shortened repeat domain (sRPT), which also forms amyloid fibrils. Here, we explored the effects of pH and protein concentration on sRPT aggregation by monitoring the intrinsic fluorescence of the sole tryptophan at position 381 (³⁸¹W). ³⁸¹ W emission properties revealed changes of local environment polarity for sRPT fibrils formed at different pH. At pH 4, fibrils formed rapidly with no lag phase. A high ³⁸¹W intensity was observed with a slight blue shift (10 nm). These fibrils underwent further structural rearrangements at intermediate pH (5–6), mirroring that of melanosome maturation, which initiates at pH 4 and increases to near neutral pH. In contrast, typical sigmoidal kinetics were observed at pH 6 with slower rates and ³⁸¹W exhibited quenched emission. Interestingly, biphasic kinetics were observed at pH 5 in a protein concentration-dependent manner. A large ³⁸¹W blue shift (23 nm) was measured, indicating a more hydrophobic environment for fibrils made at pH 5. Consistent with ³⁸¹W fluorescence, Raman spectroscopy revealed molecular level perturbations in sRPT fibrils that were not evident from circular dichroism, transmission electron microscopy, or limited proteolysis analysis. Finally, sRPT fibrils did not form at pH ≥7 and preformed fibrils rapidly disaggregated under these solution conditions. Collectively, this work yields mechanistic insights into pH-dependent sRPT aggregation in the context of melanosome maturation.

1. Introduction

The aggregation of soluble proteins into β-sheet-rich insoluble fibrils, called amyloid, is implicated in the etiology of ~50 human diseases [1]. While this has largely led to an association of amyloids with pathology, the discovery of “functional amyloids” over the past two decades has challenged this notion [2]. In human melanocytes, the pre-melanosomal protein (Pmel17, also known as SILV or gp100) forms functional amyloid fibrils that promote melanin biosynthesis [3]. This occurs within the melanosome, an acidic organelle similar to the lysosome, which undergoes four distinct stages of maturation. In vivo, Pmel17 fibrils form during the early stages of maturation (I and II), upon which melanin is synthesized during the later stages (III and IV) [4, 5]. It has been shown that the intralumenal pH increases during melanosome maturation, beginning at ~pH 4 in stages I and II and reaching near neutral pH upon full maturation [6]. Pmel17 fibrils serve as a scaffold for melanin polymerization, which also offer the benefit of sequestering toxic intermediates of melanin biosynthesis [3].

Pmel17 is a 668 amino-acid protein containing a single transmembrane domain, which undergoes several stages of processing prior to fibril formation within early stage melanosomes (Figure S1) [7–9]. The luminal domain of Pmel17, otherwise known as Mα, contains the amyloid forming region [10] and is made up of an N-terminal domain, a polycystic kidney disease-like domain, and a repeat (RPT) domain (Figure S1). While the exact domain(s) within Mα that is responsible for amyloid formation is not clearly understood [11–14], previous work has established that the RPT domain forms amyloid fibrils in vitro under pH solution conditions found in early stages of melanosomes [12, 13, 15–20].

The RPT domain comprises 10 imperfect repeats of 13 amino acids, each containing a high degree of proline, threonine, serine, and glutamic acid residues. Nichols et al. [21] have shown that alternative splicing of a “cryptic intron” within RPT leads to a shorter sequence (sRPT, Figure 1a) which lacks the 6^th and 7^th repeats and portions of the 5^th and 8^th repeats. This shorter sRPT domain has also been shown to form amyloid fibrils in acidic conditions in vitro. Notably, when exposed to neutral pH, sRPT fibrils disassemble [16], displaying reversible aggregation/disaggregation. This unique feature of RPT has not yet been documented for pathological amyloids, which are known to withstand harsh solution conditions. It is clear that the protonation/deprotonation of acidic residues (of which there are 10 Glu and 1 Asp, colored red in Figure 1a) plays an important role in the mechanism of amyloid formation of RPT [15].

Fig. 1.

Effect of pH on sRPT aggregation. (a) The one-letter amino acid sequence of sRPT. The underlined region represents the amyloidogenic core as defined in this work. Acidic and basic residues are colored red and blue, respectively. ³⁸¹W is colored in purple. (b) sRPT (30 µM) aggregation kinetics at pH 4 (red), 5 (yellow), 6 (green), 7 (cyan), and 8 (blue), monitored by ³⁸¹W fluorescence at 37 °C (n = 2 microplate experiments, 5 replicates each). (c) ³⁸¹W fluorescence at t₀ and t_end. (d) ThT binding (10 µM) to sRPT at t₀ and t_end. (e-j) Representative TEM images of sRPT aggregates formed at pH 4 (e, h), 5 (f, i), and 6 (g, j).

Here, the effects of pH on sRPT amyloid formation were examined in detail. Using the intrinsic fluorescence of the sole tryptophan residue at position 381 (³⁸¹ W) within the C-terminus of sRPT (Figure 1a), we showed that rates of aggregation increase with protein concentration and are highly dependent on pH (pH 4 > 5 > 6). Atypical for amyloid formation, sRPT fibrils formed quickly at pH 4 with no apparent lag phase, which upon a solution pH shift to 5 or 6, the fibrils reorganize at the C-terminus as indicated by ³⁸¹W spectral changes. A similar fibril maturation process, albeit at a much faster timescale and only at protein concentration ≥ 30 µM, also was observed at pH 5. Aggregation at pH 6 exhibited the expected sigmoidal kinetics trend. Corresponding secondary structural changes upon amyloid formation were investigated by circular dichroism (CD) and Raman spectroscopy. Fibril morphology was characterized by transmission electron microscopy (TEM). Both ³⁸¹W emission and Raman spectral features revealed molecular level perturbations in sRPT fibrils as a function of pH that were not evident from CD, TEM, or limited proteolysis analysis. In light of these results, we discuss the potential biological implications of the pH-dependent sRPT aggregation kinetics and fibril stability as they relate to the melanosome maturation process.

2. Materials and Methods

2.1. Materials

Unless otherwise noted, all reagents were purchased from Sigma-Aldrich.

2.2. Methods

2.2.1. sRPT protein expression and purification.

An sRPT construct containing a C-terminal hexa-histidine tag was expressed in E. coli BL21(DE3) cells and purified using affinity chromatography as described previously [16], with the following minor modifications to buffer conditions. The lysis buffer (pH 7.6) contained 6 M guanidine hydrochloride (GuHCl; Invitrogen), 100 mM NaCl, 100 mM Na₂HPO₄, and 10 mM imidazole. The Ni-NTA (Qiagen) column was equilibrated in 8 M urea, 100 mM NaCl, 100 mM Na₂HPO₄, 20 mM imidazole, pH 7.6 buffer and sRPT was eluted with the same buffer containing 200 mM imidazole. Purified protein was dialyzed against 20 mM tris(hydroxymethyl)aminomethane (Tris; Invitrogen), pH 8.0 and stored at −80 °C until use. The protein purity was first confirmed by SDS-PAGE analysis, which migrated with an apparent mass of ~50 kDa. The monoisotopic mass of 9694.5 Da was further confirmed by mass spectrometry (NHLBI Biochemistry Core), which displayed a purity greater than 95%.

2.2.2. Aggregation kinetics.

Stock solutions of sRPT were first buffer exchanged (pH 4 and 5: 20 mM sodium acetate, 100 mM NaCl; pH 6: 20 mM 2-(N-morpholino)ethanesulfonic acid, 100 mM NaCl; pH 7: 20 mM sodium phosphate, 100 mM NaCl; pH 8: 20 mM Tris, 100 mM NaCl) using a Sephadex G-25 PD-10 column (GE Healthcare) following the manufacturer’s protocol. After passing through a 100 kDa Amicon filter (Millipore), the protein concentrations were determined on a Cary 300 Series UV-Vis Spectrometer (Agilent Technologies) using the intrinsic ³⁸¹W absorbance at 280 nm (ε = 5,500 M⁻¹ cm⁻¹). Reactions of 70 µL were then made into black 384-well polypropylene flat bottom plates (Greiner Bio-One) supplemented with a 2 mm borosilicate glass bead, and sealed with a MicroAmp optical adhesive film (Thermo Fisher Scientific). Reactions were monitored using a SPARK Multimode Microplate reader (Tecan) maintained at 37 °C with constant linear shaking (6 mm, 330 rpm). Fluorescence measurements were collected from the top using excitation and emission wavelengths of 280 and 350 nm, respectively.

2.2.3. Fluorescence spectroscopy.

Fluorescence measurements were collected on a Fluorolog FL-3 instrument (Horiba Scientific). For thioflavin-T (ThT) fluorescence, the dye was added to a final concentration of 10 µM. Emission was monitored from 430–600 nm (2 nm slit width) while exciting at 415 nm (1 nm slit width). For ³⁸¹W measurements, emission was monitored from 290– 450 nm (2 nm slit width) while exciting at 280 nm (1 nm slit width). All data have their respective buffer backgrounds subtracted. In addition, ³⁸¹W data were further scaled by relative protein concentration.

2.2.4. Transmission electron microscopy.

Transmission electron microscopy (TEM) images were collected on a JEM 1200 EXII microscope (JEOL USA) equipped with an XR-60 digital camera (Advanced Microscopy Techniques) operating at 80 kV (NHBLI Electron Microscopy Core). Grids were prepared by depositing 5 µL of solution onto a 400-mesh copper grid with a formvar/carbon film (Electron Microscopy Sciences) and allowing it to adhere for 1 min. Excess solution was wicked away using grade 1 Whatman filter paper (GE Healthcare). Grids were then stained for 30 s using 5 µL of 2% (w/v) uranyl acetate, followed by wicking. Finally, grids were dried overnight at RT before collecting images.

2.2.5. Limited cathepsin L proteolysis.

Reactions were prepared in sterile, 1.5 mL microcentrifuge tubes by adding cathepsin L (CtsL; 216 nM, Sigma #C6854) to a solution of preformed sRPT aggregates (30 µM, generated at pH 4, 5, or 6 as described in section 2.2.2) in reducing buffer (DTT, 5 mM). After shaking (575 rpm) at 37 °C overnight, reactions were transferred to LC-MS vials and TFA was added to a final amount of 0.5% (w/v), followed by GuHCl to a final concentration of 4 M. Samples were then subjected to LC-MS analysis using a Agilent 6200 series ESI-TOF LC-MS instrument (NHLBI Biochemistry Core). Deconvolution of the primary eluting species was then performed using MassHunter software (Agilent Technologies) to identify peptide fragments.

2.2.6. Circular dichroism spectroscopy.

Circular dichroism (CD) measurements were collected on a JASCO J-715 instrument (NHLBI Biophysics Core) in a 1-mm quartz cuvette. Scans from 260–195 nm (1 nm data pitch) were collected at a rate of 50 nm/min with a 1-s response time. A total of three spectra were averaged. Mean residue ellipticity (θ_MRE) was calculated using the following equation, as described previously [22]:

$${\theta }_{MRE}=\frac{{\theta }_{deg}MRW}{10dc}$$

where MRW represents the mean residue weight (molecular weight/number of peptide bonds), θ_deg represents the ellipticity (degrees) as measured by the instrument, d and c represent the pathlength (cm) and protein concentration (g/mL), respectively.

2.2.7. Raman spectroscopy.

Raman spectra were collected using a home-built instrument as previously described [23]. Samples were concentrated by centrifuging at 17,000 g for 30 min, followed by deposition of a 20 µL droplet onto an 8-well #1 coverglass Nunc LabTek chamber. Spectra represent an average of 5 collections from varying spatial locations, with each collection constituting 30 accumulations (10 s integration time). Data were processed first by buffer subtraction and then by first-polynomial baseline correction using LabSpec 6 software (Horiba Scientific). For comparison, the amide I (1665 cm⁻¹) intensity was normalized to 1.

2.2.8. Dilution/disaggregation kinetics.

For dilution experiments, aggregates were diluted 6-fold into the specified buffer and aliquoted into 384-well microplates. ³⁸¹W fluorescence was then monitored as described in section 2.2.2 above. For disaggregation experiments, measurements were collected on a Fluorolog FL-3 instrument (Horiba Scientific). A magnetic stir bar was added to a 10-mm quartz cuvette, followed by the appropriate buffer. Measurements were then initiated under constant stirring (setting = 4) by monitoring the emission at 350 nm (2 nm slit width) while exciting at 280 nm (1 nm slit width, 100 ms integration) using the kinetics mode (FluorEssence V3.8, Horiba Scientific). After ~10 s, sRPT was added directly into the cuvette, such that the final dilution was 6-fold. Disaggregation kinetics were monitored for a total of 5 min.

3. Results

3.1. Effect of pH on sRPT aggregation.

The effect of pH on sRPT aggregation kinetics were evaluated by monitoring ³⁸¹W fluorescence, which has been previously shown to directly report the aggregation state of the longer RPT isoform (lRPT) [18]. We considered this to be a superior approach to the commonly utilized amyloid-binding dye thioflavin-T (ThT) [24], since ³⁸¹W is a native probe that directly reports on aggregation. ThT fluorescence was compared with ³⁸¹W, which showed similar kinetic profiles (data not shown). At pH 4, the ³⁸¹W fluorescence rapidly increased within the first two hours with no lag phase observed, then plateaued for the remainder of the reaction (Figure 1b). A similar initial increase in ³⁸¹W fluorescence was observed at pH 5, albeit a bit slower, which was then followed by a second transition before plateauing at a significantly lower fluorescence intensity (Figure 1b). We note that there was a small decrease in fluorescence intensity from 0 to 1 h as a result of temperature equilibration. The initial increase in ³⁸¹W fluorescence at pH 5 may represent the formation of an oligomeric intermediate, which was investigated and is described in detail below. In contrast, after a lag phase of ~6 h, a single transition was observed at pH 6 before plateauing at a final fluorescence intensity similar to that observed at pH 5 (Figure 1b). No changes in the ³⁸¹W fluorescence was observed over the course of the reaction at pH 7 or pH 8 (Figure 1b), which was also evident by comparing ³⁸¹W emission scans from the beginning to end of the reaction (Figure 1c). Clearly, sRPT aggregation is highly pH dependent from pH 4–6 and no aggregation occurs beyond pH 7.

Fluorescence of ³⁸¹W at the end of reactions at pH 4, 5, and 6 revealed blue shifts in the emission maxima (Figure 1c), indicating the sidechain resides in a more hydrophobic environment. This is expected for sRPT aggregation, as the protein is intrinsically disordered (IDP) in its soluble form [16] and would therefore have a solvent-exposed ³⁸¹ W at t₀. It is important to note that the magnitude of this blue-shift varies from pH 5 (23 nm, 346 to 323 nm) to pH 4 and 6 (both 10 nm, 346 to 336 nm). We attribute this observation to the formation of different fibril polymorphs, which has been shown previously for other amyloidogenic proteins as a function of pH [25–28]. In addition, the quenched emission intensity observed at pH 5 and 6 may be due to π-stacking of aromatic rings within the parallel-in-register alignment of β-strands in the final amyloid structure, indicating that fibrils formed at pH 4 may have a distinct fibril architecture from those of pH 5 and 6.

The absence of fibril formation at pH 7 and 8 was further confirmed by the lack of ThT fluorescence at the end of reactions (Figure 1d). However, ThT enhancement was seen for pH 4, 5, and 6 samples, albeit with varying intensities (Figure 1d). Interestingly, the ThT binding intensity seemed to correlate with the amount of aggregates observed by transmission electron microscopy (TEM, Figure 1e–g), with pH 4 grids being most densely populated with filamentous aggregates. Magnified images (Figure 1h–j) revealed that pH 4, 5, and 6 aggregates had a similar morphology containing twisted protofilaments. Thus, despite having ultrastructural similarities, the local environments of ³⁸¹W are distinct in these fibrils.

To investigate if the observed differences in the local ³⁸¹W environment would manifest in the protease-resistant amyloidogenic core, cathepsin L (CtsL) was used to digest fibrils formed at pH 4, 5, or 6 (Figure S2). CtsL was chosen due to it being ubiquitous in lysosomes, an acidic organelle similar to the melanosome, as well as previous reports detailing its potency to proteolyze α-synuclein amyloid fibrils [29, 30]. Upon digestion, a large fragment (7.7 kDa) constituting residues 336 to the C-terminus was identified in all samples, suggesting that the amyloid core of pH 4, 5, and 6 fibrils does not vary. This result is consistent with the previously reported proteinase K-resistant core of residues 333 to the C-terminus for sRPT fibrils formed at pH 5 [16].

3.2. Secondary structural characterization of sRPT fibrils.

To further probe whether there are other measurable differences in the sRPT fibrils formed at different pH, their secondary structures were examined using CD and Raman spectroscopies (Figure 2). As mentioned prior, sRPT is an IDP in its soluble form, which was reflected in the CD spectra collected at t0 (Figure 2a). No transition from this disordered structure was observed at pH 7 and 8 after incubation (48 h), further corroborating that no amyloid formation has occurred (Figure 2b). At pH 4, a broad minimum was observed at ~208 nm with a shoulder near 220 nm. While the latter absorption is indicative of β-sheet formation, the former band suggests the presence of α-helical conformation. A similar spectral profile was observed for pH 5 and 6 samples, with the signal intensity seeming to correlate with the amount of aggregates observed by TEM (Figure 1e–g). A nearly identical CD spectrum for sRPT was previously reported, where additional electron diffraction and solid-state nuclear magnetic resonance data suggested the formation of cross-β structure [16].

Fig 2.

Secondary structural characterization of sRPT fibrils. (a-b) CD spectra of sRPT (30 µM) aggregation reactions at pH 4 (red), 5 (yellow), 6 (green), 7 (cyan), and 8 (blue) at the beginning (t₀, a) and end (t_end, b) of the reactions. Θ is defined as the mean residue ellipticity. (c) Raman spectra of sRPT aggregates at t_end. (d) Expanded view of 1200–1380 cm⁻¹ region of the spectrum.

To bring more clarity to the secondary structure of sRPT, fibrils were characterized by Raman spectroscopy (Figure 2c). Raman spectroscopy has been previously shown to provide unique molecular fingerprints of different amyloids, including lRPT [31]. The Raman fingerprint of sRPT aggregates formed at pH 4, 5, and 6 were largely similar. A sharp peak at 1665 cm⁻¹ was observed, which corresponds to the amide I band arising from C=O stretching vibration in the protein amide backbone. This is characteristic of β-sheet secondary structure [31, 32], confirming the formation of β-sheets in sRPT fibrils. Similar bands also were observed at 1552 and 1454 cm⁻¹, which represented vibrations from Trp sidechains and C–H deformation, respectively [32]. However, upon closer inspection of the 1200–1380 cm⁻¹ region of the spectrum, differences in the relative intensities of the 1320 and 1348 cm ⁻¹ peaks could be discerned (Figure 2d). The intensity ratio (I₁₃₂₀/I₁₃₄₈) at pH 4 (0.9) was closer to 1 as compared to pH 5 and 6, which were both lower (~0.3). These vibrations have been assigned to C–H deformation frequencies of the sidechains [23]. We interpret that while the overall secondary structures of the fibrils are similar, the sidechain packing in the pH 4 fibrils are different.

3.3. Disaggregation kinetics of sRPT fibrils.

³⁸¹W fluorescence was employed to observe the kinetics of sRPT disaggregation upon dilution into pH 7 buffer (Figure 3). Fibrils (30 µM) grown at pH 4 (F_{pH 4}, Figure 3a), 5 (F_{pH 5}, Figure 3b), or 6 (F_{pH 6}, Figure 3c) were diluted 6-fold into their corresponding buffer (red, yellow, and green curves, respectively) or into pH 7 buffer (cyan curves). The vertical dashed line represents the moment when aggregates were added to a cuvette containing buffer, which was subjected to constant mixing using a magnetic stir bar. Upon self-dilution (i.e. F_{pH 4} into pH 4 buffer), no changes in ³⁸¹W kinetics were observed over the five minute reaction period for F_{pH 4}, F_{pH 5}, or F_{pH 6}. However, diluting all aggregates into pH 7 buffer resulted in disaggregation, as evident by a lack of ThT emission increase at the end of the reaction (discussed in more detail below). For F_{pH 5} and F_{pH 6}, ³⁸¹W intensity increased upon mixing and reached a plateau by the end of the reaction period. In contrast, for F_{pH 4}, the signal decreased upon disaggregation. For ease of comparison, changes in ³⁸¹W were normalized and are shown in Figure 3d. Clearly, disaggregation of F_{pH 5} (yellow) and F_{pH 6} (green) occurred immediately and proceeded with similar rates, while there was a ~7 s lag time prior to F_{pH 4} (red) disaggregation (Figure 3d inset).

Fig. 3.

Disaggregation kinetics of sRPT fibrils. Fibrils formed at pH 4 (a), 5 (b), or 6 (c) were diluted 6-fold into their corresponding buffers (red, yellow, and green curves) or into pH 7 buffer (cyan curves). The vertical dashed line represents the timepoint when sRPT aggregates were introduced. (d) Normalized ³⁸¹W fluorescence change after diluting into pH 7 buffer for fibrils formed initially at pH 4 (red), 5 (yellow), and 6 (green). (Inset) an expanded view of the first 30 s. The data are representative of two independent experiments.

3.4. Effect of protein concentration on sRPT aggregation.

The pH-dependent aggregation reactions mentioned thus far were all measured at a single protein concentration (30 µM). To investigate the concentration-dependence of the unique pH-dependent behavior, aggregation kinetics reactions were performed at lower protein concentrations (Figure 4a–c). The 30 µM reactions from Figure 1b are reproduced in Figure 4d for comparison. Similar kinetic trends were observed at pH 4 (red curves), where the ³⁸¹ W fluorescence rapidly increased within a few hours. Similar to what was observed at 30 µM for pH 5 (Figure 1a), a subsequent decrease in the ³⁸¹W intensity was observed for lower concentrations at pH 4 (Figure 4a–b), revealing a second kinetic phase. Plotting the time at which the signal intensity reached half the maximum value (t_½) revealed that at pH 4, aggregation rates increased with increasing protein concentration from 3–15 µM (Figure 4e). However, the magnitude of this concentration-dependent difference was negligible compared to that observed at pH 5 and 6 (Figure 4e).

Fig. 4.

Effect of protein concentration on sRPT aggregation. (a-d) sRPT aggregation at pH 4 (red), 5 (yellow), and 6 (green) at 3 (a), 7.5 (b), 15 (c), and 30 µM (d, reproduced from Fig 1b), respectively. (e) Mean (<t_1/2>) values for each concentration at the specified pH. Error bars are indicative of the standard deviation from the mean. n = 10, two independent microplate experiments. No aggregation was observed for 3 µM sRPT at pH 6. (f) Normalized ³⁸¹W fluorescence for sRPT aggregation at 30 (light yellow), 60 (gold), and 150 µM (black) in pH 5 buffer.

At pH 6 (green curves), from 7.5–30 µM (Figure 4b–d), the observed t_½ values decreased with increasing concentration (Figure 4e). No aggregation was observed at 3 µM using ³⁸¹ W as a probe (Figure 4a). Excluding 30 µM, aggregation at pH 5 (yellow curves) displayed a single sigmoidal decay (Figure 4a–c) similar to that observed at pH 6. To ensure this concentration-dependent behavior at pH 5 was not the result of specific aggregation conditions, reactions were performed in quiescent conditions in the absence of a glass bead (Figure S3). Similar results were observed, where biphasic kinetic trends were only observed at 30 µM (Figure S3d). While the apparent biphasic kinetics vary with protein concentration at pH 5, the t_½ displayed a linear trend (Figure 4e; the initial phase was considered when calculating t_½ at 30 µM). Interestingly, the second phase (from high to low ³⁸¹W intensity) was inhibited by increasing the protein concentration beyond 30 µM (Figure 4f), which may suggest the formation of kinetically-trapped oligomeric species. Further characterization is needed to understand this behavior at high protein concentration.

3.5. Characterization of an sRPT intermediate.

To explore the possibility of an sRPT oligomeric intermediate being formed at pH 5, 30 µM aggregation reactions were initiated as described before (Figure 5a). As a control, a reaction was started concomitantly at 10 µM, a protein concentration in which biphasic kinetics was not observed. As expected, the ³⁸¹W fluorescence of the 30 µM reaction began to increase after ~1 h and continued to increase for an additional ~1 h. No changes in the ³⁸¹W fluorescence of the 10 µM control reaction was observed in this time frame. At this point, reactions were stopped and individual samples of the 30 µM reaction were pooled together (t_int) for further characterization. Collecting ³⁸¹W emission scans showed that fluorescence intensity increased and a blue-shifted maximum was observed at t_int compared to the soluble protein at t₀ (Figure 5b). It is interesting that the blue-shift from t₀ to t_int (6 nm, 346 to 340 nm) was smaller than that of fibrils formed after several days of agitation (t_end), which had a final blue-shift of 23 nm (346 to 323 nm; Figure 5c). Moreover, a difference in ThT fluorescence was observed with the t_end point having ~4-fold higher intensity compared to t_int (Figure 5d). Samples at t_int were also visualized by TEM (Figure 5e–g), which surprisingly showed elongated fibrillar aggregates with twisted protofilaments similar to that observed at t_end. Contrary to our initial thought, the prompt increase in ³⁸¹W fluorescence does not represent oligomer formation, but rather is due to the formation of fibrils.

Fig. 5.

Characterization of an sRPT intermediate. (a) sRPT aggregation at 10 (light gray) and 30 µM (dark gray) at pH 5. After ~2 h, reactions were stopped and ³⁸¹W (b-c) and ThT (d) fluorescence measurements of the 30 µM reaction (t_int) were collected. sRPT fibrils formed at pH 5 (t_end, black) and soluble sRPT at pH 5 (t₀, light gray) are also shown for comparison. (d-f) TEM images of t_int samples. Scale bars indicate 200 nm. (g) CD spectra of sRPT samples, as described in (b-c). (h) CD spectra of the total (dark gray), supernatant (light gray), and pellet (black) fractions of t_int samples upon centrifugation at 17,000 g for 30 min. (i) Raman spectra of the t_int pellet fraction (dark gray) and t_end (black) samples.

To assess the secondary structure of samples at t_int, CD and Raman spectroscopies were employed (Figure 5h–j). As shown earlier, soluble sRPT (t₀) was initially disordered (Figure 5h). However, at t_int a distinct CD spectrum with a minimum at ~205 nm was observed compared to that of the spectrum measured at t_end, which had characteristic minima at 209 and 218 nm, respectively (Figure 5h). To ascertain if the unique CD spectra of sRPT taken at t_int could be due to a mixture of fibrillar and soluble material, the sample was subjected to centrifugation at 17,000 g, and the supernatant and pellet fractions were independently measured (Figure 5i). We found that the pellet fraction had a similar CD spectrum to that taken at t_end, indicating it contained vprimarily fibrillar aggregates. The soluble material in the supernatant remained disordered, likely still monomeric (Figure 5i).

Raman spectroscopy was used to investigate any changes in the molecular vibrations of sRPT at t_int (Figure 5j). Compared to the spectrum taken at t_end (reproduced from Figure 2c), a mostly similar fingerprint region was observed for sRPT at t_int (Figure 5j). However, the ratio of the 1320 and 1348 cm⁻¹ peaks at t_int (0.9) was closer to the value reported for F_{pH 4} (Figure 2c). While this would indicate that the structure at t_int may be more similar to F_{pH 4}, the 1200–1300 cm⁻¹ region at t_int was different from that observed for F_{pH 4} (Figure 1c). Based on these results, we conclude that the initial increase in ³⁸¹W fluorescence at high sRPT concentrations at pH 5 does not represent an oligomeric intermediate, but rather an “immature” fibrillar state.

3.6. pH-dependent stability of sRPT fibrils.

To investigate the stability of sRPT fibrils, the pH of the buffer was altered (by dilution) and ³⁸¹W fluorescence kinetics were monitored over a period of 3 days (Figure 6). Dilution of F_{pH 4} into pH 4 buffer did not result in any significant kinetics change in ³⁸¹W fluorescence intensity (Figure 6a). However, dilution of F_{pH 4} into pH 5 and 6 buffer showed a steady decrease in ³⁸¹W fluorescence intensity. Collecting ³⁸¹W emission scans of FpH 4 at the end of 3 days revealed a concomitant blue-shift and intensity decrease in emission as the pH increased from 4 to 6 (Figure 6a inset), suggesting that the environment of ³⁸¹W for FpH 4 can slowly transition into a more F_{pH 5}- or F_{pH 6}-like state. We confirmed that this observation was not a result of fibril dissociation, as all samples still showed increased ThT fluorescence intensity (Figure S4), except for pH 7 in which disaggregation occurred as described earlier in Figure 3. No changes in F_{pH 5} or F_{pH 6} were observed upon dilution at pH 4, 5, or 6 (Figure 6b–c), indicating that these fibrils are stable to pH changes.

Fig. 6.

pH-dependent stability of sRPT fibrils. sRPT fibrils (30 µM) formed at pH 4 (a), 5 (b), or 6 (d) were diluted 6-fold into buffers at pH 4 (red), 5 (yellow), and 6 (green). Averaged data are shown (n ≥ 3). (Inset) λ_max values extracted from ³⁸¹W emission scans at the end of the reaction.

4. Discussion

4.1. sRPT aggregation and fibril maturation mirror melanosomal conditions.

Described in this work are the pH- and concentration-dependent aggregation of the Pmel17 sRPT domain, which are summarized schematically in Figure 7. At pH 4, sRPT rapidly formed immature fibrils (F_I) at concentrations as low as 3 µM. This highly-acidic environment mimics that of stages I–II melanosomes, where Pmel17 fibril formation is thought to initiate [4, 5]. A protein concentration-dependent process was observed at pH 5 for sRPT, where F_I were formed at high concentrations (≥30 µM) at a rate similar to that observed at pH 4. Aggregation at lower concentrations (≤15 µM) directly led to mature fibril (F_M) formation, albeit slower. At pH 6, F_M formation proceeded much slower than those observed at pH 4 and 5.

Fig. 7.

sRPT aggregation and fibril maturation mirror melanosomal conditions. S represents soluble sRPT at pH 5, 6, or 7. F_I and F_M represent immature and mature sRPT fibrils, respectively. At pH 4, sRPT rapidly forms F_I, which converts into F_M at pH 5 and 6. This process mimics melanosome maturation, which is at ~pH 4 in stages I–II, and reaches near neutral pH upon full maturation (stage IV). sRPT displays concentration-dependent behavior at pH 5, forming F_I at concentrations above 30 µM and F_M at concentrations below 15 µM. At pH 6, sRPT forms F_M. Both F_I and F_M dissolve into their soluble form at pH 7.

To mimic the maturation process of melanosomes, sRPT fibrils made at pH 4 were diluted into buffers of increasing pH. This revealed that F_I can mature into F_M, but diluting F_M (generated at pH 5 and 6) into more acidic buffers had no effect on the environment of ³⁸¹W. Collectively, a picture emerges where fibrils are rapidly formed in early stages at pH 4, which undergo conformational dynamics as melanosome maturation proceeds to pH 6. A similar mechanism has previously been proposed for lRPT [18]. While no obvious morphological differences in the fibrils formed could be discerned by TEM under the solution conditions examined here, reorganization was revealed in the C-terminus by ³⁸¹W fluorescence, which was corroborated by the differences in C–H deformation frequencies in the Raman spectra. In the context of melanogenesis, it remains to be investigated if similar aggregation mechanisms would be observed in the lipid-rich environment of the melanosome, or if melanin would bind to one fibril polymorph (F_I or F_M) more strongly than the other. However, it is noteworthy that F_I formed at pH 4 has the highest ThT response which we inferred as having a stronger affinity for ThT, which Fowler et al. have suggested to resemble the molecular features of a melanin precursor [3].

4.2. The importance of the C-terminus in sRPT aggregation/disaggregation.

McGlinchey et al. have previously shown that sRPT fibrils dissolve upon exposure to neutral pH [16], which may be an evolutionary mechanism designed to prevent toxicity should Pmel17 fibrils escape the melanosome and enter the cytosol [17]. Here, ³⁸¹W was used to monitor the disaggregation process in real-time. This revealed that F_M immediately disassembled with no observable lag time, while a short 7-s lag time was observed for F_I at pH 4, indicating that dissolution is more difficult for F_I compared to F_M. The reverse was observed for the formation of the fibrils, where F_I is faster than F_M.

The pH-dependent aggregation/disaggregation process of sRPT further substantiates the importance of protonation/deprotonation events, which has been described in detail for lRPT [15]. Specifically, the residue ⁴²²E (³⁸⁰E in sRPT) was the primary (de)protonation site driving aggregation/disaggregation of lRPT. While we do not directly investigate this for sRPT, it is noteworthy that ³⁸¹W lies directly on the C-terminal side of this residue. Given the similar spectral changes in ³⁸¹W observed during aggregation reported here, it is likely that sRPT aggregation proceeds in a similar fashion, where the protonation of ³⁸⁰E facilitates sRPT fibril formation at acidic pH. However, because we observe a protein concentration-dependent aggregation kinetics for sRPT at pH 5, the mechanism may be more nuanced than lRPT and further investigation is needed.

4.3. Biological implications for sRPT in melanogenesis.

The conserved pH-dependent aggregation/disaggregation observed for both sRPT and lRPT, along with work from Nichols et al. showing the two isoforms can be expressed independently in MNT-1 cells [21], leads us to speculate an interplay between the two isoforms in vivo. In bacterial functional amyloid systems, specific genes within the operon serve as nucleators of amyloid formation. In E. coli, the CsgB protein rapidly forms a fibrillar template upon which CsgA binds and elongates to form the biofilm matrix [33, 34]. Likewise, a similar role for FapB has been put forward for amyloid formation in P. aeruginosa [35, 36]. Because sRPT rapidly forms fibrils at pH 4, it is reasonable to propose that it could serve as a nucleator of lRPT (and therefore, Pmel17) amyloid formation within the melanosome. Future work testing this hypothesis could reveal an evolutionary-conserved mechanism of functional amyloid formation from bacteria to humans.

Abbreviations:

Pmel17

pre-melanosomal protein

sRPT

short repeat domain

F_I

immature fibril

F_M

mature fibril

ABOUT THE AUTHORS

Dr. Dexter Dean joined the Lee lab as a postdoctoral fellow after completing his Ph.D. in Biochemistry at The University of Southern Mississippi in 2018. Prior to his Ph.D., he received a B.A. in Biochemistry from Huntingdon College in 2013. He is interested in using biophysical tools to understand mechanisms of amyloid formation as it relates to both function and disease.

Dr. Jennifer C. Lee is a Senior Investigator at the National Heart, Lung, and Blood Institute, National Institutes of Health. Her research efforts are dedicated towards the elucidation of mechanisms of amyloid formation through biophysical and biochemical approaches. She is interested in understanding the chemical nature of intermolecular interactions that drive amyloid formation in a cellular environment. Prior to joining the NIH, she was a Beckman Senior Research Fellow at the Beckman Institute Laser Resource Center at the California Institute of Technology. Jennifer obtained a Ph.D. in Chemistry from Caltech in 2002 working in the laboratory of Harry Gray and has undergraduate degrees in Chemistry and Economics from the University of California at Berkeley.