The intrinsically disordered region of the human parathyroid hormone controls functional amyloid properties

Abstract

Human parathyroid hormone (PTH), which regulates the calcium and phosphate levels in blood, forms functional amyloid fibrils, which are believed to be the storage form during secretion from parathyroid glands. We investigated by various biophysical approaches the role of the intrinsically disordered C-terminal section 35 to 84 of the peptide hormone, which is conserved in mammals but not required for activation of the target G protein–coupled PTH receptors. The flanking intrinsically disordered region (IDR) of PTH, which is not part of the cross-β fibrillar core structure, impairs the lateral filament order and reduces the thermodynamic stability to an extent that monomer release from the fibrils reaches a physiological level. Concentration-dependent fibrillation kinetics revealed that the IDR increases the critical fibrillation and oligomer concentration and screens secondary nucleation. We propose that the IDRs of PTH shape the generic property of the hormone to form amyloid fibrils toward the physiological requirement of monomer release after cellular storage.

The human parathyroid hormone (PTH) secreted by the parathyroid glands plays a pivotal role in regulating the serum phosphate and calcium levels in the blood by stimulating cellular receptors in bones, kidneys, and intestines (1, 2). Modulation of the calcium level in the body is mediated via calcium receptors that sense the level of calcium in blood, and these receptors are predominantly expressed in the parathyroid gland and kidneys (3, 4). The 84 residues comprising PTH activate G protein–coupled receptors (GPCRs) in target cells by first binding to the extracellular domains of these class B GPCRs and then activating the transmembrane part (5, 6). For this function, the α-helical N-terminal residues 1 to 34 (PTH₃₄) are sufficient (Fig. 1A). Therefore, both PTH and PTH₃₄ are Food and Drug Administration–approved drugs against osteoporosis (7, 8). Before release into the bloodstream, PTH gets stored in secretory granules, and it has been proposed that upon storage, amyloids are formed (9). Such functional amyloid structures have been observed to serve as a storage form for many peptide hormones, including pituitary hormones (10, 11, 12). The increasing number of identified functional amyloids highlights their diverse roles in various biological processes, such as structural support, information carrier functions, functional suppression, and signaling (13).

Figure 1.

Secondary structures of PTH and PTH₃₄. A, primary structure, NMR structure of PTH₃₄ (PDB code: 1ZWA, left), and a RaptorX model of full-length PTH (right). B, CD spectra of 20 μM PTH (black), PTH₃₄ (blue), and PTH_35–84 (dotted blue) in solution before fibrillation has started. The sum of the PTH₃₄ and PTH_35–84 resembling full-length PTH is depicted in gray. C, CD spectra of amyloid fibrils of PTH (black) and PTH₃₄ (blue). D, ATR–FTIR spectroscopy analysis of PTH (black) and PTH₃₄(blue) in solution before fibrillation has started and (E) of fibrillated PTH (black) and fibrillated PTH₃₄ (blue). D and E, the second derivative of the ATR–FTIR spectra is depicted with the typical wave numbers for secondary structures as indicated in gray. The corresponding spectra before mathematical treatment are shown in Figure S2. All samples were measured in 50 mM sodium phosphate buffer, pH 7.4, at room temperature. ATR, attenuated total reflectance; PTH, parathyroid hormone.

In vitro, PTH exhibits strong amylogenic properties and forms curvilinear amyloid (9). In many well-studied amyloids, the core of the cross-β structure is formed by almost the entire protein, including the Alzheimer's β peptide and the islet amyloid polypeptide (14, 15, 16). In other cases, such as α-synuclein and the tau protein, the fibrillar core is formed only by a segment of the protein flanked by N- and/or C-terminal extensions (16). The latter was also found for PTH, where a small central region (25R-37L) was identified as the core region of the fibrils by protease digestion of the flanking residues, which is involved in monomer–monomer interaction and formation of amyloid assemblies (9). Such flanking regions can be of globular and/or disordered nature and might influence amyloidogenesis by either accelerating or decelerating fibril formation (17, 18, 19). In the case of acceleration, the effect was attributed to intermolecular interactions toward the additional disordered region. This effect is known as the "fly-casting" mechanism, wherein the large flexible disordered surface of the amyloid fibril effectively captures the flanking disordered regions of the monomer (20). In other cases, the presence of disordered regions hinders the process of amyloid fibrillation because of their presumed high flexibility, resulting in an elevation of the entropic barrier (21) or steric impedance (22).

During PTH fibrillation, two nucleation mechanisms compete (23). Primary nucleation generates new elongating fibrils in the lag phase of fibrillation, whereas secondary nucleation forms nuclei from monomers catalyzed by the surface of existing fibrils. Concentration-dependent thioflavin T (ThT) fibrillation kinetics combined with NMR, small-angle X-ray scattering (SAXS), and native mass spectrometry revealed further details, including various productive oligomeric states with different molecular weights (23, 24). However, the complex behavior displayed a clear concentration dependence: at PTH concentrations below the critical oligomer concentration, secondary nucleation emerged as the dominant process, whereas at higher PTH concentrations, elongation and secondary nucleation were inhibited. Mature PTH fibrils are cross-β structured according to CD and IR spectroscopy, SAXS and wide-angle X-ray scattering (WAXS), as well as transmission electron microscopy (TEM) (9, 12, 23).

In solution, ¹³C NMR chemical shifts of the C_α and C_β carbons revealed for residues H14–N16 and S17–Q29 of PTH transient α-helical conformations and for D30–Q84, an intrinsically disordered character (25). The α-helical content is also documented by the NMR structure of PTH₃₄ (26) depicted in Figure 1A, left. For the graphical representation of PTH in Figure 1A, right, the predicted secondary structures by the program RaptorX have been used. In the fibrillar state, PTH was protease resistant only for the core section 25R-37L, indicating that both the flanking N- and C-terminal sections are intrinsically disordered (9). PTH fibrils are fundamentally different from many neurodegenerative fibrils in their property to dissolve upon dilution because of their low thermodynamic stability (9, 25). For functional fibrils and as a storage form of the PTH hormone, this property is required for monomer release into the bloodstream for secreted PTH to reach the target receptors in bone and kidney tissue (4).

In order to examine how the C-terminal intrinsically disordered region (IDR) of PTH influences the fibrillation kinetics and fibril properties, we compare in this report structural properties, monomer-fibril equilibria, and fibrillation kinetics of PTH₃₄, which lacks the C-terminal IDR, with full-length PTH. We employed IR spectroscopy, WAXS, and TEM imaging for fibril characterizations and ThT fluorescence and monomer release assays to study the molecular mechanism of fibrillation and the thermodynamic stability of the mature fibrils. We found that the C-terminal IDR increases the critical concentration of fibrillation and decreases fibril stability, shifts the predominance from primary to secondary nucleation at higher monomer concentrations, and changes the fibril morphology and filament compactness. We discuss these findings in a physiological context of hormone monomer release after a possible secretory storage of PTH as functional amyloids.

Results

Structural differences between soluble and fibrillated PTH and PTH₃₄

The two functional regions of PTH responsible for its interactions with G protein–coupled PTH receptors 1 and 2 comprise α-helical conformations (25) (Fig. 1A). While section N16–F34 binds to the extracellular domain of the receptor, S1–H14 binds to its transmembrane domain (5, 6). Hence, the CD spectra of PTH at pH 7.4 (Fig. 1B, black) shows two distinct structural components: an α-helical contribution (evidenced by a negative ellipticity at 222 nm) and a random coil contribution (indicated by a minimum at 201 nm), as reported earlier (24, 25). The CD spectrum of the N-terminally truncated variant PTH_35–84 shows the characteristic negative ellipticity of a random coil at 200 nm (Fig. 1B, blue dotted line). In contrast, the C-terminally truncated variant PTH₃₄ (Fig. 1B, blue) shows a lower amount of random coil and a shift of the minimum to 204 nm, plus the typical signature of an α-helical contribution with minimum ellipticity at 222 nm, along with positive values below 195 nm. The sum of the CD spectra of PTH₃₄ and PTH_35–84 (Fig. 1B, gray) nicely corresponds to the spectrum of full-length PTH. These experimentally recorded CD spectra of soluble PTH and its truncated variant PTH₃₄ are consistent with earlier reported NMR analyses of PTH (25) and the solution structure of PTH₃₄ (1zwa.pdb). These NMR analyses of PTH in solution have shown α-helical propensity for residues 1 to 34 and an intrinsically disordered segment for residues 35 to 84 (24, 25), which has been confirmed now by including PTH_35–84 in the CD analyses.

During amyloid fibril formation, PTH undergoes a conformational coil/helix-to-cross-β sheet conversion, which can be followed by CD and IR spectroscopy (9, 27). These conformational changes in the secondary structure toward fibrillated PTH are documented by the CD spectra displaying a minimum ellipticity at wavelengths 210 to 215 nm and a maximum ellipticity around 195 nm (Fig.1C, black), which is characteristic of a β-sheet structure. The CD spectrum of fibrillated PTH₃₄ also resembles a β-sheet structure. We suspect that the differences between the CD spectra of PTH and PTH₃₄ originate from residues 35 to 84 of PTH, which flank the core cross-β-sheet structure of PTH (9) and are expected to be unstructured (28).

Next, we compared the soluble and fibrillar forms of PTH and its C-terminally truncated variant PTH₃₄ by infrared spectroscopy. In attenuated total reflectance (ATR)–FTIR spectroscopy, the amide I spectral region (1600–1700 cm⁻¹) provides information about the stretching mode vibrations of the backbone carbonyl groups and thus is sensitive to the secondary structure of the polypeptide chain (29, 30). In hydrated films, typically α-helical conformations absorb at 1656 cm^-1, random coils at 1641 cm^-1, cross-β-sheets range between 1624 and 1633 cm^-1, and β turns between 1664 and 1678 cm^-1 (31). As IR bands are usually broad and overlapping, Figure 1, D and E displays the second derivatives of derived IR spectra in order to determine the dominant vibrational modes of PTH and PTH₃₄ before and after fibrillation.

Before fibrillation had started, a strong absorption of PTH₃₄ was found at 1654 cm⁻¹ (blue line in Fig. 1D), resulting from its α-helical content. The additional absorption at 1638 cm^-1 for PTH (black line in Fig. 1D) reports about the contribution from the disordered region toward the C terminus (9). The spectra recorded for fibrillar solutions of PTH₃₄ (Fig. 1E, blue) show absorption maxima at 1629 cm^-1 and 1665 cm^-1, indicative of β-sheet and β-turn conformations, respectively. In full-length PTH, the shift from 1629 cm^-1 to 1617 cm^-1 indicates an increase in rigidity and size (27), whereas the additional absorption at 1696 cm^-1 is indicative of an antiparallel β-sheet content (32) in PTH (Fig. 1E, black). The additional absorption band at 1647 cm^-1 is attributed to intrinsically disordered residues toward the C terminus flanking the fibrillar cross-β core structure.

The disordered C-terminal region of PTH affects the fibril structure

Next, we asked the question: To what extent does the disordered C-terminal region of PTH modify the structural organization of the corresponding amyloid fibrils? Morphologically, fibrils formed by PTH₃₄ appear straight and more compact (Figs. 2A and S3A) compared with those of PTH, which have the archetypical thin curvilinear appearance (Figs. 2B and S3B) (9, 23). The latter was found in previous studies to be mainly composed of three filamentous structures, each of 5.5 nm diameter (23). Fibrils of PTH₃₄ depicted in Figure 2B, at the same resolution as those of PTH, showed larger diameters, indicating that more filaments align within the straight morphology.

Figure 2.

Impact of the C-terminal IDR region on PTH fibril morphology. Electron micrographs of (A) PTH₃₄ and (B) PTH fibrils. The scale bar in the TEM images represents 100 nm. C, WAXS spectra of PTH (black) and PTH₃₄ (blue). Inset, 2D image of PTH₃₄ with the meridional and equatorial peak reflections at 10.8 Å and 4.75 Å, consistent with a cross-β structure. IDR, intrinsically disordered region; PTH, parathyroid hormone; TEM, transmission electron microscopy; WAXS, wide-angle X-ray scattering.

WAXS enables the characterization of the internal structural organization and dimensions of amyloid fibrils, including their packing density and architecture. The hallmarks of a cross-β structure are meridional reflections (4.7–4.8 Å) and equatorial reflections (10 Å) in oriented fibril samples (33). Meridional reflections (outer ring) in WAXS of amyloid fibrils originate from a structural pattern along the direction parallel to the long axis of the fibrils and indicate the arrangement of the β-strands. Equatorial reflections indicate a pattern perpendicular to the fibril axis and give information about the spacing between stacked β-sheets perpendicular to the fibril axis.

In Figure 2C, the WAXS spectra of PTH and PTH₃₄ are compared. The meridional reflections at 1.35 Å^-1 scattering vector q, indicating the alignment of β-strands, are very similar for fibrils of both PTH variants. However, the equatorial reflections (q between 0.5 Å^-1 and 0.6 Å^-1) appeared broader and smaller for PTH as compared with PTH₃₄, which could indicate a higher lateral order within the fibrillar bundles of PTH₃₄, also seen in the EM images (Fig. 2A). We attribute this observed difference to the presence of the C-terminally flanking IDR in PTH fibrils.

Critical fibrillation concentration of PTH and PTH₃₄

After these structural elucidations, the influence of the C-terminal IDR of PTH on the thermodynamic stability of the fibrillar state was investigated. The free monomer concentration at equilibrium with amyloid fibrils, also known as the critical concentration c_crit (23, 24) for PTH₃₄ and PTH, was determined by UV absorption at 280 nm after spinning down the fibrillar moiety after 24 h of incubation. The corresponding fractions of released monomers are depicted in Figure 3A, resulting in a c_crit of 8.83 ± 1.5 μM for PTH₃₄ and 70.3 ± 9.5 μM for PTH. c_crit can be related to the Gibbs free energy of fibril formation (ΔG⁰), using Equation 3 (see the Experimental procedures section). PTH₃₄ fibrils revealed a value of ΔG⁰ = −28.9 ± 1.6 kJ/mol and hence a higher thermodynamic stability compared with PTH fibrils with ΔG⁰ = −23.7 ± 1.6 kJ/mol. This indicates that the flanking C-terminal IDRs decorating the PTH cross-β core are destabilizing the fibrils.

Figure 3.

Thermodynamic stability of PTH and PTH₃₄ fibrils. A, monomer release from fibrils of PTH (black) and PTH₃₄ (blue) after an overnight incubation in 50 mM sodium phosphate, 150 mM sodium chloride, pH 7.4. The fraction of monomer corresponds to the ratio of released monomers and the amount of PTH in the initial fibrils. B, the same monomer release as in (A) but in the presence of urea. The error bars correspond to the standard deviation of three independent replicates. C, time-dependent monomer release of PTH from fibrils. The amount of monomeric PTH was quantified by consecutive 1D ¹H NMR spectra recorded over 35 h after dilution of PTH fibrils. Each 1D ¹H NMR spectrum was integrated between 1.0 ppm and 2.4 ppm. The green line corresponds to a single exponential fit to these NMR integrals, resulting in a build-up rate constant of 0.46 ± 0.01 h^-1. PTH, parathyroid hormone.

To further explore fibril stability, we added urea to the fibrils and monitored the concentration of monomers when equilibrium was reached. For PTH₃₄ fibrils (blue symbols in Fig. 3B), increasing urea concentrations up to 5 M did not lead to a significant rise in monomer levels, indicating high fibril stability. In contrast, fibrils formed from full-length PTH were markedly less stable. Already at 1 M urea, the monomer fraction exceeded 0.9 (gray symbols in Fig. 3B), demonstrating that these fibrils are much less stable under denaturing conditions.

To monitor the kinetics of monomer release, we employed 1D ¹H NMR spectroscopy in solution because this allows us to solely detect the monomer moiety, as PTH in the fibrillar state broadens the NMR resonances beyond detection (Fig. 3C). We observed an exponential increase of monomeric PTH with a rate constant of 0.46 ± 0.01 h^-1.

Fibrillation kinetics of PTH and PTH₃₄

To study the influence of the C-terminal IDR of PTH on the fibrillation mechanism, ThT fibrillation assays were set up for concentrations ranging from 100 to 600 μM for PTH (Fig. 4A) and 75 μM to 400 μM for PTH₃₄ (Fig. 4B) under identical buffer and shaking conditions on 96-well plates. The concentration ranges were chosen according to the critical concentration of fibrillation of PTH and PTH₃₄. The fibrillation kinetics of PTH with a two-step transition at low protein concentrations and a decreasing lag time with increasing concentrations reproduced earlier reported data (23).

Figure 4.

Amyloid fibril formation of PTH and PTH₃₄. ThT fluorescence monitored fibrillation kinetics of (A) PTH at concentrations between 100 μM and 600 μM in steps of 50 μM or 100 μM and (B) of PTH₃₄ at concentrations between 75 μM and 400 μM in steps of 25 μM or 50 μM. Fibrillation was carried out in 50 mM sodium phosphate buffer, pH 7.4, supplemented with 150 mM NaCl, at 37 °C. PTH, parathyroid hormone; ThT, thioflavin T.

The lag times observed for PTH₃₄ ranged between 5 h and 40 h (Fig. 4B) and were significantly shorter than those for PTH (between 40 h and 80 h, Fig. 4A). We attribute this difference to the lack of the C-terminal IDR of PTH₃₄. To further test this assumption, we performed cross-seeding experiments. Here, quiescent conditions were used to prevent fibril formation of the PTH and PTH₃₄ monomers (gray ThT fluorescence in Fig. S4). When adding PTH₃₄ seeds to PTH monomers, an immediate ThT fluorescence increase was observed, which grew exponentially for 20 h (Fig. S4A, orange). Vice versa, after adding PTH seeds to PTH₃₄ monomers, the ThT fluorescence increased sigmoidally after a long lag time of 60 h to 80 h (Fig. S4B, green). In principle, the addition of seeds fosters secondary nucleation. This is strongly retarded in the presence of the C-terminal IDR of PTH, indicating again that the flanking IDR decorating the fibrillar seeds delays fibril formation. To impair PTH fibrillation, this IDR has to be part of the hormone, because the corresponding PTH_35–84 peptide did not change the ThT fluorescence kinetics of PTH₃₄ significantly (Fig. S6).

Also, as reported earlier for PTH (23), the ThT kinetics at increasing concentrations of PTH₃₄ showed decreasing lag times and a more pronounced slow exponential decay after reaching the plateau. The former arises from the concentration-dependent nucleation processes (see below for a quantitative analysis toward rate constants), and the latter arises from a reduced fluorescence quantum yield of bound ThT during slow fibril clustering (23). All ThT kinetics of PTH₃₄ showed much higher ThT fluorescence intensities compared with PTH, which can be seen from the raw data in Figure S5. This observation can be addressed to the flanking IDRs, which decorate the fibrils and impair ThT binding. We had similar findings for PTH extended at the N terminus or C terminus by thermoresponsive polymers, where we found fibrils in TEM micrographs but no ThT binding (34).

To gain more information about the mechanism underlying the formation of PTH fibrils, individual curve fittings were performed on the ThT-monitored fibrillation kinetics of both PTH and PTH₃₄ using Equation 1. This approach will give insights into the contributions of the C-terminal IDR of PTH on effective rates of both the sum of primary (λ) and secondary (κ) fibrillation processes. For PTH, we could reproduce earlier findings (23) that the prefactor A, which describes the nucleation-to-growth ratio, showed at all studied concentrations values far below 0.33, indicating that secondary nucleation dominates the fibrillation mechanism (Fig. 5, C and G, left). In addition, the effective rates λ and κ and thus the correlated prefactor A did not change significantly at concentrations up to 400 μM (Fig. 5, A–C). At concentrations between 400 μM and 600 μM, λ strongly increased by one order of magnitude (Fig. 5A), whereas κ (Fig. 5B) slightly decreased, leading to an overall enhanced A value. These findings were previously explained by concentration-dependent processes that give, on the one hand, access to new pathways of primary nucleation (i.e., via oligomerization or phase separation), whereas the autocatalytic effects of secondary nucleation are reduced (i.e., reduced accessibility to fibrillar interaction surfaces by transiently bound oligomers (18)). Therefore, this 400 μM concentration refers to the critical oligomer concentration of PTH, as discussed earlier (23).

Figure 5.

Concentration-dependent primary and secondary nucleation processes. The fittings to the individual ThT kinetics shown in A and B have been performed using (Equation 1). A, B, and C, respectively, represent primary processes λ, secondary processes κ, and the kinetic prefactor A (Equation 2) for PTH. D–F, respectively, represent primary processes λ, secondary processes κ, and the kinetic prefactor A for PTH₃₄. All data points represent the mean and the standard deviation of three individual replicates. G, schematic representation of PTH fibril formation, in which secondary processes dominate over primary processes, but the former get screened by the IDR not present in PTH₃₄ (gray arrows). IDR, intrinsically disordered region; PTH, parathyroid hormone; ThT, thioflavin T.

To investigate the impact of the C-terminal IDR toward the fibrillation mechanism of PTH, the same kinetic analyses were applied to ThT-monitored fibrillation kinetics of PTH₃₄. Here, the rate describing the primary nucleation processes, λ, continuously increases with peptide concentration for all the investigated concentrations (Fig. 5D), whereas the secondary processes are slowed down, indicated by κ (Fig. 5E). As the reduction of κ is less pronounced compared with the increase of λ, this results in a positive slope of prefactor A (Fig. 5F). This finding is comparable to full-length PTH above the critical oligomer concentration of 400 μM. It suggests that the critical oligomer concentration of PTH₃₄ is reduced (Fig. 5G, right) to a value below 75 μM, which is the lowest concentration studied here to give reproducible ThT kinetics.

Discussion

The family of fibril-forming proteins is not restricted to well-characterized short pathological polypeptides, such as the Alzheimer β-peptide or β-endorphin, for which the majority of residues are part of the cross-β structure. Longer primary sequences lead to flanking N- and/or C-terminal sections, which are not part of the fibril core but decorate it at the corresponding side (16). This generic property has been found for both amylogenic proteins related to neurodegenerative diseases and for the here-studied functional amyloids. The PTH is a well-suited system to elucidate the role of such flanking sections because of its inherited bifunctional property of residues 1 to 34, which are sufficient to activate the corresponding GPCRs; residues 25 to 37 forming the core of functional amyloids during hormone secretion, and the intrinsically disordered flanking residues 35 to 84.

The comparison between concentration-dependent fibrillation kinetics of PTH and PTH₃₄ revealed that the C-terminal IDR increases the critical concentration of fibrillation from 9 μM to 70 μM, as well as the critical oligomer concentration from a value below 75 μM to 400 μM. This IDR retards nucleation processes because it decorates the cross-β core structure as a disordered flanking peptide chain and thus most likely screens secondary nucleation. Furthermore, the IDR of PTH impairs the lateral filament order of the fibrils, which probably reduces their thermodynamic fibril stability by 5.2 kJ/mol. We expect that the main entropic barrier and steric hindrance of PTH fibril assembly arise from the IDR section because the fibril core–forming residues already have some secondary structure content in the monomers. Similar findings have been reported for fibrils of C-terminally IDR-truncated barnase M1 peptides, which exhibited a pronounced thermodynamic stability unlike the wildtype peptide in the presence of denaturing agents (28).

The impact of the PTH IDR on individual contributions of primary and secondary processes at mechanistic detail is possible by comparing the rates of primary and secondary nucleation λ and κ and the pre-exponential factor A = λ³/(3κ³). A can also be taken as a measure of the minimum fibrillar mass required to initiate secondary processes (35). At all studied concentrations, the κ rates of PTH were below those of PTH₃₄, indicating that the IDR retards secondary nucleation. Still, the value of A is very small for both PTH and PTH₃₄, confirming that secondary processes dominate the overall fibrillation process (23). The presence of the IDR in PTH ensures basal primary nucleation regardless of concentration, with increasing competition only at high peptide concentrations. The hormone concentration in secretory granules, for example, of prolactin was found to be 200 times higher compared with the endoplasmic reticulum (36) and ranges generally between 100 and 150 mg/ml (37). We expect similar conditions for PTH in the corresponding secretory granules and consider, therefore, the here-studied concentration range as physiologically relevant.

It is important to note that the effect of IDR truncation on fibrillation of a given protein can be complex and context dependent. In proteins, such as α-synuclein, the C-terminal disordered region serves as a protective barrier for the aggregation-prone non-amyloid-β component region; hence, truncation of the C terminus leads to enhanced fibrillation (38). Similarly, the tau protein is a fully intrinsically disordered protein. The core region of tau fibrils comprises residues 297 to 391 decorated by N- and C-terminally disordered sections (39). Truncation of residues 150 to 230, or the last 50 amino acids, promoted its aggregation and hence its pathological activities (40, 41). In contrast to this, in certain neurodegenerative instances, regions with structural disorder adjacent to amyloid-prone sequences may actually promote fibrillation by facilitating interactions between the end of the existing amyloid structures and elongating monomers (42, 43). The here-presented comparisons of PTH and truncated PTH₃₄ extend the first group of proteins by this hormone.

As many other hormones, PTH belongs to the family of functional amyloid-forming peptides (10). Amyloids formed by this family have evolved to disassemble according to environmental changes (13). This behavior also applies to PTH. The C-terminal IDR reduces fibril stability and therefore puts monomer release from PTH fibrils at a rate constant of 0.46 ± 0.01 h^-1 (Fig. 3C) to the physiologically expected time window. PTH fibril formation also strongly depends on the pH value. At pH 9.0, close to its isoelectric point, amyloid fibrils form within a few minutes (9), whereas at pH 5.5, which corresponds to the value in the secretory granules, high salt or heparin is required to form fibrils at all (12, 44). The same applies to the N-terminal IDR extended pro form of PTH, which prevents premature fibril formation by Coulomb repulsion of four positively charged lysine residues before cleavage by the convertase furin in the Golgi apparatus (25). The N- and C-terminal flanking IDRs, which modulate PTH fibril stability, are evolutionarily conserved (Fig. S1), further underscoring their physiological role to maintain PTH function not only restricted to residues 1 to 34, which are sufficient to activate the target receptor. The neuropeptide hormone β-endorphin, which forms amyloids in secretory granules, shows very similar properties to PTH (45). Fibrils only form at pH 5.5 in the presence of high salt or glycosaminoglycans, which dissolve in blood (pH of 7.4 and lower salt as well as glycosaminoglycan concentration) with a rate constant of about 1 h^-1. β-endorphin comprises 31 residues, which all end up in the β-solenoid conformation, forming the cross-β amyloid structure (46) without flanking IDR sections. Their morphologies (45, 46) resemble closely the appearance of the here-studied PTH₃₄ fibrils. Monomer release of β-endorphin is mainly governed by deprotonation of glutamate 8 upon shift from pH 5.5 for storage to pH 7.4 in the blood (47).

In conclusion, the presented findings highlight the crucial function of the C-terminal IDR in PTH. For PTH fibrils to serve as functional amyloids and a hormone reservoir, the presence of a disordered region is important, as this allows fibrils to reversibly release monomeric PTH into the blood. These properties are not given for the short PTH₃₄ peptide, although it is sufficient to activate the cognate PTH receptor and thus is a Food and Drug Administration–approved drug (teriparatide to treat osteoporosis). Compared with PTH, fibrils of PTH₃₄ form faster and are thermodynamically more stable with a morphologically higher lateral order. Together with the lower critical oligomer and fibrillation concentrations, monomer release from PTH₃₄ fibrils is retarded by at least a factor of 10. We suspect that because of these properties of the IDR of PTH, for which, to the best of our knowledge, a clear function has not yet been reported, this C-terminal extension is conserved in mammals. It shapes the amyloid fibril-forming and monomer-releasing property of PTH toward the physiological requirements. In addition, we suspect that PTH₃₄ fibrils might serve as an osteoporosis drug with a prolonged pharmacokinetic and therefore as a long-term depot, which are future aspects to be investigated.

Experimental procedures

Protein purification

Human PTH protein was cloned and purified according to previous protocols with some modifications (48). In brief, pET SUMO adapt vectors containing PTH and an N-terminal SUMO His-tag were transformed into Escherichia coli BL21 (DE3) Codon Plus RIL cells. The purification process of the SUMO fusion proteins involved initial purification using nickel–nitrilotriacetic acid column chromatography, followed by SUMO cleavage using a specific SUMO protease at a 1/100 ratio. Subsequently, hydrophobic interaction chromatography (HiTrap Butyl FF column) was employed to separate SUMO from the target protein. PTH then underwent purification through S-75 gel filtration chromatography. In contrast to an earlier protocol (9), a final step of cation exchange chromatography (using a sulphopropyl resin) was added to the purification step (23), followed by subsequent buffer exchange to 50 mM sodium phosphate (pH 7.4) using 3.5 kDa dialysis bags. Lyophilized recombinant human PTH₃₄ was obtained from Sandoz and dissolved in the same buffer used for PTH. It was then subjected to an S-30 gel filtration purification step. Subsequently, it was concentrated using a cation exchange chromatography column, similar to the process for PTH, and underwent buffer exchange to 50 mM sodium phosphate (pH 7.4) using 1.0 kDa dialysis bags. PTH_35–84 was synthesized by the Core Peptide Unit at the University of Leipzig.

Wide-angle X-ray scattering

WAXS experiments were performed in transmission mode using a SAXSLAB laboratory setup (Retro-F) equipped with an AXO microfocus X-ray source. The AXO multilayer X-ray optic (AXO Dresden GmbH) was used as a monochromator for Cu–Kα radiation (λ = 0.154 nm). A two-dimensional detector (PILATUS3 R 300K; DECTRIS) was used to record the 2D scattering patterns, which were subsequently reduced to 1D scattering curves by angular averaging. The scattering q range was between 0.2 Å^-1 and 3 Å^-1. For WAXS experiments, the fibril suspension was ultracentrifugated (200,000 g, 10 min), and the thus obtained pellet was transferred into a ring-shaped aluminum holder (2 mm thick and with a central hole of 1.5 mm diameter) and left to dry overnight. The scattering measurements were performed at room temperature in a vacuum.

ATR–FTIR spectroscopy

ATR–FTIR spectra were recorded with an ALPHA 2 Platinum-ATR FTIR spectrometer. The sample (2 μl) was applied on the diamond crystal and dried for up to 30 min at 37 °C to decrease background contributions arising from water. The increased protein concentration because of drying did not affect peak positions or relative amplitudes, as verified by comparative measurements predrying and postdrying. The final spectrum was determined by 32 scans, measured over the range of 400 to 4000 cm^-1 at a spectral resolution of 2 cm^-1. The spectral data of the amide I band were baseline corrected and normalized with respect to the amide II absorbance band. Second derivative analyses were applied to improve the separation of overlapping bands. The resulting peaks were assigned to secondary structure elements based on references given for hydrated films (31).

ThT kinetic assay

ThT kinetic assays were carried out on a FLUOstar Omega (BMG Labtech GmbH) reader using Greiner 96 F-bottom well plates (Greiner Bio-One GmbH). ThT fluorescence was monitored at 480 nm after excitation at 450 nm. The experiments were performed in our standard fibrillation buffer, 50 mM sodium phosphate, 150 mM NaCl, pH 7.4, and 37 °C. The samples were shaken at 300 rpm for 150 s prior to each excitation, and every 1200 s, a measurement point was recorded. The final sample volume in each well of the plate was set to 150 μl, and measurements were performed in triplicate.

The overall fibrillation process can be divided into two steps (49). De novo generation of structured nuclei is called primary nucleation, whereas secondary nucleation involves the formation of nuclei on the surface of already existing fibrils. The time-dependent increase of the ThT fluorescence, F(t), was analyzed using Equation 1 (12, 20, 21, 23).

$$F\left(t\right)=\left(1-\frac{1}{\frac{{\lambda }^3}{3{\kappa }^3}\left(e^{\kappa t}-1\right)+1}\right)\Delta F_{\text{plateau}}$$ (1)

where ΔF_plateau is the plateau fluorescence, and λ and κ denote primary and secondary processes of nucleation and growth, respectively, allowing to retrieve the individual contributions of these two processes. The ratio λ³/(3κ³), later referred to as prefactor A, describes the nucleation-to-growth ratio (50). While A > 0.5 describes the hyperbolic aggregation kinetics of seeded fibrillation, smaller values of A are indicative of nonseeded aggregation and sigmoidal fibrillation curves. Moreover, in the case of A < 0.33, secondary processes dominate.

Monomer release in the presence of buffer or urea

The monomer release from PTH and PTH₃₄ fibrils was performed in accordance with the published protocol for PTH fibrils (9, 25). After the completion of the fibrillation process in 400 μM samples, centrifugation was performed at 16,100g for 60 min to separate the amyloid fibrils (in the pellet) from monomeric protein (supernatant). The fibrils (in pellet) were then diluted and mixed with buffer (50 mM sodium phosphate, 150 mM sodium chloride, pH 7.4) or buffer supplemented with urea (1 M, 3 M, or 5 M). The amount of monomer released after 24 h was calculated from the absorbance at 280 nm by UV spectroscopy using a molar extinction coefficient of ε₂₈₀ = 5500 M^-1 cm^-1 for PTH and PTH₃₄. Monomer-release kinetics of PTH were followed by 1D ¹H NMR spectroscopy under NMR-compatible conditions (10 mM sodium acetate [pH 5.3] and 300 mM sodium sulfate at 25 °C) 15 min after fibril pellet dilution in the NMR tube. 1D ¹H NMR spectra of released PTH monomers were recorded for 35 h. Each 1D spectrum was integrated between 1.0 ppm and 2.4 ppm, leading to one time point of the kinetics. These NMR integrals were normalized to 0.0 at time 0 and 1.0 for the largest integral. A single exponential fit to the buildup of the NMR integral with time resulted in the rate constant of monomer release because only PTH in the monomeric but not the fibrillar state contributes to the 1D ¹H NMR spectrum in solution.

Critical concentration determination

Fibrillated samples that had reached a plateau value in the ThT fluorescence assay, and hence in thermodynamic equilibrium, were centrifuged at 16,100g for a duration of 1 h at room temperature. Critical concentration is the concentration of free peptide monomer in the supernatant fraction, which can be measured using UV–visible spectroscopy at 280 nm (ε₂₈₀ = 5500 M^-1 cm^-1), whereas the concentration of free ThT was measured at 412 nm (ε₄₁₂ = 36,000 M^-1 cm^-1) (51). ThT contributes to the UV absorption at 280 nm; therefore, a correction factor was included (Equation 2) where OD_280,ref and OD_412,ref are the corresponding absorption values of ThT in a reference sample without peptide. Using Equation 3, ΔG⁰ can be directly quantified, where m₀ is a reference concentration of 1 M, m is the free monomer concentration at equilibrium, R represents the gas constant, and T is the absolute temperature (52).

$$O{D}_{280,\text{peptide}}=O{D}_{280}-O{D}_{412}\frac{O{D}_{280,\text{ref}}}{O{D}_{412,\text{ref}}}$$ (2)

$$\frac{\left[m\right]}{\left[m_0\right]}=e^{-\frac{\Delta G^0}{RT}}$$ (3)

CD spectroscopy

Far UV CD measurements were carried out on a Jasco J-810 spectrophotometer (Jasco Deutschland GmbH) using a 0.01 cm pathlength quartz cuvette (Hellma GmbH & Co KG). PTH and PTH₃₄ concentrations were adjusted to 20 μM in sodium phosphate buffer (50 mM, pH 7.4). The scan speed was set to 50 nm/min, scanning mode continuous, data pitch 1 nm, and the collected spectra represent an average of 25 scans at 25 °C.

Transmission electron microscopy

The protein samples were diluted to a concentration of 25 μM, and a 5 μl droplet of an individual fibril sample was pipetted on the formvar carbon–coated copper grid (Plano GmbH). The grid was then washed three times with 60 μl of water droplets. The residual water on the grid was removed with a Whatman filter paper. Grids were then stained with 30 μl of 1% (w/v) uranyl acetate, which was subsequently removed, and the copper grids were then air dried. Finally, transmission electron images were taken using an electron microscope (EM 900; Zeiss) at 80 kV acceleration voltages.

Data availability

The here-presented data are to be shared upon request from the corresponding author (jochen.balbach@physik.uni-halle.de).

Supporting information

This article contains supporting information.

Conflict of interest

The authors declare that they have no conflicts of interest with the contents of this article.

Reviewed by members of the JBC Editorial Board. Edited by Wolfgang Peti