Unraveling the underlying mechanisms of reduced amyloidogenic properties in human calcitonin via double mutations

Abstract

The therapeutic efficacy of peptide‐based drugs is commonly hampered by the intrinsic propensity to aggregation. A notable example is human calcitonin (hCT), a peptide hormone comprising 32 amino acids, which is synthesized and secreted by thyroid gland parafollicular cells (C cells). This hormone plays a vital role in regulating blood calcium levels and upholding bone integrity. Despite its physiological importance, utilizing hCT as a drug is hampered by its inclination to form amyloid. To address this limitation, an alternative is provided by salmon calcitonin (sCT), which possesses a lower aggregation propensity. Although sharing the same disulfide bond at the N terminus as hCT, sCT differs from hCT at a total of 16 amino acid positions. However, due to the dissimilarity in sequences, using sCT as a clinical replacement occasionally results in adverse side effects in patients. Earlier investigations have highlighted the significant roles of Tyr‐12 and Asn‐17 in inducing the formation of amyloid fibrils. By introducing double mutations at these sites, the ability to hinder aggregation can be significantly augmented. This study delves into the oligomerization and helical structure formation of the hCT double mutant (Y12LN17H hCT, noted as DM hCT), as well as two single mutants (Y12L and N17H), aiming to elucidate the mechanism behind hCT fibrillization. In addition, computational prediction tools were employed again to identify potential substitutes. Although the results yielded were not entirely satisfactory, a comparison between the newly examined and previously found hCT double mutants provides insights into the reduced aggregation propensity of the latter. This research endeavor holds the promise of informing the design of more effective therapeutic peptide drugs in the future.

1. INTRODUCTION

Human calcitonin (hCT) is a 32‐residue peptide hormone originating from the thyroid and exhibits amyloidogenic properties. This peptide and islet amyloid polypeptide (IAPP) secreted from β‐cells of islets both belong to calcitonin gene‐related peptide families (Austin & Heath, 1981; Barakat et al., ¹⁹⁹⁴). They share a common disulfide bond near the N terminus and an amidated C terminus. Islet amyloid formed by IAPP was frequently found in patients with type 2 diabetes (Hull et al., 2004; Westermark et al., ²⁰¹¹). Although local amyloid formations of calcitonin have predominantly been observed in patients with medullary thyroid carcinoma (MTC), recent findings extend this amyloid presence to kidney biopsies of MTC patients with nephrotic syndrome, as well as abdominal fat pad biopsy specimens, indicating a potential connection of calcitonin to systemic amyloidosis (Khurana et al., 2004). Calcitonin's primary physiological role involves maintaining bone integrity and regulating blood calcium levels by inhibiting osteoclast activity, counteracting the effects of parathyroid hormone produced by the parathyroid glands (Copp et al., 1962; Talmage et al., ¹⁹⁸⁰). This function positions hCT as a therapeutic option for conditions including osteoporosis, Paget's disease of bone, and hypercalcemia. However, the propensity of hCT to aggregate diminishes its bioactivity and restricts its viability as a therapeutic agent. Presently, the active pharmaceutical ingredient in drug products is salmon calcitonin (sCT), a highly potent variant (Lee et al., 2011). sCT also features a disulfide bond at the N terminus but differs from hCT at a total of 16 different positions. The utilization of sCT may lead to undesirable immune responses and potentially severe side effects. Prior studies have revealed that hCT exhibits superior bioactivity compared with sCT when its aggregation propensity is significantly curtailed under specific conditions (Cudd et al., 1995). Consequently, modifying critical residues to prevent protein aggregation shows promise for developing hCT analogs with enhanced or similar therapeutic efficacy. A variant called polar‐hCT (phCT) was created by incorporating five residues from the nonamyloidogenic sCT. The use of the nuclear magnetic resonance (NMR) technique confirmed that phCT's solution structure closely resembled sCT (Andreotti et al., 2011). However, fibril formation rates were significantly reduced when three aromatic residues (Tyr‐12, Phe‐16, and Phe‐19) were replaced with Leu residues Triple‐Leucine hCT (TL‐hCT), but the activity of TL‐hCT remained (Itoh‐Watanabe et al., 2013). Despite various attempts, no successful therapeutic alternatives to hCT have emerged from these endeavors.

The precise mechanism underlying hCT's amyloid formation remains incompletely understood. Based on solid‐state NMR spectroscopy, Naito's group revealed some structural information about the conformational change of hCT during aggregation (Kamihira et al., 2000). First, the hCT monomer was suggested to consist of an N‐terminal loop, central helix, and C‐terminal random coil. The central helix region of hCT starts from residues 10 to 18, whereas sCT adopts a stable and longer central helix from residues 9 to 19 (Andreotti et al., 2006). The less stable central helix of hCT may convert into a β‐hairpin during oligomerization. This new β‐strand may incorporate more residues from the C‐terminal disordered region and gradually become fibrillar with rich β‐sheet structures. The α‐helix spanning residues 9–19 has been proposed as a structural factor influencing receptor binding. Moreover, an excess helical structure originating from sCT variants was identified to impact the affinity for binding. These findings suggest that striking a favorable equilibrium between the desired helix and bioavailability is of significant significance. Previously, our group has shown that two residues of hCT, Tyr‐12 and Asn‐17, play crucial roles in inducing the fibrillization of hCT (Chen et al., 2019). When the residues at positions 12 and 17 in hCT were replaced with the amino acids present in sCT, these minor changes greatly reduced the amyloidogenicity of hCT but did not affect the physiological function. Moreover, Y12LN17H hCT (denoted as DM hCT) can serve as a peptide‐based inhibitor that suppresses amyloid formation induced by hCT. Later, we embarked on a study aimed at deciphering the inhibitory influence of DM hCT with a focus on the formation of helical structures. Our investigations yielded intriguing insights through assessments of four pertinent fragments of the DM hCT peptide. Among the two C‐terminal truncated DM hCT peptide fragments synthesized for this study, DM 1–22 but not DM 1–18 stood out as the peptide capable of amyloid formation (Chuang et al., 2023). Therefore, the disordered C‐terminal domain of DM hCT was considered to restrain fibril formation. The introduction of DM hCT potentially stabilizes hCT through helix–helix interactions, retarding its transition into β‐sheet formations.

Despite DM hCT exhibiting significantly reduced aggregation tendencies compared with hCT, it was still observed to form amyloids on hydrophobic surfaces. In contrast, sCT remained nonaggregated. It is worth noting that the stability and bioactivity of sCT was unparalleled so far. In this ongoing study, we persist in our pursuit of a superior alternative to hCT, building upon insights from prior research. Our exploration commenced by investigating two single mutants (Y12L and N17H) to gain a deeper understanding of the intricate mechanisms underlying hCT fibrillization. This effort aims to enhance our comprehension of the factors contributing to the reduced amyloid‐forming propensity of DM hCT. Subsequently, we employed Waltz (Oliveberg, 2010), a predictive tool for amyloid formation, in conjunction with (Montgomerie et al., 2008), a web server designed for protein structure prediction, to generate additional potential substitutes. Synthesizing these comprehensive outcomes, we put forth another model and energy diagram to elucidate the diminished aggregation tendency of DM hCT. We believe that these findings will prove instrumental in the development of more efficacious therapeutic peptide drugs.

2. RESULTS AND DISCUSSION

2.1. Single mutation is less capable of preventing amyloid formation

Previous studies have harnessed prediction software like Waltz, TANGO (Fernandez‐Escamilla et al., 2004; Rousseau et al., ²⁰⁰⁶), and Zipper DB (Goldschmidt et al., 2010) to engineer hCT into a less aggregation‐prone variant by altering specific residues (Andreotti et al., 2011). DM hCT was categorized as a nonamyloidogenic peptide by both Waltz and TANGO, but Y12L received differing predictions from these two software tools. Waltz and TANGO were developed by the Switch lab under the leadership of Dr. Rousseau and Dr. Schymkowitz. Waltz is a computational algorithm specifically designed to predict amylogenic regions within protein sequences (Oliveberg, 2010). It has been fine‐tuned through extensive training on a comprehensive dataset of experimentally characterized amyloid‐forming peptides, enhancing its accuracy in identifying potential amyloid‐forming regions. In contrast, the TANGO algorithm is tailored to predict cross‐β aggregation in unfolded polypeptide chains. Although cross‐β sheet formation is a hallmark of amyloid fibrils, it is important to note that the presence of cross‐β sheets does not necessarily imply amyloid formation. To gain a deeper understanding of how mutations alter the aggregation properties of hCT, we synthesized the Y12L single mutant and conducted parallel experiments using previously reported hCT variants (Figure 1).

FIGURE 1.

Primary sequence of human calcitonin (hCT), salmon calcitonin (sCT), and hCT variants. Each of these peptides features a disulfide bridge connecting Cys‐1 and Cys‐7, and they all exhibit an amidated C terminus. Residues that differ from the hCT sequence are highlighted in red.

For probing amyloid formation, we employed a fluorescent probe called thioflavin‐T (ThT), a widely used tool in amyloid studies. ThT is characterized by its benzothiazole structure and exhibits a significant increase in fluorescence when it binds to amyloid structures. Consequently, ThT is a valuable means of detecting the presence of amyloid fibrils in a solution (Krebs et al., 2005). The aggregation mechanism of hCT is believed to resemble that of many amyloidogenic proteins, involving a nucleation‐polymerization process leading to the formation of irreversible amyloid aggregates (Chatani & Yamamoto, 2018). This process comprises three distinct stages: nucleation, elongation, and saturation. During the initial nucleation phase (also called lag time), protein monomers come together to form nucleation seeds, which is considered a rate‐limiting step in the overall process. Following this, there is a rapid elongation phase once these seeds are generated. During elongation, protein oligomers continually interact with monomers, promoting the formation of amyloid fibrils. It is important to note that the comparison of aggregation propensity among hCT and its variants has primarily relied on observing lag times, which correspond to the nucleation phase. ThT, as a probe, does not develop strong fluorescence during this lag time. Initially, we quantified the ThT intensity for hCT and its variants in a 50 mM, pH 7.4 phosphate buffer. All samples were prepared at a concentration of 60 μM and incubated in a sealed low‐binding 384‐well microplate under quiescent conditions. As depicted in Figure 2a, the lag time observed for hCT under these conditions was approximately 16 h. The lag time for Y12L was roughly twice as long as that for hCT alone. In contrast, no significant ThT fluorescence was detected for DM hCT and N17H, indicating that they exhibit lower amyloidogenicity compared to hCT and Y12L. Given that amyloid formation was not observed for DM hCT and N17H under quiescent conditions, we conducted additional time‐course ThT measurements for hCT samples under agitation (see method details). Once again, hCT displayed the highest propensity for aggregation, with readily detectable ThT fluorescence. Unfortunately, we were unable to distinguish the rate of amyloid formation between DM hCT and N17H using these two preparation methods (Figure 2b). However, based on our observations, hCT aggregation appeared to be promoted when the peptide solution was incubated in polystyrene microplates. Utilizing this incubation environment, we were eventually able to detect amyloid formation by N17H and DM hCT. N17H forms amyloid faster than DM hCT. Notably, sCT did not exhibit any ThT fluorescence, emphasizing its remarkable resistance to aggregation compared to hCT and its variants (Figure 2c). In summary, the aggregation propensity of hCT and its variants follows the order: hCT > Y12L > N17H > DM hCT > sCT.

FIGURE 2.

Comprehensive thioflavin‐T (ThT) kinetic studies aimed at assessing the aggregation tendencies of human calcitonin (hCT) and its variants. (a) ThT fluorescence measurements for peptides that were incubated in nonbinding 384‐well microplate under quiescent conditions. (b) ThT fluorescence measurements for peptides that were incubated in microtubes under agitated conditions. Blue and green curves shown in (a) and (b) overlap due to very low ThT signals. (c) ThT fluorescence measurements for peptides that were incubated in nontreated polystyrene 384‐well microplate under quiescent conditions. The peptides examined include hCT (black), Y12L (Red), N17H (blue), DM hCT (green), and salmon calcitonin (sCT) (gray). All peptides were prepared at a concentration of 60 μM in a 50 mM, pH 7.4 phosphate buffer.

Besides, we employed a photo‐induced cross‐linking strategy to explore the oligomeric populations of hCT and its variants during the early stages of aggregation. This technique, known as photo‐induced cross‐linking of unmodified proteins, is commonly utilized in biochemistry and structural biology to investigate protein–protein interactions and the structures of proteins in their native, unmodified states (Rahimi et al., 2009). It has also been frequently used in amyloid‐related studies to gain insights into the potential oligomeric species naturally present in freshly prepared protein samples. We have previously employed similar approaches in our studies. Prior reports indicated that hCT and N17H tend to form monomers to tetramers of peptides, whereas DM hCT predominantly exists in a monomeric state with a minor dimer population (Chen et al., 2019). Based on these findings, we speculated that this difference contributes to why DM hCT is much less prone to aggregation compared to hCT and N17H. However, when we included Y12L in this series of examinations, we made an intriguing discovery (Figure S1). Similar to DM hCT, Y12L also predominantly exists in a monomeric state with a small amount of cross‐linked dimers. Furthermore, when we prepared samples at a lower concentration (30 μM), we noted a reduction in trimer and tetramer species for hCT and N17H. Remarkably, Y12L, like DM hCT again, remained predominantly in a monomeric form. However, despite these similarities in monomer‐dimer populations, Y12L displayed a notably higher inclination for aggregation compared with DM hCT and N17H. This unexpected discrepancy challenges our initial expectations and suggests that the formation of oligomers might not be the pivotal step in hCT amyloid formation. Further investigations are currently underway to shed more light on this aspect.

2.2. Y12L mutation enhances the helix formation of peptide and biological activity

The formation and endurance of the helical conformation play a crucial role in the aggregation and bioactivity of calcitonin. In particular, the persistent helical structure in sCT serves as a barrier against amyloid formation and confers substantial bioactivity (Amodeo et al., 1999). In our prior study, we gauged the extent of helical conformations in DM hCT relative to hCT and sCT using circular dichroism (CD) measurements. Additionally, we ascertained the bioactivity of DM hCT by measuring stimulated cyclic adenosine monophosphate (cAMP) levels in MCF‐7 cells treated with the peptide. However, our previous findings lead us to the conclusion that helix–helix association likely represents a pivotal step in the aggregation of hCT and the inhibition of hCT aggregation by DM hCT may occur through this form of association (Chuang et al., 2023). It is likely that the combined ability of DM hCT to adopt a helical structure and directly interact with hCT collaborates in reducing hCT aggregation. These various pieces of evidence show understanding the capability of peptide forming helical structures is important. Following this, we proceeded to track the conformational changes of individual peptides under varying percentages of trifluoroethanol (TFE) and estimated the helical components using BeStSel (Micsonai et al., 2015). Initially, the freshly prepared hCT and hCT variants primarily existed as random coil structures with a limited extent of helical conformation, but the helical component of sCT was more obvious than others (Figure 3a and Figure S2). Upon the addition of TFE to the solution, the peptides gradually transitioned toward adopting helical structures. TFE is a solvent that is known to induce the formation of helical structures in proteins (Arunkumar et al., 1997). This phenomenon is primarily attributed to its ability to disrupt the hydrogen bonding interactions in proteins and promote the formation of helical secondary structures. However, it is important to note that TFE's ability to induce helical structures is highly dependent on the protein's primary sequence and the presence of helix‐promoting amino acids. The extent of helix formation of hCT and hCT variants certainly can be compared via the induction of TFE as the initial discovery of the helical structure in hCT and sCT stemmed from solid‐state NMR studies. Although all the examined peptides displayed an increase in helical components when induced by TFE, the degree of helix formation varied among them. Notably, sCT exhibited the highest inclination to adopt helical conformations, followed by DM hCT. Y12L consistently displayed a higher helical conformation than hCT and N17H under each condition, suggesting that the Tyr to Leu mutation aids hCT in forming α‐helices. This outcome aligns with general expectations in peptide structure. Amino acids with small, compact side chains and a propensity for hydrogen bond formation tend to favor α‐helical structures. For instance, amino acids such as Ala, Leu, and Glu possess relatively high helix propensities and are frequently found in α‐helices (Pace & Scholtz, 1998). Therefore, the substitution of Tyr with Leu likely enhances the helix formation of hCT and may also contribute to the increased helical component observed in DM hCT.

FIGURE 3.

Y12L mutation enhances the helix formation of peptides and assists in binding to the calcitonin receptor. (a) The percentage composition of helix, estimated using circular dichroism fitting software BeStSel, for each condition with different concentrations of trifluoroethanol (TFE). (b) Intracellular cyclic adenosine monophosphate (cAMP) levels measured after stimulation with freshly prepared human calcitonin (hCT) and its variants, including Y12L. sCT, salmon calcitonin.

Furthermore, these results also support our assessment for peptide binding to human calcitonin receptors. It is known that hormone calcitonin has the ability to reduce blood calcium levels through two distinct mechanisms: first, hCT inhibits osteoclast activity within bones, thereby maintaining bone structure (Chambers et al., 1986); second, hCT inhibits the reabsorption of calcium and phosphate by renal tubular cells, allowing them to be excreted in urine. hCT achieves its effects by interacting with the calcitonin receptor, which is primarily located in osteoclasts, thus stimulating the production cAMP by adenylate cyclase (Martin et al., 1980). In this context, we assessed the bioactivities of hCT and hCT variants by measuring the levels of stimulated intracellular cAMP in T‐47D breast cancer cells treated with these peptides. The highest concentration of stimulated cAMP was obtained by treatment of DM hCT, followed by Y12L (sCT was not included in comparison). The level of cAMP generated from these two peptides are much higher than the one from hCT and N17H (Figure 3b). This finding was well correlated to peptide helical conformation.

2.3. Protein structure prediction was utilized to create a more aggregation‐resistant hCT double mutation

Based on our examination results, it became evident that a single point mutation is insufficient to prevent hCT aggregation. Previous research demonstrated that a single point mutation could effectively suppress the amyloid‐forming ability of IAPP and convert it into a potent inhibitor of IAPP fibrillization. This discovery was inspired by proline scanning studies of the highly amyloidogenic region from 20 to 29 in IAPP, as proline is not favorable for β‐sheet structure (Abedini et al., 2007). In this context, we previously synthesized Y12P hCT, and this single mutant was also considered non‐amyloidogenic by prediction tools like Waltz and TANGO (Hsieh et al., 2022). Y12P did indeed show promise in preventing amyloid formation. However, it encountered solubility issues, tending to form large aggregates as observed through dynamic light scattering. More critically, Y12P completely lost its ability to bind to the hCT receptor, indicating that this variant could not strike a good balance between amyloidogenicity and bioactivity. Considering the importance of adopting a helical conformation for hCT bioactivity, we also employed another prediction server called “PROTEUS2” to suggest appropriate residues for mutation other than those found in the sCT sequence (Montgomerie et al., 2008). Initially, our aim was to ensure that the newly designed hCT variants had a high propensity for helix formation (Table S1). PROTEUS2 is designed to support comprehensive protein structure prediction and structure‐based annotation. However, when we submitted various sequences for Y12X (X representing 20 different amino acids), none of them were suggested to have a high score for helix component. This was somewhat unexpected since Y12L readily formed a helix when induced by TFE and was able to bind to the hCT receptor, as our experiments showed. However, N17M was suggested to adopt a helical structure with the highest score. Based on this result, we continued by submitting different sequences (Y12XN17M) to PROTEUS2. Some double mutations achieved the same score as N17M, whereas others caused a decrease in the helical conformation score. Initially, Y12V, Y12L, or Y17I combined with N17M were found to have a better preference for forming a helix. Given that the side chains of Val and Ile are quite similar to Leu, and our data showed that the Tyr to Leu mutation enhances helix conformation, we chose and synthesized Y12LN17M for further studies. Second, Y12F and Y12W combined with N17M were also found to prefer a helical structure. However, Tyr, Phe, and Trp are all aromatic amino acids found in nature. For the time being, we did not consider creating Y12FN17M or Y12WN17M. Lastly, Y17MN17M appeared to be a promising candidate since a previously reported helix propensity scale, based on experimental studies of proteins and peptides, suggests that Met appears with high frequency in a helical structure. Following this, we submitted the Y12LN17M and Y12MN17M peptide sequences to Waltz again, and they were both suggested to be non‐amyloidogenic (Table S2). Later, we synthesized these two peptides and performed further examination.

2.4. Y12LN17M and Y12MN17M fail to prevent amyloid formation

To understand the amyloidogenicity of the newly designed double mutant peptides (Y12LN17M and Y12MN17M), ThT assay was conducted once more, with wild‐type hCT included as an important control. The results showed that Y12LN17M and Y12MN17M, started to exhibit ThT fluorescence after approximately 20 h of incubation in the nonbinding plate (Figure 4a). Confirmation of amyloid fibril formation by these peptides was obtained by acquiring transmission electron microscope (TEM) images of the ThT end products (Figure 4b–d). Since these two double mutants were initially selected for their preference for forming a helical structure, we further measured their helical component after induction with TFE. Regardless of the percentage of TFE used, both Y12LN17M and Y12MN17M exhibited a much higher ratio of α‐helix than wild‐type hCT. In particular, Y12LN17M showed a similar extent of helical structure as DM hCT in each condition, suggesting that the incorporation of Leu at residue 12 facilitates the peptide to adopt a helical conformation. Although the helical component of Y12MN17M was significantly higher than that of wild‐type hCT, it was slightly lower than that of Y12LN17M and DM hCT (Figure 4e and Figure S3). Since TFE can enhance the helical contents of the peptide, we performed the ThT assay again with the conditions containing 30% TFE. We confirmed that fibril formation still occurred by using these two variants although the lag time was also prolonged (Figure S4), especially for Y12LN17M. Regarding peptide bioactivity, these two double mutants were found to stimulate cAMP to a similar extent as DM hCT, indicating that their binding affinity to the receptor falls within a similar range and is significantly higher than that of wild‐type hCT (Figure S5).

FIGURE 4.

The newly designed double mutant peptides form amyloid fibrils after incubation. (a) thioflavin‐T (ThT) fluorescence measurements for human calcitonin (hCT) (black), Y12LN17M (brown), Y12MN17M (orange), and DM hCT (green) in 50 mM phosphate buffer at pH 7.4. The peptide concentration was fixed at 60 μM. (b–d) Transmission electron microscope images for hCT, Y12LN17M, and Y12MN17M. Samples were taken from the end of the ThT assays. The scale bar represents 500 nm. (e) The percentage composition of helix for hCT (black), Y12LN17M (brown), Y12MN17M (orange), and DM hCT (green) with different concentrations of trifluoroethanol (TFE).

2.5. The crucial role of Asn‐17 in forming hCT amyloid

Understanding the mechanism of amyloid formation, especially for a protein like hCT that has exhibited amyloidogenic features, is indeed a challenging task. The research via molecular dynamics simulations also was employed to enhance our understanding aggregation properties of hCT and its mutated variants (Liu et al., 2023; Paul & Paul, ²⁰²¹). Y12L and N17H were considered to have more favorable interpeptide energy than sCT and DM hCT, suggesting more peptide self‐assembly occur in these two single mutations. Our examination here indeed confirmed both single mutants are amyloidogenic. Sun's group with help of all atom discrete molecular dynamics revealed that wild‐type hCT monomers mainly adopted unstructured conformations with dynamic helices around the central region and readily self‐assembled into helical oligomers followed by helix‐to‐sheet conformational conversion to form fibrils. Substitution with leucine in TL‐hCT and replacement of hydrophilic amino acid at C terminus in phCT mainly suppress β‐sheet propensities only in central region or C‐terminal regions. Only in sCT, both regions were stabilized by mutations. Therefore, enhancing central helix stability and decreasing C‐terminal hydrophobicity would be two important strategies for design of hCT analogs. Although DM hCT has shown resistance to amyloid formation, it has not proven to be a suitable replacement for the current pharmaceutical ingredient, sCT, in osteoporosis applications. Nevertheless, the experimental data obtained from designing and studying new double hCT mutants have contributed to a deeper understanding of hCT amyloid formation. In previous studies, the D₁₅FNKF₁₉ fragment was identified as the minimum sequence required for fibril formation and was found to play a crucial role in hCT amyloid formation (Bertolani et al., 2017). Specific residues, such as Asp‐15 and Lys‐18, were suggested to mediate the orientation of β‐sheet motifs in the fibrillar structure. Solid‐state ¹³C NMR spectroscopy confirmed that residues Gly‐10 to Phe‐22 of hCT could form an antiparallel β‐sheet structure at pH 7.5 (Kamihira et al., 2000). The π–π interaction between the phenyl rings of Phe‐16 and Phe‐19 on the same side of the β‐sheet was noted for enhancing fibril stability. Additionally, crystal packing of the iodinated derivative of this fragment revealed that the amide side chains of Asn‐17 stacked within a sheet, forming an “Asn ladder,” which is also significant for peptide assembly (Tsai et al., 2005).

Comparing the kinetic data of different hCT variants, it became evident that Asn‐17 plays a role at the interface of peptide–peptide associations, affecting the affinity of peptide interaction and fibril formation. Variants N17H and DM hCT significantly delayed amyloid formation. Although mutations like Y12L, Y12LN17M, and Y12MN17M increased the helix conformation of hCT, they may not be stable enough to prevent the conformational change required to form cross β‐sheets. The substitution of Asn with His in N17H changes two key properties of the amino acid side chain: size and charge. Histidine has a larger side chain volume than asparagine Asn and provides a partial positive charge at pH 7.4. Both changes are typically considered unfavorable for peptide assembly (Figure 5). Combined with previous findings that helix–helix association is a critical step for hCT amyloid formation and inhibition (Chuang et al., 2023), the N17H substitution could make helix–helix association more difficult or make the helix‐to‐β‐sheet conversion more difficult to delay fibril formation. Because hCT cannot form a stable helix like sCT or DM hCT and lacks repulsion forces to prevent peptide association, it is more prone to forming amyloids. Overall, these findings contribute to a deeper understanding of the factors influencing hCT amyloid formation and provide insights into the design of variants with improved properties for potential therapeutic use.

FIGURE 5.

A schematic diagram help illustrate the differences in aggregation kinetics between human calcitonin (hCT) and the N17H variant. In hCT, the formation of helix–helix associations is more favorable, which promotes the aggregation process. However, in the N17H variant, the helix–helix association is less favorable due to the changes introduced by the histidine substitution. This reduced propensity for helix–helix association slows down the aggregation kinetics, delaying amyloid formation. His residues may also perturb the conformation change from helix to β‐sheets.

3. CONCLUSION

In previous investigations, it has been established that mutations can be utilized to alter the aggregation characteristics of peptides. As a result, this strategy holds promise for a diverse array of peptide‐based therapeutic applications, and it is not limited to a specific protein or disease. It can be applied across various therapeutic proteins and peptides, offering a versatile approach to improving the characteristics of a wide range of biopharmaceuticals. Protein aggregation can introduce variability in drug performance, affecting dose–response relationships. Aggregation‐resistant modifications contribute to a more predictable and consistent drug response, ensuring that patients receive the intended therapeutic effect. Besides, modifying critical residues to prevent aggregation can help minimize immunogenicity, making the therapeutic protein more tolerable and suitable for long‐term use. Therefore, it is crucial to identify the specific amino acid residues that govern peptide–peptide interactions and protein aggregation. Our research has revealed that introducing just two mutations at positions 12 and 17 is able to greatly slow down the formation of hCT fibrils. However, it is important to note that DM hCT is not entirely devoid of amyloid‐forming potential. Pre‐existing amyloid fibrils could induce fibril formation by DM hCT, but not by sCT (Chen et al., 2019).

In this investigation, it became apparent that single mutations occurring either solely on residue 12 or solely on residue 17 were ineffective in preventing hCT aggregation. However, these mutations did yield valuable structural insights that contributed to our understanding of the reduced aggregation propensity observed in DM hCT. In parallel, we endeavored to design an improved double mutation variant with the assistance of predictive software. Past experiences have taught us that although we can engineer peptides with low aggregation tendencies, their practical application is contingent upon retaining bioactivity. Consequently, we needed to strike a balance between bioactivity and reduced aggregation properties in our peptide design. To achieve this, we adopted an approach for the first time, utilizing the “PROTEUS2” protein structure prediction web tool to identify peptides with a high propensity for adopting a helical structure. Subsequently, we employed the “Waltz” amyloid‐forming peptides prediction software to identify peptides with non‐amyloidogenic characteristics. However, it is worth noting that Y12LN17M and Y12MN17M, although initially displaying a high helical component when induced by TFE, ultimately formed amyloid structures after incubation. Previously, “Waltz” has successfully identified amyloidogenic regions in challenging targets such as yeast prions and functional amyloids, often considered difficult to predict. The software has also identified numerous potential amyloidogenic sequences in disease‐associated proteins, with an impressive 80% success rate upon experimental validation (de Groot et al., ²⁰⁰⁵; Seidler et al., ²⁰¹⁸; Tenidis et al., ²⁰⁰⁰; Tsiolaki et al., ²⁰¹⁵). It is important to mention that the presence of a disulfide bond bridge in hCT may have added complexity to the prediction process. We utilized the software's guidance for identifying non‐amyloidogenic peptides but did not employ it for predicting the amyloidogenic regions, as this appeared to contradict the intended purpose of the Switch laboratory. In future research, we shall also focus on a comprehensive characterization of the amyloidogenic regions within Y12LN17M and Y12MN17M hCT, contributing data to expand the “WALTZ‐DB 2.0” database (Louros et al., 2020).

Although our new peptide designs did not fully achieve our intended goal of creating a more stable and bioactive alternative to hCT, this endeavor has provided valuable insights into the behavior of hCT variants. Notably, the substitution of Tyr‐12 with certain amino acids that have a strong preference for helical structures has indeed facilitated the adoption of a helical conformation by hCT. Moreover, these variants, including Y12L, Y12LN17M, and Y12MN17M, have demonstrated the ability to interact with the hCT receptor. However, it should be noted that these variants may not exhibit the same degree of stability as sCT, which is critical in preventing the process of helix–helix association and the eventual formation of β‐sheet‐rich amyloid structures. Intriguingly, the mutation of Asn‐17 to His appears to be more effective in preventing hCT aggregation. As we move forward, our research will place greater emphasis on this particular residue, aiming to extend our understanding of the aggregation mechanisms associated with this hormone peptide. We firmly believe that through these foundational investigations, we will be better equipped to achieve improved protein designs in the future, ultimately contributing to advancements in the field of peptide therapeutics.

4. MATERIALS AND METHODS

4.1. Peptide synthesis, purification, and preparation

Synthetic hCT, sCT, and hCT‐related variants (Y12L, N17H, DM hCT, Y12LN17M, and Y12MN17M) were synthesized with the assistance of a microwave peptide synthesizer. Fmoc‐Rink amide ProTide (0.16 mmol/g, CEM Corporation) served as the solid support and facilitated the amidation of the C terminus after peptide cleavage. A comprehensive account of the peptide synthesis procedure has been previously documented (Chen et al., 2019). In brief, the synthesis involved several steps. Initially, the Fmoc protecting group was removed from the resin using a 10% piperazine (w/v) solution prepared from a mixture of ethanol and N‐methylpyrrolidone (10:90). The carboxylic acid group of the first amino acid, after activation with 0.25 M diisopropylcarbodiimide in dimethylformamide, underwent a coupling reaction to establish a new amide bond, attaching it to the reaction resin. Subsequently, a series of iterative actions (deprotection–activation–coupling) were employed to synthesize the peptide, progressing from the C terminus to the N terminus. Finally, a cleavage reaction, involving a cocktail of trifluoroacetic acid, water, triisopropylsilane, and 3,6‐dioxa‐1,8‐octanedithiol (in a ratio of 92.5: 2.5: 2.5: 2.5), was performed to detach the peptide from the resin. The resulting crude peptides were precipitated using cold ether after the evaporation of trifluoroacetic acid. All peptides underwent further oxidation to establish disulfide bonds between Cys‐1 and Cys‐7 using I₂ dissolved in methanol. Purification was carried out via reverse‐phase high‐performance liquid chromatography (HPLC) utilizing a Proto 300 C18 semipreparative column (10 mm × 250 mm, Higgins Analytical). Two solvent solutions were employed: solution A, composed of 100% H₂O and 0.045% HCl (v/v), and solution B, consisting of 80% acetonitrile, 20% H₂O, and 0.045% HCl. Peak fractions were separately collected and subsequently subjected to lyophilization. The identity of the purified products was validated by determining their molecular weights using a Bruker matrix‐assisted laser desorption/ionization time‐of‐flight mass spectrometry. HPLC chromatograms and mass spectra for the aforementioned peptides were provided in Figures S6–S12. Prior to experiments, approximately 0.1 mg of peptide powder treated with hexafluoroisopropanol before was dissolved in 300–400 μL of 50 mM phosphate buffer at pH 7.4. The solution was then centrifuged at 15,000 rpm for 10 min to remove any possible preformed aggregates, and the supernatant was transferred to another microtube. To determine protein concentration, 10 μL of each peptide solution was employed as suggested in a bicinchoninic acid protein assay kit (Thermo Fisher Scientific, USA).

4.2. ThT kinetic assay

Amyloid formation by hCT, sCT, and hCT‐related variants (Y12L, N17H, DM hCT, Y12LN17M, and Y12MN17M) was assessed through ThT assays. Typically, peptide solutions were prepared at a concentration of 64 μM in a 50 mM phosphate buffer containing 16 μM ThT at a pH of 7.4. These solutions were then incubated in microtubes (Axygen, MCT‐150‐C) with a total volume of 400 μL with continuously shaking at 500 rpm, or alternatively, in two different sealed 384‐well microplates (Corning #3575 and Greiner #781076) with 40 μL of solution per well. It is worth noting that Corning #3575 microplates feature nonbinding surface polystyrene. Measurements were conducted using a multimode microplate reader (SpectraMax M2, Molecular Devices, USA) with excitation at 430 nm and emission at 485 nm. To ensure the reliability and consistency of the experimental results, each condition was examined in triplicate, and samples from at least two different batches were employed for analysis.

4.3. Photo‐induced cross‐linking of unmodified proteins

hCT and its variants were freshly prepared at concentrations of 30 or 60 μM in a 50 mM phosphate buffer at pH 7.4, and the solutions were maintained at a temperature of 25°C. To facilitate cross‐linking, samples were treated with 1 mM Ru(bpy)₃ ²⁺ and 20 mM ammonium persulfate (APS) in a 10 mM sodium phosphate solution, with a ratio of peptide to Ru(bpy)₃ ²⁺ to APS set at 1:2:5. The cross‐linking reaction was initiated by illuminating the samples using a 150 W incandescent light bulb for a duration of 5 s. To halt the reaction, sodium dodecyl sulfate sample dye was added, and the samples were subsequently separated through electrophoresis using a 13.5% Tris‐tricine sodium dodecyl sulfate (SDS) polyacrylamide gel. Visualization of the gel was achieved through silver staining, utilizing the Invitrogen SilverQuest Silver Staining Kit.

4.4. CD measurements

CD experiments were conducted utilizing a JASCO J‐715 CD spectrometer. Peptide solutions were prepared at a concentration of 64 μM as previously outlined, but in a 50 mM phosphate buffer containing varying percentages of TFE. Generally, a 1 mm path length quartz cell, holding 280 μL of peptide solution, was employed for CD measurements. Spectra were recorded in the wavelength range of 200 to 260 nm with 1 nm intervals at a temperature of 25°C, utilizing a scan speed of 50 nm/min. The data acquisition process involved averaging results from 10 scans and subtracting the background spectrum. To determine the percentage of helical content for each peptide under different conditions, BeStSel, a method for secondary structure determination, was employed.

4.5. Transmission electron microscope

TEM was conducted at the instrumentation center of National Taiwan University using a Hitachi H‐7100 transmission electron microscope, operated at an accelerating voltage of 120 kV. The TEM samples were prepared from the same solutions that were used for ThT assays. Ten microliters of alcohol were initially applied to a carbon‐coated Formvar 300‐mesh copper grid. Excess alcohol was carefully removed from the grid using filter paper. Subsequently, 10 μL of the solution, which was transferred from the end of the ThT assays, was blotted onto the grid and allowed to sit for 2 min. The grid was then washed twice with distilled deionized (DDI) H₂O. Finally, the grid was negatively stained with saturated uranyl acetate for another 1 min.

4.6. cAMP assay

The T47D human breast adenocarcinoma cell line was procured from the Bioresource Collection and Research Center in Hsinchu, Taiwan. These cells were cultured in 10 cm dishes using RPMI 1640 medium supplemented with 10% fetal bovine serum and 0.1% insulin. Cell cultures were maintained at 37°C in a 5% CO₂ incubator. In general, cells were seeded into 96‐well culture plates at a density of 5 × 10⁴ cells per well and allowed to grow for 18–24 h in a 37°C incubator. The culture medium was removed, and cells were washed with phosphate‐buffered saline. After then, cells were then incubated in a medium containing bovine serum albumin (BSA), and the BSA medium was removed just prior to sample treatments. The cells were treated with hCT and hCT‐related variants at concentrations of 1 nM for a duration of 15 min at 37°C. Subsequently, the cells in the plates were placed on ice and lysed by adding 60 mL of cell lysis buffer. The lysed cell suspension was centrifuged at 3000 rpm for 5 min to remove cell pellets. Plate primers were added (25 μL) to wells that were coated with an antibody to capture sheep immunoglobulin G. Standards or samples (50 μL) and sample diluents (50 μL) were pipetted into the primed wells. Next, cAMP–HRP (horseradish peroxidase) conjugate and sheep anti‐cAMP antibody were added into each well using a repeater pipet. After a 2‐h incubation, each well was washed four times with 300 mL of wash buffer. To initiate color development, 100 μL of Tetramethylbenzidine (TMB) substrate was added to the wells and allowed to develop for 30 min without shaking. The reaction was terminated by adding 50 μL of 1 M HCl. The developed signals, indicated by absorbance at 450 nm, were recorded using a microplate reader (SpectraMax M2, Molecular Devices, USA).

AUTHOR CONTRIBUTIONS

Ling‐Hsien Tu: Conceptualization; formal analysis; supervision; project administration; visualization; writing – review and editing; writing – original draft; funding acquisition; investigation. Yu‐Pei Chang: Data curation; investigation; formal analysis. Pei‐Chun Pan: Investigation; data curation; validation.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

Supporting information

Figure S1. Oligomeric species generated from (a) 30 μM and (b) 60 μM of hCT and its variants were stabilized using PICUP and subsequently detected through silver staining. This experimental approach allowed for the identification and visualization of the oligomeric forms of the peptides generated under these specific conditions.

Figure S2. CD spectra of 60 μM (a) hCT, (b) sCT, (c) N17H, (d) Y12L, and (e) DM hCT in a 50 mM phosphate buffer containing varying percentages of TFE.

Figure S3. CD spectra of 60 μM (a) Y12LN17M and (b) Y12MN17M in a 50 mM phosphate buffer containing varying percentages of TFE.

Figure S4. ThT fluorescence measurements for hCT (black), Y12LN17M (brown), and Y12MN17M (orange) in 50 mM phosphate buffer containing 30% TFE at pH 7.4. Peptide concentration was fixed at 60 μM.

Figure S5. Intracellular cAMP levels were measured after stimulation with freshly prepared hCT and its variants, including Y12LN17M and Y12MN17M.

Figure S6. (a) The HPLC chromatogram and (b) mass spectrum of hCT. The monotopic molecular weight determined for hCT is 3416.309.

Figure S7. (a) The HPLC chromatogram and (b) mass spectrum of sCT. The monotopic molecular weight determined for sCT is 3430.436.

Figure S8. (a) The HPLC chromatogram and (b) mass spectrum of Y12L. The monotopic molecular weight determined for Y12L is 3366.633.

Figure S9. (a) The HPLC chromatogram and (b) mass spectrum of N17H. The monotopic molecular weight determined for N17H is 3438.546.

Figure S10. (a) The HPLC chromatogram and (b) mass spectrum of DM hCT. The monotopic molecular weight determined for DM hCT is 3388.916.

Figure S11. (a) The HPLC chromatogram and (b) mass spectrum of Y12LN17M. The monotopic molecular weight determined for Y12LN17M is 3383.278.

Figure S12. (a) The HPLC chromatogram and (b) mass spectrum of Y12MN17M. The monotopic molecular weight determined for Y12MN17M is 3400.432.

Table S1: Protein secondary structure prediction by PROTEUS2.

Table S2: Amyloid formation predictions by Waltz. Calculations were performed by submitting two different lengths of sequence, residue 9 to 22 and full length.

Chang Y‐P, Pan P‐C, Tu L‐H. Unraveling the underlying mechanisms of reduced amyloidogenic properties in human calcitonin via double mutations. Protein Science. 2024;33(4):e4952. 10.1002/pro.4952