Sourav
Kalita
a,
Sujan
Kalita
a,
Ashim
Paul
a,
Manisha
Shah
b,
Sachin
Kumar
b and
Bhubaneswar
Mandal
*a
aLaboratory of Peptide and Amyloid Research, Department of Chemistry, Indian Institute of Technology Guwahati, Assam-781039, India. E-mail: bmandal@iitg.ac.in
bDepartment of Biosciences & Bioengineering, Indian Institute of Technology Guwahati, Assam-781039, India
First published on 15th January 2021
Amylin or hIAPP, together with insulin, plays a significant role in glucose metabolism. However, it undergoes β-sheet rich amyloid formation associated with pancreatic β-cell dysfunction leading to type-2 diabetes (T2D). Recent studies suggest that restricting β-sheet formation in it may halt amyloid formation, which may limit the risk for the disease. Several peptide-based inhibitors have been reported to prevent aggregation. However, most of them have limitations, including low binding efficiency, active only at higher doses, poor solubility, and proteolytic degradation. Insertion of non-coded amino acids renders proteolytically stable peptides. We incorporated a structurally rigid β-amino acid, Anthranilic acid (Ant), at different sites within the central hydrophobic region of hIAPP and developed two singly mutated hIAPP8–37 peptidomimetics. These peptidomimetics inhibited the amyloid formation of hIAPP substantially even at low concentration, as evident from in vitro ThT, CD, FT-IR, TEM, and Congo red staining birefringence results. These peptidomimetics also disrupted the preformed aggregates formed by hIAPP into non-toxic species. These β-amino acid-based peptidomimetics can be an attractive scaffold for therapeutic design towards T2D or other protein misfolding diseases.
Although hIAPP complements insulin in keeping the glucose equanimity by suppressing glucose secretion and regulating gastric emptying, it is highly amyloidogenic.7 Development of an effective inhibitor of hIAPP aggregation is a challenging task. Until now, no cure for T2D exists. However, one of the most popular strategies is β-sheet breakers, where one recognition moiety targets the protein of interest, and the activity element averts β-sheet formation. One positive aspect of peptides is their ability to bind a large target surface efficiently and selectively, which is the prerequisite of protein–protein interactions (PPI).8–10 This feature of the peptides enables their use as a therapeutic agent, which has grown rapidly over the decades, including metabolic and cardiovascular diseases.11
We previously demonstrated the inhibitory efficacy of β-sheet breaker hybrid peptidomimetics (BSBHPs) by inserting one breaker element, isomers of aminobenzoic acid, into the core hydrophobic region of hIAPP (hIAPP22-27). We found 2-Aminobenzoic acid or Anthranilic acid (Ant) is an efficient breaker element.12Ant induces conformational restriction in the peptidomimetics due to its structure. In the structure of Ant, both the amine group and the carbonyl group are directly connected to the aromatic moiety, constituting a planar structure with a fixed dihedral angle, ∅ = 0°. The structural rigidity of peptides containing Ant are significantly enhanced due to the π-stacking ability of the aromatic moiety.13,14 Due to its higher structural rigidity, we inserted Anthranilic acid as a β-breaker element within the central hydrophobic region of hIAPP peptide. Furthermore, insertion of Ant in peptide sequences favors either a turn or a helix conformation.15,16
Earlier, Raleigh et al. observed that a single point mutation by proline in hIAPP converts it into a highly effective inhibitor.7 Further, a three proline containing analog of hIAPP, with less tendency to aggregate, called Pramlintide (PM), has been approved for clinical applications. However, it also suffers from low solubility, particularly at physiological pH, preventing its co-formulation with insulin.17 Moreover, peptides are highly susceptible to proteolytic degradation; but introducing β-amino acids instead of α-amino acids to the peptide sequence decreases the proteolytic degradation.12,18
The core hexapeptide, NFGAIL (hIAPP22–27), is highly amyloidogenic and a significant driver for hIAPP aggregation. The Ant containing small BSBHPs was found useful to inhibit the aggregation of hIAPP and disrupt its preformed fibrillar assemblies at a relatively higher dose (10-fold molar excess). These BSHBPs are stable towards proteolytic degradation due to non-coded β-amino acid Ant.12 However, these BSBHPs may not effectively bind with the full-length hIAPP due to smaller size and lack of residual interaction. Probably, therefore, a high dose was necessary for the inhibition of aggregation.
Although the mechanism is not fully understood, it is hypothesized that the inhibitors need to bind with the growing fibrils of the aggregating peptide to inhibit aggregation, and a mutant of the full-length hIAPP may bind more tightly than the small fragment of hIAPP.7,19 Therefore, we decided to synthesize a new set of peptidomimetics, comprising of hIAPP8–37 with a single point mutation at different positions, which are expected to exhibit inhibitory effect at fewer molar ratios compared to the smaller hIAPP22–27 variant (Scheme 1).
Code | Sequence | Functions |
---|---|---|
Note: standard amino acids are represented by one-letter code, X = Anthranilic acid (Ant). | ||
A | 8ATQRLANFLVHSSNNFGAILSSTNVGSNT37Y | Aggregating |
B1 | 8ATQRLANFLVHSSNNFXAILSSTNVGSNT37Y | Breaker |
B2 | 8ATQRLANFLVHSSNNFGAXLSSTNVGSNT37Y | Breaker |
C | 22NFGAX27L | Control |
As the entire hexapeptide is responsible for forming β-sheet containing amyloid fibril, we may select any one position to incorporate the breaker element inside the peptide sequence. Hence to maintain sequence homology as a standard control breaker peptide, we selected I26 for mutation with Ant instead of G24.12,20,21
The designed peptidomimetics were synthesized by standard solid-phase peptide synthesis method using Fmoc/t-Bu strategy on Rink Amide MBHA resin, purified by reverse-phase HPLC, and characterized by MALDI mass spectrometry (ESI S1–S8, ESI†).23 Commercially available wild type hIAPP1–37 was used as a native aggregating system for the present study.
The appearance of fibrillar assembly under an electron microscope is a characteristic property of amyloid.24 Similarly, the presence of green gold birefringence under cross-polarised light upon staining with Congo red is another characteristic feature of amyloid formation.24A showed a clear fibrillar structure under TEM and green gold birefringence under cross-polarised light (Fig. 1(a and b)); however, no such characteristic appearance was noted for B1, B2, and C, respectively (Fig. 1(a and b)).
Fig. 1 (a) TEM and (b) Congo red-stained birefringence images of A (i), B1 (ii), B2 (iii), and C (iv), respectively. Images were taken after five days of incubation in PBS at pH 7.4 and 37 °C. The scale bar for TEM and Congo red stained birefringence is 200 nm and 20 μm, respectively. (c) CD and (d) FTIR spectra of A (black), B1 (red), B2 (blue) and C (magenta). Spectra were recorded after five days of incubation in PBS at pH 7.4 and 37 °C. A zoom-in inset image for the FTIR spectra has been incorporated in Fig. 2(b). |
After five days of incubation, A displayed a positive band at ∼195 nm and a negative band centered at ∼225 nm (black, Fig. 1(c)) in CD, indicating its β-sheet rich conformation. Similarly, we observed a strong amide I band at 1639 cm−1 for A (black, Fig. 1(d)) in the FTIR profile, which is also a characteristic feature of a β-sheet rich conformation.24 In contrast, the other three peptides, B1, B2, and C, did not show any such characteristic features for β-sheet rich conformation in CD and FTIR, suggesting their non-amyloidogenic nature.15,24
The above results indicate that B1, B2, and C do not form amyloid under physiological conditions. Nevertheless, A, which lacks turn-inducing Ant in the backbone, is highly amyloidogenic similar to wild-type hIAPP.
After seven days of incubation, hIAPP alone exhibited clear fibrillar morphology (Fig. 2(c)-(i)) when viewed under TEM. However, hIAPP, when incubated with a 2-fold molar excess of B1 (Fig. 2(c)-(ii)) and B2 (Fig. 2(c)-(iii)), no such fibrils were observed, indicating substantial inhibition of aggregation. On the contrary, in the presence of 2-fold molar excess of C (Fig. 2(c)-(iv)), some fibrillar assembly was observed, indicating less efficiency of C. Further, hIAPP alone showed green gold birefringence under cross-polarised light, after staining with Congo-red (Fig. 2(d)-(i)), indicating amyloid formation. On the other hand, in the presence of 2-fold molar excess of B1 (Fig. 2(d)-(ii)) and B2 (Fig. 2(d)-(iii)), no such green-gold birefringence was observed, which indicated significant inhibition of amyloid aggregation by the peptidomimetics. However, in the presence of C (Fig. 2(d)-(iv)), some characteristic green-gold birefringence was observed, implying relatively less efficiency of the same to inhibit amyloid aggregation. After seven days of incubation, hIAPP alone exhibited clear β-sheet rich conformation, as evident from CD and FTIR analyses. But, hIAPP, when incubated with a 2-fold molar excess of B1, B2, and C, such β-sheet conformation was not observed in both CD and FTIR spectra, indicating inhibition of β-sheet formation. Similar results were obtained with 0.5- and 1-fold molar excess of peptidomimetics when co-incubated with hIAPP with relatively less efficiency (ESI S10 and S11, ESI†). All the above results collectively indicate that B1, B2, and C were able to inhibit the aggregation of wild type hIAPP. However, B1 and B2 emerged to be better inhibitors than C. No significant difference in the inhibition capability of B1 and B2 was noted, which implies the position of Ant does not alter aggregation inhibition capability significantly. Therefore, we proceeded with B1 only for other experiments.
From the dose-dependent studies, it was inferred that in the presence of 0.5-, 1- and 2-fold molar excess of B1 (red, Fig. 3(b)), hIAPP exhibited 51%, 61%, and 72%. In contrast, with the same doses of C (blue, Fig. 3(b)), it exhibited 27%, 32%, and 38% of fibril disruption, respectively. Hence, the ability of fibril disruption by the designed peptidomimetics increased in a dose-dependent manner (ESI S12, ESI†).
After seven days of incubation, hIAPP alone (Fig. 3(c)-(i)) showed fibrillar assembly under the electron microscope (TEM). Whereas, in the presence of 2-fold molar excess of B1 (Fig. 3(c)-(ii)), hIAPP did not show any such fibrillar assembly indicating significant disruption of preformed amyloid. However, a 2-fold molar excess of C (Fig. 3(c)-(iii)) could not disrupt the preformed fibril completely. Similarly, hIAPP (Fig. 3(d)-(i)), when incubated alone, exhibited clear green-gold birefringence under the cross-polarised light after staining with Congo-red dye, indicating amyloid formation. But, in the presence of 2-fold molar excess of B1 (Fig. 3(d)-(ii)), no such characteristic birefringence was observed, indicating significant disruption of preformed amyloid. However, in the presence of the same equivalent of C (Fig. 3(d)-(iii)), some green gold birefringence persisted, which indicated incomplete amyloid disruption.
The fibril disrupting-ability was also investigated using CD and FT-IR. When hIAPP was incubated alone for 7 (2 + 5) days at physiological condition, β-sheet conformation was observed in the CD profiles. However, in the presence of 2-fold molar excess of B1, a random coil conformation was observed, which indicated disruption of the preformed amyloid of hIAPP.
On the other hand, when C was present in 2-fold molar excess, β-sheet rich conformation was observed, indicating inefficiency (ESI S13, ESI†). Similarly, after seven days of incubation, hIAPP alone showed a sharp band at 1634 cm−1 in FT-IR spectroscopy, a characteristic amide I band for aggregated β-sheet. Again, when B1 was present in 2-fold molar excesses, the band shifted to a 1654 cm−1, indicating its β-sheet disrupting capability. In the presence of 2-fold molar excess of C too, the band shifted to 1648 cm−1. The disruption of preformed amyloid improved in a dose-dependent manner (ESI S13 and S14, ESI†). The above systematic studies collectively indicated B1 was highly efficient in disrupting the preformed amyloid of hIAPP with 2-fold molar excesses only, whereas C was significantly less effective at that concentration.
We noted that both B1 and C disrupted preformed amyloid. To investigate whether they converted the fibrils into membrane-damaging toxic species after disruption, we carried out a dye leakage assay using carboxyfluorescein dye entrapped Large Unilamellar Vesicles (LUVs). LUVs acted as an artificial cell membrane. We prepared five sets of samples to perform the vesicle leakage assay, including the untreated LUVs (without any peptide) as a control. B1 and C were added separately to the preformed fibrillar assemblies of hIAPP, i.e., after 48h of incubation, and co-incubated for an additional five days (168 h = 48 h + 120 h). The different samples prepared for the dye leakage assay was brought up as follows:
Sample 1 – untreated LUVs
Sample 2 – hIAPP (after 5h incubation) + LUVs
Sample 3 – hIAPP (after 7 days incubation) + LUVs
Sample 4 – hIAPP: B1 (1:2) + LUVs
Sample 5 – hIAPP: C (1:2) + LUVs
The complete dye leakage was achieved by the addition of Triton X-100 (10 μL) and considered it as total fluorescence. The percentage of dye leakage was calculated as29
However, disruption of fibrils by B1 and C refrained them from pore formation on the LUVs significantly, as the increment of corresponding fluorescence intensity (sample 4, magenta, and sample 5 olive, Fig. 5(b and c)) was as low as that of the untreated LUVs (sample 1, black, Fig. 5(b and c)). Hence, these results indicate that the disrupted fibrillar assembly of hIAPP by B1 and C does not damage the cell membrane, which may be related to their non-toxicity.
In the DLS study, the size distribution in terms of hydrodynamic diameter (d) of hIAPP particle was observed to change in ascending order from 1 nm to several micrometers, broadening the d value. At 0 h, hIAPP exhibited a size distribution at around 1 nm, which indicated the formation of monomeric species (Fig. 5(a)-(i)). Oligomeric intermediates with size distribution centered at ∼100 nm (Fig. 5(a)-(ii)) was noted after 6 h. Further incubation up to 48 h resulted in the gradual increase in the size of hIAPP particles ranging from 100–10000 nm (Fig. 5(a)-(iii–v)). In contrast, when hIAPP was incubated in the presence of B1, the fibrillization process was modulated, as in 0 h, two types of species, one having size distribution centered at 100–1000 nm and the other with a reduced size distribution at around 1 nm were observed instantly (Fig. 5(b)-(i)). The appearance of a size distribution at 100–1000 nm indicates that hIAPP assembles with B1 to generate some specific aggregated species, altering the native aggregation pathway of hIAPP. As B1 restricted the native fibrillation of hIAPP, the initial aggregated species gradually reduced in size, evident from the substantial change in the size distribution with time (Fig. 5(b)-(ii–v)).
Further, a transitional growth of hIAPP from smaller to longer fibrils, finally leading to a more extensive fibrillar network, was observed under TEM. At 0h of incubation, hIAPP existed entirely in dot-like form. It slowly developed intermediate smaller fibrillar species at 6–12 h, followed by the formation of the mature fibrillar network after 24 h (Fig. 5(c)-(i–v)). On the contrary, in the presence of B1, the entire process of hIAPP fibrillization was modulated. Some aggregated species of altered morphology along with a minute amount of fibrils were initially generated upon binding of hIAPP to B1. The fibrillar assembly was not observed anymore after six hours, gradually transforming the non-fibrillar aggregates into smaller dot-like species (Fig. 5(d)-(i–v)).
In the dye leakage assay, hIAPP in the absence and presence of B1 were incubated for 0, 1, 6, 12, 24, 36, and 48 h in PBS and then added to LUVs to obtain a final concentration of 50 μM of the lipids. We observed the most significant leakage of carboxyfluorescein dye from the 6h incubated hIAPP sample (red, Fig. 6). However, the leakage decreased gradually upon increasing the incubation time of hIAPP from 12 to 48 h, indicating the higher toxicity arising from the oligomeric species generated at around 6 h and lower toxicity of the mature fibrils. In contrast, the dye leakage by hIAPP co-incubated with B1 (blue, Fig. 6) was not significant as the corresponding fluorescence intensity appeared as low as the untreated LUVs (black, Fig. 6).
Therefore, from the observation of size distribution and morphology, it can be inferred that binding of the breaker peptide (B1) with hIAPP triggers the generation of non-fibrillar aggregates, which prevents the growth of hIAPP oligomers associated with amyloid fibrils.30,31 Also, it was observed in the dye leakage assay that the aggregates that appeared in the presence of B1 were not membrane damaging, thus probably non-toxic. These results are in strong correlation with the results obtained from the other biophysical studies of inhibition and disruption. Hence effective binding of the breaker peptide, B1, might influence the anti-amyloidogenic processing of hIAPP; however, further investigations are required to understand the mechanism correctly.
Here, we have demonstrated the superiority of Ant mutated hIAPP8-37 at different positions (B1 and B2) over its smaller Ant mutant variant (C) for modulating aggregation of hIAPP. The designed peptidomimetics, B1, B2, and C were non-amyloidogenic. However, polypeptide A, which did not contain any Ant, showed amyloid aggregates similar to hIAPP in the same condition. Also, B1 and B2 were better inhibitors of amyloid formation than C and worked at a lower concentration. However, the position of mutation (G24X and I26X, X = Ant) did not change their ability to inhibit amyloidogenesis significantly.
Moreover, both B1 and C disrupted preformed amyloid, and B1 was more effective than C. Such amyloid disruption did not result in toxic smaller oligomers, as evident from LUV studies. Systematic DLS and TEM studies revealed that the aggregation pathway of hIAPP gets altered by B1, and oligomers thus generated do not rupture lipid membrane significantly. Most importantly, a gradual decrease in aggregate size was noted. MTT assay revealed that Ant-containing peptidomimetics are non-toxic to RIN-5F cells and rescue from hIAPP mediated toxicity. The improved efficiency of the longer peptidomimetics (B1 and B2) may be due to their tight binding and more effective sequence recognition to the fibrillar species generated by the wild-type hIAPP.7,13 These peptidomimetics are highly efficient aggregation inhibitors and disrupt the preformed amyloid at fewer molar ratios. Therefore, these peptidomimetics can be a lead scaffold for therapeutic design towards T2D. Most importantly, these results give hope that an Ant mutant of hIAPP may work similarly to Pramlintide; of course, further studies are required to confirm it.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/d0cb00178c |
This journal is © The Royal Society of Chemistry 2021 |