{"paper_id":"be724dab-0ec7-4be6-b243-4515ebc31d67","body_text":"The gonadotropin releasing hormone (GnRH)\nantagonists are short\npeptide analogues of GnRH, which bind and block the action of GnRH\nreceptors directly, with a rapid decrease in luteinizing hormone,\nfollicle-stimulating hormone, and testosterone in men and estradiol\nin women. \n − \n \n \n  This is therapeutically valuable for conditions that are controlled\nby sex hormone secretion, including prostate cancer, benign prostatic\nhyperplasia, and endometriosis. \n − \n \n  Teverelix (Tv) is a decapeptide\nGnRH antagonist that has been developed as an effective treatment\nfor prostate cancer, which is currently under phase III clinical trials. \n , , ,\nTeverelix is formulated as a trifluoroacetic acid (TFA) salt, which,\nat the high concentrations used for injection, forms a microcrystalline\nsuspension. This unusual behavior has only been observed for a few\npeptides and sometimes only with specific counterions. For example,\nthe acetate salt of teverelix does not form a microcrystalline state.  For other systems like insulin, the microcrystalline\nsuspensions have already been used as a formulation method. \n − \n \n  This state is essential for the effective formulation and use of\nteverelix as it enables a solution of the drug at high concentration\nin a small volume to be injected into a patient. \n , , ,\nAfter\nsubcutaneous injection, teverelix exhibits a biphasic, long-acting\nrelease profile that is attributed to two processes. It is known that\nthere is rapid aggregation of teverelix at the site of injection resulting\nin the formation of a depot, which acts as a slow-release mechanism. \n , , , \n  In\naddition, some of the teverelix is absorbed quickly into the bloodstream,\neither because it is a state that can be rapidly transported into\nthe circulation or it simply has not yet had the time to aggregate\nalong with the rest of the teverelix. In either case, this leads to\nthe initial rapid release of some of the teverelix, followed by the\nslow release of more teverelix from the depot. \n , , , \n  As other GnRH\nantagonists have been shown to form amyloid-like fibrils under specific\nconditions, it is thought, but not known, that this depot may be amyloid\nin nature. \n −\nMany peptides, including a significant number of therapeutic\npeptides,\nare known to aggregate when formulated at high concentrations, this\nbeing one of the most common and troubling processes encountered in\nalmost all phases of biological drug development. \n − \n \n  Aggregates\ncan be amorphous or highly structured, for example, amyloid fibrils. \n − \n \n  In some cases, it has been shown that the reduction of physical\nstability of peptide/protein-based therapeutics not only leads to\nloss of activity, but also causes other, potentially severe, problems\nsuch as toxicity and immunogenicity. \n − \n \n  However, in other cases,\naggregation into amyloid fibrils in particular may have certain advantages.\nFor example, the use of amyloids in a pharmaceutical application for\nthe formulation of long-acting drugs (through the formation of an\namyloid depot) has been explored.  In\nthe case of teverelix, understanding the formation of an amyloid-like\nslow-release depot, which is formed in vivo, is essential for optimizing\nits therapeutic use and efficacy. Despite this, there have been no\ndetailed studies on the self-assembly of teverelix or the factors\nthat govern which physical state dominates.  The single published study simply determined the critical aggregation\nconcentration of teverelix under one condition.  Until now, the focus has been on its pharmacological behavior. \n −\nHere, we report a comprehensive study of the self-assembly\nof teverelix\ninto a fibrillar state at concentrations between 0.05 and 10 mg/mL\nusing multiple different biophysical approaches. N.B. We do not study\nthe formation of the microcrystalline state here, which requires much\nhigher peptide concentrations. We show that the peptide does form\namyloid fibrils under a wide range of conditions, despite its unusual\nsequence, which contains both  L - and  D -amino acids\nas well as unnatural and, in some cases, large side chains. Kinetic\nstudies were employed to understand the rate and mechanism by which\nteverelix forms amyloid fibrils, and transmission electron microscopy\nexperiments provided detailed structural information about the morphology\nof fibrils and higher-order structures formed.\nIn this study,\na range of spectroscopic methods were employed to\nprobe the aggregation of teverelix, including thioflavin T (ThT) assays.\nThT is capable of binding to the cross-β sheet structure of\namyloid fibrils, and, in this case, an increase in its fluorescence\nintensity at around 480 nm is observed; therefore, it has been widely\nused as a method to monitor fibrillation processes.  However, it should be noted that there are a considerable\nnumber of cases where an increase in ThT fluorescence is not due to\nthe formation of fibrils, so additional methods to verify fibril formation\nare essential.  X-ray fiber diffraction\n(XRD) was used to determine whether the Tv fibrils are amyloid or\nnot, \n , \n  and the fibrils were further characterized\nusing transmission electron microscopy (TEM) and various spectroscopic\ntechniques. In addition, the starting state (freshly prepared solutions\nof teverelix) was investigated using different experimental methods\nand shown to be dimers with similar secondary structure to the fibrils,\nexplaining the rapid fibril formation observed under some conditions.\nOur studies do not start with the normal design of buffer experiment\nas teverelix has already undergone several clinical trials. Instead,\nand somewhat unusually, we start by studying the self-assembly of\nteverelix without buffer, as this is how it is currently formulated\nin clinical trials. This has the advantage of (i) better mimicking\nthe formulation conditions and, importantly, (ii) enabling us to detect\nwhether there are any protonation/deprotonation processes associated\nwith fibril formation. Subsequently, buffered conditions were employed\nto gain further understanding of how Tv concentration, pH, and TFA\nand NaCl concentrations affected the rate of self-assembly.\n\nTeverelix (Tv), Ac- D -Nal­(2)- d -Phe­(4Cl)- D -Pal­(3)-Ser-Tyr- D -Hcit-Leu-Lys­(iPr)-Pro- d -Ala-NH 2  in the form of a TFA salt (white powder),\nwas supplied by\nAntev (Antev Ltd., UK) with a purity of 93.2% and a molecular weight\nof 1460 kDa. Tv was stored in a −20 °C freezer and used\nwithout further purification.\n\nTv powder was dissolved\nin the appropriate buffer in a 1.5 mL Eppendorf tube (Eppendorf International,\nGermany). To aid the dissolution of Tv, the samples were transferred\ninto a Thermomixer compact (Eppendorf International, Germany) with\n300 rpm agitation at 37 °C for 30 s. The concentration of the\nTv solution was determined spectroscopically on a Cary 60 UV–vis\nspectrophotometer (Agilent Technologies, USA) by using the Beer–Lambert\nLaw and a calculated extinction coefficient of 5426 M –1  cm –1  at 280 nm. The Tv samples were sealed with\nparaffin film (Fisher Scientific, USA) and covered with aluminum foil\nto prevent solvent evaporation and exposure to sunlight. Incubation\nwas performed at room temperature without further agitation, unless\nin the ThT assays or as stated otherwise.\nThe intrinsic fluorescence measurements\nwere performed on a Cary Eclipse fluorescence spectrophotometer (Agilent\nTechnologies, USA). Spectra were acquired from 300 to 400 nm using\nan excitation wavelength of 280 nm and a wavelength step of 1.0 nm.\nBoth excitation and emission bandpasses were kept at 10 nm with an\nappropriate voltage on the photomultiplier tube ranging from 550 to\n750 V. All samples were measured in a 120 μL quartz cuvette\n(Hellma Analytics, Germany) at 25 °C.\nThe circular dichroism measurements\nwere performed on a Chirascan CD spectrophotometer (Applied Photophysics,\nUK). All samples were measured with a 1 nm wavelength step and 1 nm\nspectral bandwidth at 25 °C. The far-UV CD spectra were acquired\nfrom 190 to 250 nm with samples in a 0.1 mm (or 0.01 mm) path length\ncuvette, while near-UV CD spectra were acquired from 250 to 350 nm\nwith samples in a 0.2 mm path length cuvette. The result for each\nmeasurement was obtained by averaging three scans, followed by the\nsubtraction of the buffer background.\n2.5 μL\nof Tv samples with appropriate concentrations and conditions was spotted\nonto a glow-discharged, carbon-coated copper grid (Agar Scientific,\nUK) for 30 s. The glow discharge was performed on a Quorum Technologies\n(UK) GloQube system prior to the sample application. The excess sample\nsolution was removed by blotting the edge of the grid with a filter\npaper. The sample was further stained by loading 2.5 μL of 2%\n(w/w) aqueous uranyl acetate solution onto the grid for 1 min followed\nby the removal of excess staining fluid with a filter paper. 2.5 μL\nof water was then used to wash the grid to remove the salts presented,\nand the grid was dried in air. TEM analysis was performed using a\nThermo Scientific (USA) Talos F200X G2 transmission electron microscope\nwith an acceleration voltage of 200 kV.\nThioflavin T assays\nwere performed on a microplate reader FLUOstar Omega (BMG Labtech,\nGermany). Samples were transferred into a 96-well half area plate\n(Corning 3881, USA) and sealed with tape (Costar Thermowell) to prevent\nthe evaporation of samples. Samples were incubated in the plate reader\nat 37 °C with a final concentration of ThT of 50 μM. ThT\nfluorescence was measured using an excitation filter at 440 nm and\nan emission filter at 480 nm. Bottom reading of the plate was performed\nevery 30 min with 5 min shaking (orbital shaker at 600 rpm) prior\nto each measurement. Fluorescence was measured at a gain of 500 with\n8 flashes per well.\nThe results from ThT\nbinding assays were fitted with the following equation: \n y = y 0 + A 1 + exp ( − k ( t − t 1 / 2 ) ) + b · t \n 1 \n where  y \n 0  is the initial fluorescence,  A  is the amplitude\nof the transition,  t \n 1/2  is the half-time\n(the time at which the ThT fluorescence reaches half of the final\nplateau value),  k  is the apparent growth rate, and  b  is the slope of the final baseline. The lag time was calculated\nusing the parameters obtained from the best fit of the data to  eq  \n , using the following equation: \n t lag = t 1 / 2 − 2 k \n 2\nThT fluorescence was measured\non a Cary Eclipse fluorescence spectrophotometer (Agilent Technologies,\nUSA). Spectra were acquired using an excitation wavelength of 448\nnm, and emission was recorded from 460 to 600 nm with a wavelength\nstep of 1 nm. The final concentration of ThT was kept at 50 μM,\nwhich was the same as that used in the ThT assays. Both excitation\nand emission bandpasses were kept at 10 nm with an appropriate voltage\non the photomultiplier tube ranging from 550 to 750 V.\nAnalytical size-exclusion\nchromatography (SEC) was performed using an AKTA/FPLC system (GE Healthcare,\nUSA) with a Superose 12 10/300 column (GE Healthcare, USA). A 200\nμL injection loop was used for sample loading. All samples were\nfiltered through 0.22 μm filters before being loaded onto the\ncolumn. All samples were eluted with a flow rate of 0.75 mL min –1  at room temperature and an upper pressure limit of\n3 MPa. The elution process for each run was monitored by using a UV\nabsorbance detector at 280 nm through a 0.5 cm flow cell.\nTv samples (5 mg/mL) were prepared\nin 25 mM citrate buffer (pH 3.0) and incubated at room temperature\nfor 1 week without shaking. For comparison, samples were incubated\nat room temperature without shaking and in an incubator at 37 °C\nwith shaking for a week. Tv fibrils were prepared by hanging 10 μL\ndroplets of Tv between two wax-tipped capillaries, followed by overnight\ndrying (also known as the alignment procedure). The dried Tv fibrils\n( Figure S1 ) were transferred into a diffractometer\n(Malvern Panalytical, UK) with care to avoid breakage of the fibril\nbundle samples. The recorded X-ray diffraction patterns were converted\nto the JPEG format and subsequently analyzed by X-ray fiber diffraction\nanalysis program CLEARER  to extract the\nvalues of the equatorial and meridional reflections.\n\nTeverelix (Tv) is a synthetic\npeptide GnRH antagonist, featuring\nan unusual primary structure. It has several large, unnatural side\nchains (Nal, Cpa, Pal, hCit, and Lys­(iPr)) as well as being comprised\nof both  L - and  D -amino acids ( Figure  \n A). Here, we show the results of studies\non the self-assembly of Tv over a range of peptide concentrations\nfrom 0.05 to 10 mg/mL and from pH 2.7 to 5.0. In some cases, the peptide\nsolutions contained no buffer (to mimic the conditions currently used\nin the drug formulation and to probe changes in pH during self-assembly).\nIn other cases, an appropriate buffer was used to maintain the pH\nof the solution throughout self-assembly.\nChemical structure of\nteverelix and images of the fibrillar states\nit adopts. (A) Chemical structure of teverelix. (B) TEM image of the\nfibrils formed by a 2 mg/mL solution of Tv in dd H 2 O incubated\nfor 7 days at 37 °C. A typical narrow filament is shown in the\nyellow rectangle (diameter 10 nm), and a wide filament (diameter 20\nnm) assembled from two narrow filaments is shown in the red rectangle.\n(C) TEM image of a network of Tv fibrils in a gel-like sample. One\nmg/mL Tv in 25 mM citrate incubated for 14 days at 37 °C was\nused (pH 3.0). (D) Alignment of Tv fibrils as observed in a 3 mg/mL\nsolution of Tv in ddH 2 O incubated for 14 days at 37 °C\n(pH 2.73). (E) Inverted microcentrifuge tube showing the gel-like\nsample of Tv formed by 10 mg/mL Tv in 25 mM citrate incubated for\n14 days at 37 °C (pH 3.0).\nIn our initial experiments,\nsolutions of Tv were found to form fibrils over a range of different\npeptide concentrations from 1 to 3 mg/mL in both unbuffered and buffered\nsolutions from pH 2.73 to 3.5 after incubation for 1–2 weeks.  Figure  \n B–D shows\ntransmission electron microscopy (TEM) images of the fibrils formed.\nIn these cases, the samples were incubated at room temperature under\nquiescent conditions; the process did not, therefore, require elevated\ntemperatures, pressures, or agitation. For solutions of 0.2 mg/mL\nTv at pH 4.1 and 1.0 mg/mL Tv at pH 3, fibrils were observed directly\nin freshly prepared samples after approximately only 30 min of incubation\nat room temperature,  Figure S1A,B , establishing\nthat fibril formation can be rapid.\nTwo types of fibrils with\ndifferent widths were observed by TEM (shown in the yellow and red\nboxes in  Figure  \n B).\nUnder conditions used for most of the X-ray fiber diffraction and\nTEM microscopy experiments, samples were incubated at room temperature\nunder quiescent conditions. Narrow filaments of Tv were the major\nspecies observed (width 10 nm), while occasionally, larger wide Tv\nfilaments were also observed (width 20 nm), which appeared to be twisted\npairs of narrow filaments. Both were twisted into a helix with a repeat\napproximately 100 nm. Thus, the dimensions of the fibrils of teverelix\nare consistent with those observed for other amyloid fibrils. \n , \n  Interestingly, wide filaments were the dominant species when samples\nof Tv were incubated at 37 °C with constant agitation ( Figure S1C,D ). However, under the conditions\nused in all of the experiments described here, including the kinetic\nexperiments in which there was periodic but not constant agitation,\nnarrow filaments were the dominant species. In the rest of the paper,\nthe narrow filaments will be referred to as fibrils. All of the fibrils\nobserved within these studies had the same overall structure and morphology\nin terms of widths and twist.\nAt sufficiently high Tv concentrations\nand over time, the physical state of Tv samples changed from a nonviscous,\nsolution-like state to a viscous gel-like state as self-assembly occurred\n( Figure  \n E). In the\ngel-like state (formed at higher Tv concentrations with long incubation\ntimes), TEM was used to image the sample, which showed that a high\nconcentration of fibrils formed a random network in which fibrils\ncross over each other ( Figure  \n C), like many other fibril-forming peptides.  Under different conditions (moderate Tv concentration with\nincubation times of approximately 2 weeks), the fibrils formed by\nTv were found to strongly align forming a much less heterogeneous\nstate ( Figure  \n D).\nAlthough there was a pipetting step in the preparation of the TEM\ngrid, the alignment of Tv fibrils did not appear to need any additional\nexternal force, e.g., flow of the solvent as is the case for a number\nof other peptides. \n − \n \n  It should be noted that the alignment was not observed\nin every sample, even though every sample prepared for TEM had a pipetting\nstep, suggesting that this result is not an artifact of the pipetting\nstep but that Tv fibrils align in solution only under a specific set\nof conditions.\nThe TEM results show that Tv self-assembles into\nfibrils over a wide range of conditions and that the fibrils formed\nhave dimensions like those observed in amyloid fibrils.  They also establish that fibrils of Tv can adopt\ndifferent higher-order structures, forming either a network of criss-crossed\nfilaments or fibrils that are highly aligned. Despite the power of\nTEM to image the fibrils of Tv, these results do not prove that the\nfibrillar state of Tv is an amyloid.\nAmyloid fibrils are well-known to display a cross-β\ndiffraction pattern by X-ray fiber diffraction. \n , , \n  A sample of Tv fibrils was aligned to form\na bundle of fibrils, and an X-ray diffraction pattern was collected\n( Figure  \n A). The well-oriented\ndiffraction pattern shows a strong 4.8 Å meridional reflection\narising from the hydrogen bonding distance between β-strands\nin a β-sheet. Several weak reflections were observed on the\nequator, as well as a strong, sharp 15.3 Å equatorial reflection,\nwhich was interpreted to arise from the spacing between the β-sheets.\nFrom these values, a schematic showing arrangement of the peptide\nand the wide β-sheet spacing was constructed and is shown in  Figure  \n B.\nX-ray fiber diffraction\npattern and model of β-sheet spacing.\n(A) X-ray fiber diffraction pattern of 5 mg/mL solution of Tv in 25\nmM citrate incubated for 7 days at 37 °C without agitation (pH\n3.0). The meridional (M) and equatorial (E) reflections extracted\nare shown in red, which were 4.8 and 15.8 Å, respectively. (B)\nModel of the β-sheet spacing of the typical cross-β structure\nfor amyloid. The interstrand and intersheet spacings are indicated\n(4.8 and 15.8 Å) based on the values obtained in A.\nThe diffraction pattern is consistent with cross-β\nconformation,\nalthough the equatorial reflection is larger than is typical (∼12.0\nÅ).  The sharp reflections in both\nthe meridional and equatorial axes suggest a highly ordered structure\nfor the peptide within the fibrils. Since there are several bulky\nside chains in the Tv structure ( Figure  \n A), which will impact the stacking of the\nβ-sheets within the fibril, the difference in equatorial reflection\nis reasonable, and we conclude that Tv forms amyloid-like fibrils.\nInitial experiments on the self-assembly of Tv focused on the behavior\nof unbuffered solutions of the peptide, as this more closely mimics\nthe current formulation condition of the peptide. In these cases,\nthe acidic counterion TFA content largely determines the pH of the\nsolution, and therefore, it is expected that the pH varies with the\npeptide concentration. Using unbuffered solutions enabled experiments\nto determine if there is any change in the protonation state of the\npeptide during self-assembly. The pH of the Tv samples was measured\nboth prior to, and after, aggregation, both in Tv samples made up\nin ddH 2 O as well as those dissolved in a buffer to control\nthe pH (to ensure that buffer concentrations were sufficient to maintain\nthe pH throughout the self-assembly process when buffers were used).\nThe results are shown in  Figure  \n A and  Table S1 .\npH decrease\nof Tv upon incubation and estimated critical aggregation\n(cac) from pH 3.0 to 5.0. (A) pH values of freshly prepared and seven-day\nincubated Tv samples from 0.2 to 5 mg/mL in ddH 2 O at room\ntemperature. Due to the high viscosity of gel-like samples formed,\nthe pH values of 4 and 5 mg/mL Tv in ddH 2 O incubated for\n7 days were not measured. (B) Estimated critical aggregation concentration\nof freshly prepared Tv in 25 mM citrate at 25 °C (pH from 3.0\nto 5.0). A series of Tv (0.05–1 mg/mL) was freshly prepared\nin 25 mM citrate and measured in order to obtain the corresponding\ncac values (pH 3.0–5.0). For more details, see  Figures S2–S4 .\nFor the nonbuffered samples, the initial pH recorded\nimmediately\nafter preparation of the sample was found to vary with the concentration\nof the peptide as expected as the TFA counterion is acidic. In addition,\nthe pH was observed to decrease significantly after self-assembly,\nas shown in  Figure  \n A. These results indicate that a deprotonation step is associated\nwith the formation of the Tv amyloid fibrils. Calculations establish\nthat a single deprotonation event occurs (see the footnote to  Table S1 ). Considering the p K \n a  of all possible protonation sites in Tv, we think that\nthe pyridine side chain, which has a p K \n a  of 5.22, is the most likely one to be involved; however, we cannot\ncompletely rule out the possibility that it is another side chain\nwith a perturbed p K \n a  value. At the low\npH values used in this study, the pyridine side chain will be in a\nprotonated form (pyridinium ion) and will carry a positive charge\nat the start of the reaction. The results show that deprotonation\n(and elimination of the positive charge) of the pyridinium group is\nessential for the self-assembly process to occur. Thus, it is highly\nlikely that the pyridine side chain is buried in the fibrillar state,\npotentially acting as a H-bond acceptor. These results suggest that\nthe self-assembly of Tv is highly likely to be pH-dependent.\nDeprotonation/protonation events on fibril formation have been\nobserved before in studies on other peptides including glucagon.  In this case, depending on whether the peptide\nis in acidic or alkaline conditions, it either protonates or deprotonates\nto have a net charge of zero in the fibrillar state. Thus, this study\nestablished that the p K \n a  of side chains\ncan be dramatically different in the fibrillar state compared to in\nsolution.\nTo establish whether the self-assembly of Tv is pH-dependent, the\ncritical aggregation concentration (cac) of Tv was estimated by using\nThT fluorescence. Although the ThT dye can bind to a number of non-amyloid-like\nstates, in cases where amyloid formation has been shown by other methods,\ne.g., X-ray fiber diffraction, ThT is a more reliable measure of the\nextent of amyloid fibril formation and has been used by many groups\nas a measure of the amount of amyloid fibrils in solution.  ThT fluorescence of a series of Tv samples (0.01–1\nmg/mL) in buffered solutions over a range of pH values from 3.0 to\n5.0 were measured ( Figures S2–S4 ). ThT dye was added to each sample prior to the fluorescence measurement,\nand the cac was estimated from the data. It should be noted that the\ncac values are estimates, as the system is not in equilibrium. They\ncan be considered the cac values after 1 h of incubation at room temperature.\nOver longer periods of time, the starting state will further convert\ninto fibrils and the apparent cac will therefore decrease over time.\nDespite this, the values still give a clear indication of how fibril\nformation depends on pH. The data show that the cac values decrease\nas the pH increases from 3 to 4.5, reaching a plateau between pH 4.5\nand 5.0 ( Figure  \n B),\nsuggesting that the fibrillar form is more stable at higher pH values.\nThese results are consistent with the decrease in pH observed on fibril\nformation as discussed above, and the pyridinium side chain being\ndeprotonated in the fibrillar state. Given that the stability of the\nfibrils depends critically on pH, it is highly likely that the kinetics\nof fibril formation will also be pH-dependent.\nThe kinetics\nof fibril formation for Tv were investigated using ThT assays over\na wide range of conditions and pH values using buffered solutions.\nIt is well-known in the literature that an increase in ThT fluorescence\ncan occur even in the absence of amyloid-like fibrillar species. Therefore,\nfor many of our kinetic runs, TEM imaging was employed at the end\nof the ThT assay to confirm the presence of fibrillar species ( Figure S5 ), thus establishing that ThT is a reliable\nindicator of fibril formation.\nAt low Tv concentrations and/or\nlow pH (pH 3.0–4.0), sigmoidal growth curves showing a lag\nand growth phase followed by a plateau were observed ( Figure  \n A,B). Where possible, the ThT\ncurves were fit to extract the key kinetic parametersthe lag\ntime ( t \n lag ),  t \n 1/2,  the time at which the ThT signal is 50% of its final value, and  k , the apparent growth rate, which corresponds to the steepest\npart of the growth phase. The effects of peptide concentration, pH,\nTFA concentration, and mannitol (which is present in the formulation\nof Tv) were all investigated to obtain information about the factors\naffecting the rate of fibril formation as well as mechanistic insight.\nKinetic\ndata from ThT assays monitoring the formation of amyloid-like\nfibrils by teverelix under different conditions. (A) ThT assay of\n0.2–3 mg/mL Tv in 25 mM citrate at pH 3.7 at 37 °C. (B)\nThT assay of 0.1 mg/mL Tv in 25 mM citrate pH from 3.0 to 5.0 at 37\n°C. The fluorescence data has been normalized, see  Materials  and  Methods  for further\ndetails. No half-life time ( t \n 1/2 ), lag\ntime ( t \n lag ) and apparent growth rate ( k ) values were calculated at pH 3.0. For pH 4.0, these values\nwere 748 min ( t \n 1/2 ), 308 min ( t \n lag ) and 0.0046 ( k ). For pH\n5.0, these values were 652 min ( t \n 1/2 ),\n64 min ( t \n lag ) and 0.0034 ( k ). (C) ThT assay of 0.5 mg/mL Tv in 100 mM citrate pH 3.0 with 0\nto 11.0 mM TFA added at 37 °C. The fluorescence data has been\nnormalized, see  Materials  and  Methods  for further details. (D) ThT assay (normalized) of\n0.5 mg/mL Tv in 100 mM citrate pH 3.0 with 0–11 mM NaCl at\n37 °C. (E) Normalized half-life ( t \n 1/2 ) versus [TFA] or [NaCl]. (F) Normalized apparent growth rate ( k ) versus [TFA] or [NaCl]. The kinetic parameters  t \n 1/2  and  k  were obtained from\nthe best fit of the triplicate ThT assays shown in (C) and (D) to  eqs  \n  and  . The error bars in parts (E) and (F) are the standard deviations\nbased on triplicate data.\nThe ThT assays of a range of Tv concentrations from 0.2 to 3 mg/mL\nat pH 3.7, which were acquired in a 96-well plate in a fluorescence\nplate reader, are shown in  Figure  \n A. For the lowest concentration of Tv studied, 0.2\nmg/mL (red), typical sigmoidal kinetics consistent with a nucleation–polymerization\nmechanism were observed. At pH 3.7, all higher peptide concentrations\nhad a nonzero initial ThT intensity and no lag phase was observed,\nindicating that fibrils had already formed in the solution before\nthe ThT plate reader assay was started. N.B. it took approximately\n30 min to prepare the samples and load them onto the plate reader\nbefore the program was initiated and detection began. Thus, in these\n30 min, for most Tv concentrations at pH 3.7, fibril formation is\nrapid, and the lag phase is over within the 30 min deadtime of the\nexperiment and thus is not observed ( Figure  \n A). Notably, the initial and final ThT intensities\nare proportional to the Tv concentration, suggesting that ThT fluorescence\nintensity is a good measure of fibril formation and that the equilibrium\nlies well over toward that of the fibril (i.e., there is relatively\nlittle of the starting state (be it a monomer/dimer) left in solution\nat the end of the reaction). Additionally, from these data, it can\nbe observed that the higher the Tv concentration, the shorter the\ntime required to reach the final plateau, i.e., shorter  t \n 1/2 . Though it is difficult to get the actual  t \n 1/2 ,  t \n lag , and  k  values from fitting data at higher Tv concentrations due\nto the nonsigmoidal curves, it is clear from the data obtained that\nTv concentration is an important factor in determining the rate of\nfibril formation as expected for a peptide that forms amyloid fibrils\nthrough a nucleation–polymerization mechanism.\nIn addition,\nto check that Tv fibrils formed following a nucleation–polymerization\nmechanism at higher peptide concentrations, additional experiments\nwere undertaken monitoring ThT fluorescence in a single cuvette in\na standard fluorimeter. This enabled us to have a shorter dead time,\nas only one sample needed to be prepared at one time. Two higher concentrations\nof Tv, 4 and 5 mg/mL, were used, and, in these cases, lag phase and\nsigmoidal kinetics were observed ( Figure S6 ), consistent with a nucleation–polymerization mechanism.\nUnfortunately, as the agitation used in the plate reader assays could\nnot be reproduced in the fluorimeter, the results could not be compared\ndirectly to those obtained above.\nThe kinetics of fibril formation were investigated at pH 3.0, 4.0,\nand 5.0 at a peptide concentration 0.1 mg/mL Tv using citrate buffer\nto maintain the pH throughout the reaction. The results are shown\nin  Figure  \n B. No ThT\nsignals were observed at pH 3.0 at any peptide concentration, suggesting\nno, or very few, fibrils form at this low pH. In contrast, at pH 4.0\nand 5.0, a sigmoidal increase in ThT fluorescence was observed and\nthe data were fit to  eqs  \n  and   to obtain the kinetic parameters  t \n 1/2 ,  t \n lag , and  k , which are 12.5 ± 0.2 h, 5.1 ± 0.3 h, and 0.0046\n± 0.0001, respectively, at pH 4.0, and 10.9 ± 0.4 h, 1.1\n± 0.5 h, and 0.0034 ± 0.0002, respectively, at pH 5.0. These\ndata establish that the rate of fibril formation is also pH-dependent\nand faster at higher pH values, consistent with the cac measurements\ndescribed above. These results suggest that the pyridinium side chain\nin Tv must be deprotonated for the formation of critical species on\nthe pathway to forming fibrils, e.g., an aggregate that can act as\na nucleus, as well as in the fibril itself.\nThe teverelix used was the TFA salt,\nand it is known that the molar ratio of TFA to Tv is approximately\n2.1 or 2.2 to 1 in its lyophilized state.  It is also known that only the TFA salt forms a microcrystalline\nstate at high concentrations.  Thus, counterion\nTFA may also play a role in the self-assembly of Tv. The effect of\ndifferent TFA concentrations on the rate of fibril formation was investigated\n( Figure  \n C).  t \n 1/2  values decrease and  k  increase\nwith increasing TFA concentration, suggesting that TFA plays a role\nin fibril formation. However, as increasing the TFA concentration\naffects the ionic strength of the solution, a series of control experiments\nusing NaCl were also undertaken to establish whether ionic strength\nalso affected the rate of fibril formation ( Figure  \n D). These data showed a similar trend to\nTFA; however, the effects of NaCl were smaller than for TFA. The results\nfor NaCl agree with other studies, which have shown that ionic strength\nfrequently impacts the rate of fibril formation. \n − \n \n  To compare\nthe effects of TFA and NaCl quantitatively, the kinetic parameters\nobtained from the best fit of the ThT assays shown in  Figure  \n C,D to  eqs  \n  and   were normalized,\nand the results for TFA and NaCl are shown together in  Figure  \n E,F. It is clear that TFA and\nNaCl both result in the same overall trends, but the effect of TFA\nis considerably larger than that of NaCl. These results suggest a\nspecific role of TFA in stabilizing key species on the aggregation\npathway such as oligomers, nucleus, and/or fibrils. The fact that\nthe TFA concentration also affects the apparent growth rate indicates\nthat Tv may add to an elongating fibril as the TFA salt.\nHaving established the conditions\nunder which Tv forms fibrils very slowly, i.e., low Tv concentrations\nand low pH (<4.0), experiments were undertaken under these conditions\nto characterize the starting state of the peptide in freshly prepared\nsolutions before any significant self-assembly had occurred. These\nsamples were then characterized using size-exclusion chromatography\n(SEC) and far- and near-UV CD.\nSize-exclusion chromatography was used to determine the oligomeric\nspecies populated in freshly prepared solutions under conditions where\nself-assembly into fibrils is slow, i.e, low peptide concentrations\n(1.0 mg/mL) at pH 3.0 ( Figure  \n A). Two peaks were identified with elution volumes at 17.4\nand 19.9 mL, and based on the calibration curve shown in  Figure S7 , the major peak eluting at 19.9 mL\n(fraction A) is from what is likely to be a dimer, while the minor\npeak eluting at 17.4 mL (fraction B) is from what is estimated to\nbe a pentamer. N.B. note that as the calibration curve was calculated\nwith globular proteins and Tv is a peptide with what is likely to\nbe an extended (β) conformation, the oligomeric states are only\napproximate. The pentamer may well be either a tetramer or hexamer,\ngiven that the smallest stable oligomer observed is highly likely\nto be a dimer. As expected, as the Tv concentration increases from\n0.5 to 1 mg/mL, the proportion of the dimer decreased, while the proportion\nof the pentamer increased ( Figure  \n A).\nSEC profile of freshly prepared Tv and far/near-UV CD\nresults.\n(A) Elution profile of Tv dimers and pentamers on a Superose 12 Increase\n10/300 column in 25 mM citrate at pH 3.0. Elution profiles of freshly\nprepared 0.5, 0.8, and 1 mg/mL Tv in 25 mM citrate at pH 3.0 are overlaid.\nAs the concentration of Tv increases, the intensity of dimers (19.9\nmL) decreases, indicating a conversion from dimers to larger species.\nThe intensity of the pentamer (17.4 mL) remains steady regardless\nof the Tv concentration. The percentage of each fraction is indicated\nin corresponding color. (B) Far-UV CD of freshly prepared 0.2, 0.4,\n0.6, 0.8, 1.0, 1.2, 1.4, 1.6, 1.8, and 2 mg/mL Tv with 25 mM citrate\nat pH 3.0. All samples were prepared and measured within 5 min. A\nshift in the single minimum from 228.5 to 234.5 nm was observed as\nthe concentration increased from 0.2 to 2 mg/mL. N.B. there was no\npeak shift from 0.2 to 0.8 mg/mL Tv. (C) Near-UV CD of freshly prepared\n0.2, 0.8, and 2 mg/mL Tv in 25 mM citrate at pH 3.0. All samples were\nprepared and measured within 5 min. A single minimum at 260 nm was\nobserved for all samples. (D) Far-UV CD of seven-day incubated 1 mg/mL\nTv with 25 mM citrate at pH 3.0, 4.0, and 5.0. A single minimum at\n232 nm (consistent with B) was observed for all samples. (E) Near-UV\nCD of seven-day incubated 1 mg/mL Tv with 25 mM citrate at pH 3.0,\n4.0, and 5.0. Single minima at 283.5 nm for pH 3.0 and 287 nm for\npH 4.0 and 5.0 were observed. In addition, a small shoulder at 322\nnm was observed for all samples. The data were normalized based on\nthe largest ellipticity value for each sample to make it easier to\ncompare.\nIn  Figure  \n B, the far-UV CD spectra of freshly prepared solutions\nof Tv at ten different peptide concentrations from 0.2 to 2 mg/mL\nat pH 3.0 are shown, and a single minimum shifting from 228.5 to 234.5\nnm was observed. This is consistent with the naphthalene side chain\nin Tv being in a fixed conformation and chiral environment in the\nsmall oligomers (dimers/pentamers) that are formed in freshly prepared\nsolutions at low pH. The shift in the minimum to higher wavelengths\non increasing the peptide concentration indicates that the environment\naround the naphthalene changes as the equilibrium position of the\nsystem moves away from dimers toward larger oligomeric forms, consistent\nwith the results from the SEC experiments.\nThe near-UV CD spectra\nfor three different Tv concentrations at pH 3.0 were also recorded\nat three different peptide concentrations ( Figure  \n C), and a single minimum at 260 nm, which\ndid not shift position over the concentration range 0.2–2 mg/mL,\nwas observed. In this case, the near-UV CD spectrum comprises possible\nsignals from the tyrosine, pyridine, phenylalanine, and naphthalene\nside chains. This result establishes that some of these aromatic side\nchains are already in a fixed, chiral environment in the freshly prepared\nsample (dimer/pentamer); however, it is not possible to say which.\nBoth far- and near-UV CD confirm that some of the aromatic side\nchains in Tv are fixed in a chiral environment in the starting state\nbefore any self-assembly has occurred, i.e, freshly prepared samples\nat pH 3.0. This is consistent with the SEC results, which show that\nthe predominant species in solution under these conditions are dimers\nand other small oligomers (possibly pentamers).\nFar- and near-UV CD were\nalso used to characterize the environment of the aromatic side chains\nof Tv in the fibrillar state. For this, three samples of 1 mg/mL Tv\nat pH 3.0, 4.0, and 5.0 were studied after a seven-day incubation\nat 25 °C under quiescent conditions. In the far-UV CD spectra,\na single minimum at 232 nm was observed in all cases ( Figure  \n D), which is between wavelengths\nof 228.5 and 234 nm, which were observed for the dimer and pentamer.\nThis result suggests that the environment of the naphthalene residue\nremains similarly buried in the fibril compared with the dimer/pentamer\nstarting state but may have a slightly different environment, although\nit is not possible to say exactly what is different. In contrast,\nthe near-UV CD spectra of Tv recorded after a seven-day incubation\nat pH 3.0, 4.0, and 5.0 ( Figure  \n E) were all different from the starting state observed\nat pH 3.0 ( Figure  \n C). N.B. near-UV CD spectra of the starting state at pH 4.0 and 5.0\ncould not be recorded as fibril formation is relatively fast at higher\npH values. Instead of a minimum at 260 nm, a major minimum at 290\nnm and a smaller minimum at 322 nm were observed ( Figure  \n E). These results indicate\nthat at least one of the tyrosine, pyridine, or phenylalanine side\nchains undergoes a conformational change/change in the environment\non fibril formation.\nFT-IR was also used to check for any significant\ndifferences in the secondary structure in the lyophilized Tv powder\n(created directly after its initial synthesis) and lyophilized Tv\nfibrils. Two major peaks were identified for both samples. The amide\nI band was at 1643 cm –1  and the amide II band at\n1535 cm –1  ( Figure S8 ),\nwhich suggests that the peptide chain is in an extended conformation\nand adopts a β-structure in both samples.\nThe results described above can be used to propose a model for the\nformation and structure of amyloid-like fibrils by Tv. We believe\nthat it is likely that the mechanism and structure of fibril formation\nwe have characterized in vitro here are like the process that occurs\nin vivo after injection of Tv. From studies at low Tv concentrations\nand low pH, we know that the peptide exists in solution in a largely\ndimeric form, which we propose is stabilized by the burial of some\nof the large hydrophobic side chains including the naphthalene side\nchain and possibly some of the other hydrophobic side chains and aromatic\ngroups ( Figure  \n A).\nThis dimer is in equilibrium with a slightly larger oligomeric state,\ngiven the apparent instability of the monomer, likely to be either\na tetramer or hexamer. In this state, the naphthalene side chain group\nhas changed its environment somewhat, as seen by the observed shift\nin the wavelength minimum in the far-UV CD. Most likely, it has undergone\nfurther burial. As the population of the tetramer/hexamer increases\nwith peptide concentration, as does the rate of fibril formation,\nwe think it likely that the tetramer/hexamer is on the pathway to\nfibril formation; however, without further extensive experiments,\nit is impossible to rule out other more complex pathways, e.g., involving\ndifferent on- and off-pathway oligomers.\nSchematic showing a putative\nstructure of Tv within a narrow amyloid-like\nfilament and a possible assembly pathway to the fibrillar state from\nthe dimeric starting point. (A) Putative structure of a Tv dimer within\na narrow amyloid-like filament. The double-arrow red dashed lines\nshow possible hydrogen bonding and π–π interactions\nbetween each monomer within the dimeric unit within the narrow filaments.\nIt is proposed that the deprotonated state of the pyridinium side\nchain hydrogen bonds with the tyrosine side chain, as inferred from\nthe pH dependence of fibril formation observed. (B) Potential model\nof the self-assembly of dimeric teverelix into narrow and wide filaments\n(both amyloid-like fibrillar states capable of binding ThT). The dimeric\nstate of Tv (in this state, the naphthalene side chain is already\nburied; however, the other aromatic side chains including the pyridinium\nand tyrosine are not fully buried). Dimers then form tetramers and,\nthrough association of another dimer, hexamers in which there has\nbeen some conformational change/burial of additional hydrophobic surface\narea. These then either form larger oligomers and/or undergo a conformational\nchange to form a critical nucleus, which then rapidly elongates by\naddition of a dimer of Tv into either the narrow filaments observed\nby TEM or wider filaments consisting of two narrow filaments (also\nobserved by TEM) if there is sufficient agitation. The nucleus likely\nelongates by the addition of a dimer of Tv. The blue curved arrow\nindicates that there is a deprotonation step during fibril formation,\nalthough it is not known exactly when this occurs.\nFibril formation occurs by the very well-established\nnucleation–polymerization\nas shown by the sigmoidal ThT kinetics obtained under some conditions\nas well as the fact that key kinetic parameters such as  t \n 1/2  and  t \n lag  clearly depend\nupon Tv concentration. It is not possible to say exactly what size\nor structure the critical nucleus is. However, given that fibril formation\nis known to be rapid after formation of a nucleus and that there are\nmeasurable populations of dimers and tetramers/hexamers in the starting\nstate it is possible that the nucleus is a larger oligomer or, alternatively,\nit could also be that the tetramer/hexamer has to undergo a relatively\nslow conformational change to form a state capable of rapid elongation\nand growth ( Figure  \n B). Again, given the relative instability of the monomer and stability\nof the dimer, it is likely that nuclei/fibrils elongate by addition\nof the dimer not the monomer (although we cannot rule out addition\nof the tetramer/hexamer). The results of the kinetic experiments using\ndifferent concentrations of TFA indicate that the nucleus must contain\nTFA molecules and that TFA is also involved in the elongation step,\nwhich we assume occurs by the addition of dimers of Tv closely associated\nwith TFA molecules.\nUnusually for peptides that form amyloid-like\nfibrils, in the absence\nof significant agitation, the narrow filaments of Tv are stable and\nwide filaments consisting of two or more narrow filaments only form\nif agitation is constant. We propose that this may be directly due\nto the fact that a dimer, not a monomer, is the starting state. Thus,\nwithin a narrow filament that consists of dimers, there is significant\nburial of the hydrophobic surface area between the two extended β-sheets\nthat have formed. Presumably, some further hydrophobic surface areas\nmay be buried on formation of the wide filaments from the narrow filaments,\nbut it appears that this may not be as extensive as the burial that\noccurs on narrow filament formation. Our results show that the Tv\nnarrow filaments adopt a typical amyloid-like cross-β-structure\nas shown by X-ray fiber diffraction and FT-IR. Most importantly, the\npH dependence of the cac and kinetics suggests that it is likely that\nthe pyridinium side chain is deprotonated in the fibrillar state and\nmay act as a H-bond acceptor with what we think maybe the tyrosine\nside chain ( Figure  \n A). However, we stress that this is a putative structure only.\nAlthough relatively few studies\non the self-assembly of teverelix prior to this one have been reported,\na few papers on the self-assembly of other GnRH antagonists have been\npublished. Early studies on both degarelix and LXT-101, another GnRH\nantagonist, observed the peptide to form a depot in vivo after injection \n , \n  similar to what is seen for teverelix. Further in vitro studies\non degarelix established that residue 7 in the peptide (Leu in degarelix\nand teverelix) is critical in determining the nanostructures that\nthe peptide can form.  Both fibrils, which\nbound to Congo Red, suggesting they are amyloid in nature, as well\nas vesicles with dimensions in the order of tens of nanometers, were\nobserved by TEM depending upon the residue at position 7.  Interestingly, the fibrils formed by degarelix\nhad widths reported to be in the range of a few nanometers and therefore\nsomewhat smaller than those we observe here for teverelix. The two\npeptides differ from each other at positions 5 and 6, which are  l -Tyr- D -Hcit (a carbamylated lysine side chain) in\nteverelix and 4Aph­(L-hydroorotyl)- D -4Aph­(Cbm) in degarelix.\nThus, degarelix has a considerably larger side chain than teverelix\nat position 5, while the only difference at position 6 is the substitution\nof a −CH 2 –CH 2 – group in\nthe lysine side chain in teverelix with a benzyl group in degarelix,\nwhich has the potential to alter the side chain in terms of sterics\nand hydrophobicity. Either/both of these substitutions could cause\nthe differences in self-assembled structures observed, both in terms\nof the width of fibrils and possibly the formation of vesicles (although\nthis is most likely due to changes in the side chain of residue 7).\nOther studies with degarelix have focused on the interaction of the\npeptide with polyanions, including alginate and carboxymethyl cellulose. \n , \n  Degarelix in the absence of the polyanion was shown to form twisted\nfibrils, again potentially amyloid-like in nature. Upon addition of\nthe polyanion, these types of aggregates were observed to dissolve\nand a stable polyanion-degarelix complex formed, which varied somewhat\nin size, shape, and stability dependent upon the polyanion used. \n , \n  A direct comparison with teverelix cannot be undertaken, as we did\nnot perform any experiments with polyanions in this study.\nSelf-assembly\nstudies on LXT-101 using different solution conditions also established\nthat in water the peptide forms stable fibrils, while in the presence\nof excipients such as mannitol, dextrose, or NaCl, less stable vesicles\nwere observed by TEM. These results contrast what we find for teverelix\nhere, where the addition of mannitol or NaCl has no effect on the\nnanostructures formed, and amyloid-like fibrils are always observed.\nThus, we can speculate that teverelix has a higher propensity for\namyloid fibril formation than LXT-101 and possibly degarelix.\n\nIn this study, the amyloid-like identity\nof Tv fibrils was first\nconfirmed using X-ray fiber diffraction, and the morphology of the\nfibrils was studied by TEM. In contrast to many other amyloid-forming\npeptides, narrow filaments were the major species populated over a\nwide range of conditions for Tv, with wider fibrils only being formed\nin any significant amount when agitation was continuous. From the\nstudies of far/near-UV CD and intrinsic fluorescence, the naphthalene\nside chain of Tv is buried and in a fixed conformation even in the\nstarting state, which was shown by SEC to be largely dimeric (with\na small proportion of tetramers/hexamers; the population of which\nincreased with Tv concentration). A small change in the environment\naround the naphthalene side chain was shown to occur on forming the\nslightly larger soluble oligomers compared with the dimer, while a\nlarge change in the environment of the pyridinium and other aromatic\nside chains, including the tyrosine side chain, occurs on fibril formation.\nInterestingly, there appears to be relatively little conformational\nchange in the secondary structure of Tv between the lyophilized powder\nand the fibrils, both states containing mainly β-structure as\nshown by FT-IR. This suggests the starting conformation of Tv is already\nin a highly aggregation-prone state and maybe one reason why Tv forms\nfibrils so rapidly under certain conditions.\nThe kinetics of\nfibril formation by Tv were studied in detail to\nunderstand more about the factors that affect this process. Under\nconditions in which the fibrils formed sufficiently slowly, sigmoidal\nkinetics were observed consistent with a nucleation–polymerization\nmechanism common to many amyloid-forming peptides. The kinetics were\npeptide-concentration-dependent, with faster aggregation occurring\nat higher Tv concentrations. The stability (as shown by the critical\naggregation concentration )  and the kinetics of fibril\nformation were both shown to depend critically on pH. The critical\naggregation concentration decreases with increasing pH, and the rate\nof aggregation increases significantly at higher pH values such that\nby pH 5.0 the self-assembly of Tv is so fast it can only be monitored\nat low Tv concentrations. Additionally, it was found that the counterion\nTFA plays an important role in fibril formation, with results suggesting\nthat not only it is needed to stabilize key species on the self-assembly\npathway and that it is also stably incorporated into the fibrillar\nstate. To the best of our knowledge, this is the first paper to study\nin detail the formation of fibrils of this therapeutic peptide, a\nprocess which is critical to the long-acting action of this drug in\nvivo.","source_license":"CC-BY-4.0","license_restricted":false}