Synthesis of an analog of the thyroid-hormone-binding protein transthyretin via regioselective chemical ligation

Transthyretin is an essential protein responsible for the transport of thyroid hormones and retinol in human serum and is also implicated in the amyloid diseases familial amyloidotic polyneuropathy (FAP) and senile systemic amyloidosis (SSA). Its folding properties and stabilization by ligands are of current interest due to their importance in understanding and combating these diseases. Here we report the solid phase synthesis of the monomeric unit of a transthyretin analog (equivalent to 127 amino acids) using t-Boc chemistry and peptide ligation and its folding to form a functional 54 kDa tetramer. The monomeric unit of the protein was chemically synthesized in three parts: 1-51, 54-99 and 102-127 and ligated using a chemoselective thioether ligation chemistry. The synthetic protein was folded and assembled to a tetrameric structure in the presence of TTR’s native ligand, thyroxine, as shown by gel filtration chromatography, native gel electrophoresis, transthyretin antibody recognition and thyroid hormone binding. Other folding products included a high molecular weight aggregate as well as a transient dimeric species. This represents one of the largest macromolecules chemically synthesized to date and demonstrates the potential of protein chemical synthesis for investigations of protein:ligand interactions.

The ligand binding properties of TTR have become of major importance recently, with the discovery that certain thyroid hormone competitors (e.g. 2,4,6-triiodophenol) are able to decrease the tendency of TTR to form amyloid fibrils (15,16). They are reported to act by binding deeply within the TTR binding channel at both ligand binding sites and inhibiting the formation of amyloid by stabilizing the normal fold against the pathogenic conformational change. A range of non-steroidal antiinflammatory drugs are currently being investigated for their ability to inhibit and reverse amyloid formation (17). Whilst crystallographic information has revealed the orientations of several of these drugs bound to TTR, there are many aspects of the ligand binding that cannot be probed using crystallographic techniques. These include dynamic aspects of protein:ligand binding in solution and in the presence of competitors, or in the presence of other serum proteins, including the other thyroid hormone carriers albumin and thyroxine binding globulin. For such studies it is possible that nuclear magnetic resonance (NMR) spectroscopic methods could be employed.
NMR can be used to observe the interaction of a ligand with particular sites in the protein. Such studies currently depend upon the assignment of the specific resonances that are perturbed by the ligand (18) which limits studies to systems in which the protein signal of interest is either fortuitously distinct or has been fully assigned using isotopic labeling and heteronuclear NMR techniques. In the latter case this also requires that the protein is less than about 40 kDa due to the broad linewidths of the NMR signals and large number of signals that have similar chemical shifts. The problem can be somewhat reduced by using chemical ligation strategies in which just by guest on March 24, 2020 http://www.jbc.org/ Downloaded from one portion of the protein is isotopically labeled (19). Of even greater value would be the ability to completely control the position of the isotopic label in a protein for probing the ligand interaction.
The current study thus presents the first stage in the development of a spectroscopic method for probing a protein:ligand interaction. The strategy involves the complete synthesis of a TTR analog using solid-phase synthesis and chemical ligation techniques, and refolding of the protein to a tertiary and quaternary structure that approximates the native form. Since the structure of TTR is known to atomic resolution from X-ray crystallographic studies, an NMR active probe (i.e. 15 N-or 13 Clabelled amino acid) may be incorporated at any strategic position, to enable subsequent ligand binding studies to be manifest in NMR spectra.
The chemical synthesis of proteins of the size of the monomeric unit of TTR (127 residues) represents a significant challenge. In the past, long peptides were synthesized in a stepwise fashion, as exemplified by HIV-1 protease (20) and IL-8 (21) -but significant purification problems resulted in low yields of protein.
Currently, the synthesis techniques for proteins of this size rely on chemoselective ligation techniques, where two or more non-protected peptides are joined through a highly selective chemical reaction.
There have been several notable protein syntheses described using chemoselective ligation techniques that incorporate thioester and thioether surrogate amide bonds.
Native chemical ligation chemistry is only useful provided there are suitably positioned cysteines. In order to synthesize a protein like transthyretin without suitably located cysteines it becomes necessary to use chemically modified amino acid substitutes or amino acid mutations. The thioether ligation strategy we use here introduces -NH-CH 2 -CH 2 -S-CH 2 -CO-, which mimics a two amino acid subunit closely resembling a glycyl-glycine (-NH-CH 2 -CO-NH-CH 2 -CO-). While the thioether moiety closely resembles the spatial requirements for glycyl-glycine it may lack potential hydrogen bond donation/acceptor behavior of the di-amino acid unit, thereby potentially introducing some non-native structural characteristics.
Here we demonstrate the total chemical synthesis of an analog of human TTR through the use of the thioether strategy for the sequential ligation of three peptides. We also show that this synthetic TTR (henceforth referred to as sTTR) may be successfully refolded and reconstituted to form a 54 kDa tetrameric structure able to bind the thyroid hormone T4. This represents one of the largest active proteins made synthetically, and provides methodology for future protein:ligand investigations using NMR spectroscopic techniques. All reagents were of analytical grade.

Chemicals and Reagents
Native hTTR -Native hTTR was isolated from serum using an adapted version of the method described by Dwulet and Benson (30).
Peptide Synthesis-Peptides were synthesized using the rapid manual HBTU in-situ neutralization synthesis technique (31) or using the same technique on a modified ABI 430A peptide synthesizer (32). The thioether resins were prepared according the

HAEVVFTAND-NH-CH 2 -CH 2 -SH. (Chloroacetyl-54-99-NH-CH 2 -CH 2 -SH) (II)-The
C-terminal thiol peptide II was manually synthesized using the thiol linker attached to Boc-Ala-Pam resin in the first synthesis then Boc-Gly-Pam resin for the second. The average amino acid coupling for the syntheses was 99.5 and 99.6%, respectively. The DNP protecting group was removed followed by N-Boc and CHO group removal.
The chloroacetyl group was coupled then the peptide HF cleaved. The crude peptide was purified by preparative HPLC using a linear gradient of 0-70% B, then analyzed

Solid Phase Synthesis of H-GPTGTGESKAPLMVKVLDAVRGSPAINVAVHV-
C-terminal thiol peptide III was synthesized using machine assisted synthesis. The peptide was synthesized using the thiol linker attached to Boc-Gly-Pam resin. The average amino acid coupling was 99.6% (1st coupling) for the synthesis, which was routinely double coupled. The DNP protecting group was removed followed by N-Boc and then the formyl group. The peptide was cleaved from the resin and the crude peptide purified by preparative HPLC using a linear gradient of 0-70% B. The  Chloro-Iodo exchange of (IV) to give I-Ac-ELHGLTTEEEFVEGIYKVEIDTKSYW-   (35).

Formation of H-GPTGTGESKAPLMVKVLDAVRGSPAINVAVHVFRKAADDTWEPF-
Detection achieved was using enhanced chemiluminescence against X-ray film.
Preparation of L[ 125 I]-thyroxine -Commercially available L[ 125 I]-thyroxine was found to contain up to 5% 125 Ion the reference date. Therefore, 125 I-thyroxine was separated from 125 Iand other degradation products by reversed phase chromatography using a SepPak C-18 cartridge column (36). Purification was checked by thin layer chromatography followed by autoradiography (37).
Analysis of thyroxine binding to synthetic human TTR -Commercially purchased 125 I-thyroxine was purified from degradation products and 125 Ias described above.
Methyl cellulose charcoal was prepared as described by Chang et al. (38). In order to remove thyroxine from the solution containing sTTR, 40 µl of methylcellulosecharcoal (1%) in Tris-HCl pH 8.9 was centrifuged and the supernatant was removed.
The methylcellulose-charcoal was resuspended in a 160 µl solution containing 3.2 µg sTTR. The mixture was kept at 4 o C for half an hour, with mixing each 10 minutes.
The solution was centrifuged and the supernatant removed for analysis of 125 Ithyroxine binding to sTTR.

Synthetic strategy
The choice of ligation sites for the preparation of sTTR was based on both the amino acid sequence and the known tertiary structure of the TTR molecule (Fig. 1A,B) and involved the ligation of three peptides (Fig. 1C). Whilst the ligation of two peptides, each of 60-65 residues in length, is a possible alternative strategy, the degree of difficulty of preparing and purifying peptides of this length is comparable with the difficulty of a second ligation. In addition, since all of the residues considered for future labeling studies occur within the last 30 residues it was desirable to prepare a relatively small C-terminal fragment. It was anticipated that this approach would increase the ease of preparing several sTTR molecules with selective labels since only this fragment would need to be resynthesized.
The thioether linker spans a distance equivalent to two amino acids, is highly flexible

(peptide VI).
In addition to non-native bonds, one other modification was made to sTTR. Cysteine 10 was replaced with alanine in order to avoid any problems of competition for the iodoacetyl peptide fragment during the ligation reaction and to exclude the possibility for oxidation when the protein was folded. Alanine was chosen due to its similarity in bulk and hydrophobicity to cysteine. This was considered unlikely to have any detrimental effect on the folding or activity of TTR.

Peptide Synthesis and purification
Trial syntheses of the peptides I, II and III were carried out to see how each of the peptides would behave during the synthesis, cleavage and ligation reactions. This preliminary work showed that all three peptides could be readily synthesized, cleaved from the resin, and purified, excepting the middle fragment, which tended to retain the methylphenoxyacetic acid (AMPA) linker at the C-terminal thiol after HF treatment.
This was detected as a result of cleavage at the C-terminal amino acid attached to the by guest on March 24, 2020 http://www.jbc.org/ Downloaded from resin on which the peptide-AMPA linker was synthesized (see experimental for details). Despite this the three target peptides were obtained in good yield and purity.

Ligation of peptides I and II
Ligation of peptides I and II proceeded cleanly to give peptide IV (Fig. 2). The purification and monitoring of the ligation reaction was complicated by the fact that the ligated peptide IV co-elutes with the starting peptide II, thus the use of an excess of peptide I was required to drive the reaction to completion. After 2 h the starting peptide II could not be detected and the ligated peptide IV was present in high yield as judged by HPLC and ESMS (Fig. 3A,B). The ligation was left overnight for completion of the reaction. The broad HPLC profile of peptide II was attributed to the chloroacetyl and the thiol functionalites at the N and C-termini of the peptide. The absence of the chloroacetyl group, in particular, significantly reduces the broadness of the peak shape.

Formation of the iodoacetyl peptide V
The initial chloro-iodo exchange reaction and the subsequent ligation were carried out in-situ (peptide III was added to peptide IV in saturated KI solution). Most of the thiol peptide III was found to form disulfide dimers, and the iodoacetyl peptide V also lost a small percentage of its iodofunctionality. In subsequent ligations the iodoacetyl peptide V was either purified by RP-HPLC and the ligation reaction carried out immediately after lyophylisation or purified by rapid desalting using a PD10 column

Ligation of peptide III and V to give VI
After purification of the iodoacetyl peptide V its ligation to peptide III  was carried out as rapidly as possible under nitrogen. Samples were removed from the ligation mixture and analyzed by HPLC and ESMS. The rate of the ligation reaction, affording the ligated 1-127 TTR VI, was monitored by a slow HPLC gradient (0.5% /min B) as the ligated product eluted very close to the two starting peptides (III and V), and the disulfide-linked peptide formed by the unreacted excess of peptide III (Fig. 4A). The ligated 1-127 peptide VI was easily purified by RP-HPLC in excellent yield, and its identity confirmed by ESMS (Fig 4B).

Folding of synthetic TTR to give the 54 kDa Tetrameric Complex.
The synthetic TTR (sTTR) spontaneously folded to its tetramer complex in the presence of the ligand T4, in 0.075 M NH 4 HCO 3 . In addition to the formation of tetramer (as initially determined by its equivalent retention time to that of native TTR upon size exclusion chromatography) two other products were also formed. These corresponded to species with an intermediate MW, likely to be a dimeric form of sTTR, as well as a high molecular weight aggregated sTTR (Fig 5A). The ratio of these products altered over time in the refolding buffer, with a gradual accumulation of the high MW aggregate. After isolating the tetrameric complex it was equilibrated at 37 o C for 48 h in the presence of excess T4 ligand. Some re-equilibration between the monomeric, tetrameric and the high molecular weight aggregates occurred in the initial 2h but the ratios of tetramer to monomer and the high molecular weight aggregates remained constant after that time (Fig 5B).

Western analysis of synthetic TTR
Western analysis was employed to determine the subunit molecular mass and confirm the recognition of the sTTR by anti-TTR antiserum (Fig. 6). An aliquot of human serum was analyzed as a positive control. TTR has a subunit molecular mass of approximately 15 kDa, as estimated by SDS-polyacrylamide gel electrophoresis (39).
The interactions between monomers to form the dimer are very strong, and even after boiling in the presence of 2% SDS for 20 minutes, some TTR still exists as a dimer (38). Thus bands with molecular masses of about 15 and of about 34 kDa correspond to the TTR monomer and dimer, respectively. Bands are also apparent at higher molecular masses. These result from non-specific binding of the antibodies to other proteins in serum as is commonly observed (35).
Synthetic TTR gave rise to bands corresponding to the molecular masses of the sTTR monomer and dimer (Fig. 6, lanes 2 and 3). The bands were discrete, and no indication of partially synthesized or partially degraded sTTR was apparent.

Analysis of thyroxine binding to synthetic TTR
The correct folding and formation of the tetramer was assessed by analyzing nondenaturing polyacrylamide gel migration combined with a 125 I-thyroxine binding assay. The analysis of 125 I-thyroxine binding to proteins in human serum was used as the reference. This revealed the presence of thyroxine-binding globulin, albumin and TTR ( Fig. 7 lanes 1 and 4). indicating that the tetrameric size, shape and charge distribution of sTTR were almost identical to native TTR. There was no evidence for the existence of aggregates.

Discussion
We have shown that it is possible to chemically synthesize and correctly fold a transthyretin analog (sTTR) from its monomeric unfolded state to produce the 54 kDa tetrameric quaternary structure in the presence of TTR's strongest-binding native ligand T4. This is likely to be due to the stabilizing effect of T4, which has previously been reported to stabilize native TTR against acid denaturation leading to the formation of amyloid fibrils (15). The integrity of the final product was confirmed by native TTR antibody recognition and ligand binding studies. Since the binding site for thyroid hormone ligands is formed only upon the formation of the tetrameric species, ligand binding also demonstrated tetramer formation.
Competing folding pathways involving the formation of high molecular weight aggregate as well as the appearance of smaller molecular weight species, possibly corresponding to dimeric and monomeric sTTR, were also apparent. The former pathway was expected, considering the predisposition for TTR to form amyloid fibrils and, in particular, the increased propensity of many mutant forms of TTR to do so.
The appearance of dimeric-like species of sTTR however, was unexpected, as this species has never previously been reported. The current model of the unfolding pathway of TTR involves a transition between correctly folded TTR to its monomeric form via a perturbed tetramer and an extremely transient dimeric form (40). The monomeric TTR is thought to be in equilibrium with a molten globule-like monomeric structure or be sequestered to irreversible amyloid formation. Our refolding studies of sTTR have revealed a species that appears to have the molecular weight of the dimeric form of sTTR. It was not possible to isolate this dimeric species for further identification. Rechromatographing the species using gel diffusion chromatography only gave rise to the high molecular weight aggregate and some of the monomeric species. This highly unstable form may be unique to sTTR, or represent an alternative species that could be the basic unit of TTR-based amyloid fibrils.
It is not surprising that the sTTR construct has a high propensity to form aggregates.
Investigations into mutant forms of TTR have shown that even in the absence of apparent perturbations to the tertiary and quaternary structure of TTR, most mutant forms of TTR display a higher propensity to form amyloid fibril than native TTR (41)(42)(43). The mutations, rather than underlying an alternative TTR conformation, are thought to cause subtle perturbations to the equilibrium between the different forms of TTR. Any such movement towards the monomeric form of TTR thus results in the increased opportunity for TTR monomer to be irreversibly incorporated into amyloid.
As observed for native TTR, the tetrameric form of sTTR is its most stable form. It is not known whether the sTTR tetramer forms and then is stabilized by T4 or the tetramer spontaneously forms about the T4. The stabilization of sTTR by a high affinity ligand, however, appears analogous to the stabilization observed for TTR in the presence of several non-steroidal anti-inflammatory drugs, currently under investigation for their ability to inhibit human transthyretin amyloid disease (17).
These molecules, which include flufenamic acid, diclofenac and flurbiprofen, bind in the T4 binding site of TTR (45), forming similar polar and non-polar contacts as do the natural ligands. These interactions are thought to stabilize the native quaternary structure of TTR against pH-mediated dissociation and conformational changes associated with amyloid formation (16). It may be that some of these drugs would also promote the folding of sTTR from its unfolded monomeric state.
In conclusion, the current study reports a methodology for the total chemical synthesis of a TTR analog (sTTR) through use of a thioether strategy for the sequential ligation of three peptides and successful folding and formation of the sTTR tetramer in the presence of the native ligand T4. It is remarkable that this macromolecule can be synthesized, however, since other multimeric forms of sTTR also formed readily, the construct may not be not ideal for biophysical studies. It is possible that other TTR analogs made using thioether linkages placed in alternative positions may provide closer mimics of native structure. Such constructs potentially allow the incorporation of 15 N or 13 C-isotopic labels at TTR ligand binding sites and may facilitate future TTR:ligand studies using NMR spectroscopy. In particular, the kinetics of the interaction between sTTR and TTR-binding drugs of current interest for their antiamyloidogenic capacity may be carried out in the presence of other competitive binding proteins to provide a more complete picture of their mode of action in serum.