Angewandte
Communications
Chemie
ceutical preparations. However, novel insulin analogues that
need to be produced chemically may be commercially feasible
as therapeutics for niche markets (such as in regions where
thermal stability is an issue) provided that the synthesis is
efficient. There are already precedents for the large scale
chemical synthesis of other complex peptide-based active
pharmaceutical ingredients.[13] Therefore, replacement of the
disulfide bonds in insulin with stable isosteres to potentially
improve its thermal stability can be viable.
When deciding which disulfide mimetics to use, consider-
ation is given to geometry, toxicity and especially yields since
the chemistry should preferably be scalable. Diselenide,[14]
selenoether,[11] triazole,[15–17] lactam[18,19] and dicarba[20–22]
bridges have all been employed as cystine isosteres. However
a thioether bridge[22–24] was preferred because it more closely
mimics the disulfide bond in biologically active peptides and
proteins. Although an early attempt to incorporate this
moiety to substitute for the intra-A-chain disulfide bond of
ovine insulin via solution synthesis was briefly reported, only
a trace amount of a highly heterogeneous product was
obtained that was not chemically or biologically assessed.[9]
For this reason we undertook to develop an efficient process
of producing A-chain cystathionine insulin, A[S-CH2] human
insulin (Figure 1B) via solid phase peptide synthesis (SPPS).
The assembly of cystathionine-containing peptides has
been studied extensively using tert-butyloxycarbonyl (Boc) or
9-fluorenylmethyloxycarbonyl (Fmoc) SPPS protocols.
Desulfurization of a disulfide bond is one method available
to form the thioether moiety[25] although this can lead to loss
of chiral integrity of the cysteine a-carbon. Forming the
macrocycle via an SN2 reaction between a free thiolate and
halogenated side-chain can be performed on the solid
support[26] or in solution[11] but this can be low yielding due
to steric hindrance and/or hydrolysis of the halide. Incorpo-
ration of pre-formed thioether building blocks with orthog-
onal protection is used widely[22,23,27] and allows macrocycli-
zation to occur via lactamization which is higher yielding.
There is the need, however, to synthesize complex building
blocks. Alternatively, it is possible to alkylate the free thiol of
cysteine with a suitable halogenated amino acid on the solid
support.[28,29] A variation of the latter method—whereby the
nucleophilic attack occurs in the reverse direction—would
only require the synthesis of simple monomeric building
blocks and avoid the difficulties of the preceding methods. To
achieve this, Fmoc-protected g-bromohomoalanine (Fmoc-g-
Br-hAla-OH) was synthesized via a modified version of an
existing method (Scheme 1A).[26] A novel cysteine derivative,
Na-mono-methoxytrityl-cysteinyl-a-allyl ester (Mmt-Cys-
OAll) was also produced efficiently with suitable orthogonal
Scheme 1. A) Reagents and conditions: a) 1 equiv. Fmoc-OSu in pyri-
dine, 4 h; b) HBr/AcOH, 4 h. B) Reagents and conditions: c) 2 equiv.
Mmt-Cl, 5 equiv. DIEA in THF, 4 h; d) 2 equiv. tri-n-butylphosphine,
10 equiv. H2O in THF, 18 h; e) flash chromatography. Fmoc-OSu=
Fmoc succinate; AcOH=acetic acid; Mmt-Cl=mono-methoxytrityl
chloride; DIEA=diisopropylethylamine; THF=tetrahydrofuran.
chain on the solid support was performed to avoid epimeri-
zation of asparagine during loading.[31] Diphenylmethyl
(Dpm)[32] was the preferred protecting group for the A20
thiol since there is insufficient chemoselectivity between trityl
(Trt) and mono-methoxytrityl (Mmt) removal. After assem-
bly of I, acylation of the N-terminus with Fmoc-g-Br-hAla-
OH using carbodiimide activation proceeded efficiently (and
avoids lactone formation) to generate II. Introduction of
Mmt-Cys-OAll under basic conditions resulted in nucleo-
philic attack at the g-carbon by the thiolate to form the
thioether moiety (III). After incorporation of residues A7–
A10 with acetamidomethyl (Acm) protection at A7 (IV),
cleavage of the allyl ester with palladium(0) via allylic
transfer[33] was performed followed by Fmoc removal (V).
Lactamization was slow, most likely due to steric hindrance
on the solid support. Nevertheless, multiple treatments with
the acylating reagents to the resin-bound peptide generated
VI. No significant cross-linking was observed by reversed-
phase high performance liquid chromatography (RP-HPLC)
or matrix-assisted laser desorption ionization mass spectrom-
etry (MALDI-MS). After treatment with 1% trifluoroacetic
acid (TFA) in dichloromethane (DCM) to cleave the Mmt
protecting group from the N-terminus,[34] residues A1–A5
were incorporated to form VII. Cleavage from the solid
support and concomitant deprotection with a TFA cocktail,
followed by purification using RP-HPLC generated the
purified A-chain precursor (VIII) in a total yield of 12%
relative to crude material.
The B-chain was synthesized using standard Fmoc SPPS
protocols and S-Acm derivatization at B7 (IX). After
purification, the peptide was functionalized with the S-
pyridinesulfenyl (S-Pyr) moiety at B19 followed by
a second purification to form X in 5% overall yield. Asym-
metrical combination of A- and B-chain precursors was then
undertaken via displacement of the pyridinesulfenyl leaving
group with the b-thiolate at A20.[35] The synthetic intermedi-
ate XI was then purified in a yield of 45%. A large excess of
elemental iodine was used to efficiently form the final
disulfide bond via oxidative cleavage.[31] After 45 minutes
over 80% of A[S-CH2] human insulin was formed (XII).
Importantly, no oxidation of the thioether moiety to the
sulfoxide was detected. A yield of 25% for the final step was
protecting groups (Scheme 1B).
A similar conceptual
approach has recently been reported by Baell and co-workers
in a simple cyclic peptide system.[30] Our work improves on
this with respect to the higher yields of the amino acid
precursors and application to forming a lactam within
a complex peptidic system. The Fmoc SPPS of A[S-CH2]
human insulin, containing the homoalaninyl-type residue (or
g-carba moiety) at CysA11, is described herein.
The synthesis of the human insulin A-chain is outlined in
Scheme 2. Acylation of the Rink linker via the aspartate side-
2
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2016, 55, 1 – 6
These are not the final page numbers!
Karas, John A.
