Bisubstrate UDP–peptide conjugates as human O-GlcNAc transferase inhibitors

Inhibitors of OGT (O-GlcNAc transferase) are valuable tools to study the cell biology of protein O-GlcNAcylation. We report OGT bisubstrate-linked inhibitors (goblins) in which the acceptor serine in the peptide VTPVSTA is covalently linked to UDP, eliminating the GlcNAc pyranoside ring. Goblin1 co-crystallizes with OGT, revealing an ordered C3 linker and retained substrate-binding modes, and binds the enzyme with micromolar affinity, inhibiting glycosyltransfer on to protein and peptide substrates.


INTRODUCTION
Reversible post-translational modification of nuclear and cytoplasmic proteins with β-linked O-GlcNAc in metazoa is involved in numerous signal transduction cascades that regulate almost every cellular process [1][2][3]. O-GlcNAc cycling is governed by a pair of antagonistic enzymes existing as single-copy genes in two to three splice variants, namely OGT (O-GlcNAc:polypeptidyl transferase) and OGA (O-GlcNAc hydrolase). An ogt gene knockout in mice has been shown to be lethal at the embryonic level [4]. Aberrant O-GlcNAc profiles on certain proteins are associated with the onset and progression of neurodegenerative disease [5]. Existing inhibitors of OGA have been used to induce cellular hyper-O-GlcNAcylation both in vitro and in vivo [6]. Conversely, in-depth elucidation of the biological implications of cellular hypo-O-GlcNAcylation is hampered by the dearth of suitable effectors of OGT. To date, only a limited number of OGT inhibitors have been reported, all targeting the UDP-GlcNAc-binding site. The compound BZX {4-methoxyphenyl 6-acetyl-2-oxobenzo[d]oxazole-3(2H)-carboxylate}, proposed to be a neutral pyrophosphate mimic, was identified as a cell-permeant irreversible inhibitor of hOGT (where h denotes human) [7]. The mechanism of hOGT inhibition with BZX involves cross-linking of the activesite residues Lys 842 and Cys 917 with an S-thiocarbamate link [8]. A unique approach to hOGT inhibition was reported previously [9], employing cell-penetrant per-acetylated 4Ac-5S-GlcNAc (2-acetamido-2-deoxy-5-thio-D-glucopyranose). Once inside the cell, this compound is deacetylated by non-specific esterases producing the free thiosugar 5S-GlcNAc, which is then a substrate for the UDP-GlcNAc biosynthetic pathway, leading to incorporation of the thiosugar into the OGT donor substrate analogue UDP-5S-GlcNAc. The latter inhibits hOGT in vitro (K i = 8 μM) and in vivo, although the basis for selectivity over other GlcNAc transferases remains to be fully explained [9]. A set of non-hydrolysable substrate analogues including the glycosyl thiophosphate (UDP-S-GlcNAc) and Cglycosylphosphonate (UDP-C-GlcNAc) have also been reported [10]. Although these compounds moderately inhibited hOGT, they are not expected to be selective probes. In common with many other glycosyltransferases, OGT is subject to product inhibition, and, accordingly, the most potent inhibitor of hOGT reported to date is the reaction product UDP (K d 0.5 μM) [11]. However, this product inhibition is difficult to exploit for inhibitor design. Not only would the charged nature of the compound render it cellimpermeant, but, in addition, UDP is a central metabolite involved in nucleic acid (RNA) biosynthesis, as well as in the activation of monosaccharides as glycosyl donors for glycoconjugate biosynthesis (six out of the nine mammalian nucleotide sugars are UDP-sugars). UDP analogues consequently have the potential to interact with diverse classes of enzymes such as oxidoreductases, transferases, hydrolases, lyases and isomerases. Bisubstrate inhibitors would offer a means of engineering selectivity by capitalizing on the specific combination of substrates.
Recent structural snapshots of ternary hOGT complexes with substrate/product analogues have uncovered invaluable insights into the hOGT catalytic mechanism [11,12]. Upon binding of the acceptor substrate, hitherto solvent-exposed parts of the bound UDP-GlcNAc engage directly with the peptide backbone, aligning the incoming nucleophile and the anomeric carbon in a glycosyltransfer-competent conformation. This conformation also brings together the acceptor serine and the pyrophosphate moiety, an interaction that has been proposed to be essential for catalysis [11]. In the present study, we exploit these data for the rational design of new OGT bisubstrate inhibitors that combine elements of both substrates; an approach expected to result in selective inhibition of the enzyme compared with inhibitors limited to targeting the donor-binding site alone.

Steady-state kinetics
Reactions contained 50 nM hOGT-(312-1031) in 50 mM Tris/HCl (pH 7.5), 0.1 mg/ml BSA, 10 μM sodium dithionate and 10 μM peptide (KKENSPAVTPVSTA) and various amounts of inhibitors in a total volume of 100 μl. Reaction mixtures were pre-incubated for 15 min and started by addition of UDP-GlcNAc to a final concentration of 3.2 μM. After 30 min of incubation at 22 • C, assays were stopped by adding 200 μl of 25 mM Hepes (pH 7.4), 10 mM NaCl, 50 % (v/v) methanol and 15 μM fluorophore, a UDP-sensitive xanthene-based Zn(II) complex prepared as described in [19,20]. Product formation was detected fluorimetrically on a Gemini EM fluorescence microplate reader (Molecular Devices) at excitation and emission wavelengths of 485 nm and 530 nm respectively. A non-linear regression curve fit was performed with Prism.

Biolayer interferometry
Measurements were made on a ForteBio Octet RED384 instrument at 25 • C. Biotinylated hOGT-(312-1031) was prepared at 25 μg/ml in TBS (25 mM Tris/HCl, pH 7.5, and 150 mM NaCl) buffer containing 1 mM DTT and immobilized on superstreptavidin biosensors. Free streptavidin sites were blocked by incubation with biocytin. A parallel set of superstreptavidin biosensors were prepared with biotinylated streptavidin to act as a control. The assay was carried out in 384-well plates with a sample volume of 100 μl. Inhibitor solutions were prepared from solid stocks and dissolved in assay buffer and a concentration series of 3-fold dilutions from a top concentration of 100 μM was created. Cycles for analysis involved obtaining a 30 s baseline followed by a 60 s association step and a 120 s dissociation step. The assay was repeated with the reference biosensors to correct for non-specific interactions and the entire assay was repeated in triplicate for both compounds. Data were processed and kinetic parameters were calculated using ForteBio software.

RESULTS AND DISCUSSION
Inspection of the ternary hOGT-UDP-5S-GlcNAc-TAB1 peptide complex [11] shows that the anomeric carbon of the sugar is positioned at a distance of 3.4 Å from the modified peptide side chain ( Figure 1A). Biophysical determination of the binding affinity for product and substrate revealed a 30-fold higher K d for UDP-GlcNAc than for UDP [11], suggesting that, in the absence of an acceptor, the sugar moiety does not positively contribute to the binding affinity.
We consequently envisaged a set of bisubstrate OGT inhibitors in the form of UDP-peptide conjugates in which UDP is coupled to a variable peptide subunit by a suitable linker. The weakly binding GlcNAc moiety was omitted from the inhibitor structure and replaced with a linear three-or four-carbon tether to retain the spatial arrangement of UDP and peptide as observed in the Michaelis complex. The resulting compounds were named goblin1 (OGT bisubstrate-linked inhibitor 1) and goblin2 respectively.
Synthesis of the target compounds entailed the preparation of phosphorylated 'stretched serine' derivatives (Scheme 1) suitable for Fmoc solid-phase peptide synthesis and peptide assembly followed by reaction of the phosphorylated peptides with activated nucleoside monophosphate (Scheme 2). Synthesis of analogous peptide-nucleotide polyphosphate conjugates as protein kinase bisubstrate inhibitors using an on-resin phosphorylation-pyrophosphorylation protocol has been reported previously [21]. In the first instance, we opted for off-resin pyrophosphorylation for the reason of operational and analytical convenience. We established a robust, albeit moderately yielding, two-step procedure for the preparation of the key 'stretched serine' allyl esters 3a/b by alkylation [22] of the dianion generated from the commercially available N-Boc (N-t-butoxycarbonyl) serine 1 with MP (p-methoxyphenyl)-protected 3-bromopropan-1-ol or 4-bromobutan-1-ol 2a/2b [NaH, DMF (dimethylformamide)] followed by esterification with allyl bromide/DIPEA (N,N-diisopropylethylamine) (Scheme 1). Attempted alkylation of 1 under phase-transfer conditions [23] was unsuccessful, whereas more elaborate synthetic schemes proved to be inefficient. Ensuing N-protective group remodelling as well as removal of the terminal hydroxy group protection (3a/3b→4a/4b→5a/5b) was ensured by the perfect orthogonality of the MP protecting group, resulting in nearly quantitative yield for three steps. Finally, installation of the phosphate group with acid-labile MBn (p-methylbenzyl) protection and deblocking of the C-terminus yielded the required phosphorylated 'stretched serine' building blocks 8a/8b with very good overall efficiency. Careful choice of the phosphate protecting group was essential, as the traditional  [11] (PDB code 4AY6) and the hOGT-goblin1 complex. hOGT is shown as a molecular surface. The donor substrate UDP-5S-GlcNAc (pink/turquoise carbon atoms), the acceptor peptide (yellow carbon atoms) and goblin1 (green carbon atoms) are shown as stick models. On the right, a close-up view of the compounds is shown, centred on the linker incorporated in goblin1. For the hOGT-goblin1 complex, an unbiased (i.e. before incorporating of any ligand) four-fold NCS (non-crystallographic symmetry) averaged F o − F c electron density map is shown, contoured at 3.5σ . Peptide synthesis was performed on a microwave-assisted CEM Liberty instrument using RinkAmide MBHA (4-methylbenzhydrylamine) resin and HCTU [O-(1H-6-chlorobenzotriazole-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate] as the coupling agent (for details, see the Supplementary Online Data). The synthesis of the UDP-peptide conjugates 14a/b is illustrated in Scheme 2. Crude phosphopeptides 11a/b were reacted with a 10-fold excess of 2 ,3 -O-diacetyl-UMP imidazolide prepared from triethylammonium salt 13 and CD1 (1,1 -carbonylbisimidazole) overnight, deacetylated with methanol/triethylamine/water and finally purified by reversephase HPLC to furnish the targeted UDP-peptide conjugates 14a (goblin1) and 14b (goblin2) in fair yield. The identity of the newly synthesized compounds was confirmed by TOF-ESI-MS and NMR spectroscopy. To investigate the contribution of the separate parts of the UDPpeptide conjugate towards inhibition potency, we synthesized 'linker-only' peptide 12 using 'stretched serine' building block 10 (Scheme 1). We also prepared the matching 'linker-only' UDP analogue 17 starting from 3-methoxypropanol taking advantage of the one-pot/three-step pyrophosphorylation procedure mediated by di-(p-methoxybenzyl)-N,N-di-isopropylphosphoramidite 16 [24] (Scheme 3).
The affinity of goblin1 and goblin2 for hOGT was evaluated by biolayer interferometry ( Figure 1B, and Supplementary Figure S1 at http://www.biochemj.org/bj/457/bj4570497add.htm) yielding K d values of 7.9 and 4.9 μM respectively. In vitro glycosylation of a peptide substrate was inhibited in a dose-dependent manner with an IC 50 value of 18 μM for goblin1 and 40 μM for goblin2 ( Figure 1C). Furthermore, the ability of goblin1 to inhibit O-GlcNAcylation of a well-characterized human substrate protein, TAB1 [18], was investigated by Western blotting employing a TAB1 O-GlcNAc Ser 395 site-specific antibody ( Figure 1D, and Supplementary Figure S2 at http://www.biochemj. org/bj/457/bj4570497add.htm). Dose-dependent inhibition of hOGT activity was observed, and densitometric quantification allowed the calculation of an IC 50 value of 8 μM (Supplementary Figure S2). In a set of control experiments with 'linkeronly' compounds 12 and 17, we observed only weak hOGT inhibition by 17 (IC 50 300 μM) whereas 'linker-only' peptide 12 was proved to be neither a substrate nor an inhibitor ( Figure 1C). Attempts to inhibit OGT in cells appeared unsuccessful, probably due to the size/negative charge of the compounds.
To confirm the binding mode of goblin1, we co-crystallized the compound with hOGT and determined the crystal structure of the complex (Supplementary Table S1 at http://www.biochemj. org/bj/457/bj4570497add.htm). Electron density difference maps at 3.15 Å, improved by 4-fold non-crystallographic averaging (Supplementary Figure S3 at http://www.biochemj.org/bj/457/ bj4570497add.htm), revealed unambiguous density for the entire compound, including the ordered C 3 linker ( Figure 1A). As envisaged, UDP adopts the same conformation as observed in the hOGT Michaelis complex ( Figure 1A; maximum atom shift, 0.6 Å) and the peptide occupies the − 4 to + 2 subsites with a similar backbone conformation (maximum backbone atom shift near O-GlcNAc site, 0.9 Å). The three-carbon linker connects the two components apparently without introducing any strain, allowing both the UDP moiety and the peptide part of the inhibitor to adopt the optimal position in the binding site, effectively mimicking the natural substrates.
One of the main objectives of combining components of donor and acceptor substrate into a bisubstrate inhibitor is the expected improved selectivity of such a construct when compared with inhibitors that compete with a single substrate only. Although accessing the selectivity of the novel OGT bisubstrate inhibitors over an exhaustive panel of GlcNAc transferases is beyond the scope of the present study, we were able to establish that goblin1 is not an inhibitor of the GlcNAc transferase SmNodC (Sinorhizobium meliloti NodC) (Supplementary Figure S4 at http://www.biochemj.org/bj/457/bj4570497add.htm).

Concluding remarks
O-GlcNAc modification of proteins is abundant and essential, yet its precise cellular function remains to be uncovered. The development of small-molecule inhibitors of OGT is a prerequisite for the elucidation of the biological role of O-GlcNAc, and eventually the therapeutic intervention in diseases involving dysregulation of O-GlcNAc. With multisubstrate enzymes, such as OGT, specific inhibition can rarely be achieved with ligands that compete solely with one of the substrates. Peptide-nucleotide polyphosphate conjugates have been reported as bisubstrate inhibitors of protein kinases in the early 1990s; the discovery recently led to the identification of low-nanomolar cell-penetrant inhibitors with neutral linkers replacing the polyphosphate chain [25]. The pioneering work of Palcic et al. [26] introduced the concept of bisubstrate inhibition for glycosyltransferases, targeting fucosyltransferases [27].
In the present paper, we have reported the first examples of functional OGT bisubstrate inhibitors featuring a short linker replacing the GlcNAc moiety of the donor substrate and covalently connecting a uridyl diphosphate 'binding anchor' to an acceptor peptide. The resulting compounds, goblin1 and goblin2, exhibit low-micromolar affinity for OGT and are capable of inhibiting glycosylation of peptide as well as protein substrates in vitro. The crystal structure of the hOGT-goblin1 complex reveals the structural mimicry offered by the inhibitor, with both UDP and peptide moieties retaining the alignment observed in the Michaelis complex. As expected for a bisubstrate inhibitor, both fragments add synergistically to the inhibition: the 'linker-only' derivative of UDP was shown to be an order of magnitude less potent an inhibitor, whereas the 'linker-only' acceptor peptide was neither a substrate for the OGT nor an inhibitor. The goblin scaffold allows for further increases in potency by optimization of linker length/type, peptide sequence and nucleotide modifications, whereas cell penetrance could be addressed by extending the peptide backbone with cell-penetrating peptide/peptoid motifs [28,29] potentially bearing intracellular localization signals, to provide tools for studying OGT function in vivo.

Compound 3a
To a solution of the SerBocOH (N-t-butoxycarbonylserine) (2.25 g, 11 mmol) in DMF (40 ml) 60 % sodium hydride suspension in oil (0.88 g, 22 mmol) was added sequentially in two equal portions at 0 • C (ice-bath). When gas evolution ceased, the reaction was removed from the cooling bath and stirred at room temperature until gas evolution ceased (1 min). The reaction was returned to the ice-bath and tetra-n-butylammonium iodide (0.369 g, 1 mmol) and a solution of 1-(3-bromopropoxy)-4-methoxybenzene 2 (3.23 g, 13.2 mmol) in DMF (15 ml) were added in a succession. The reaction was removed from the icebath and stirred at room temperature for 16 h. The reaction was quenched by careful addition of a few drops of 10 % citric acid solution, diluted with ethyl acetate and washed successively with 10 % citric acid and water. The aqueous layers were back extracted with the ethyl acetate two times more. The combined organic layer was dried and concentrated. The residue was dissolved in DMF (25 ml) and treated with allyl bromide (1.3 ml, 15 mmol) in the presence of DIPEA (2.6 ml, 15 mmol) at room temperature for 16 h. The reaction was diluted with methanol and concentrated. The residue was partitioned between ethyl acetate and 1 M HCl, and the layers were separated. The organic layer was successively washed with water and a mixture of saturated NaHCO 3 solution and brine. The aqueous layers were back-extracted with ethyl acetate. The combined organic layer was dried and concentrated. The residue was adsorbed on to silica and purified by flash chromatography in petroleum ether/ethyl acetate 5-10-20 % to give 1.37 g (3.36 mmol, 30 %) of the target product as clear oil.

Compound 4a
To a cold (ice-bath) stirred solution of 3a (1.18 g, 2.9 mmol) in DCM (dichloromethane) (15 ml) was added 95 % aqueous TFA (trifluoroacetic acid) (1.5 ml). The reaction was kept for 2 h at room temperature, diluted with toluene (10 ml) and concentrated. The residue was dissolved in a mixture of chloroform and toluene (1:3, v/v) (10 ml) and concentrated. This procedure was repeated once more. The residue was dissolved in DCM (20 ml) and treated with an excess of DIPEA (2 ml) and FmocCl (Fmoc chloride) (0.905 g, 3.5 mmol) at 0 • C (ice-bath). The reaction was kept at room temperature for 16 h, quenched with methanol (0.1 ml), stirred for 30 min and concentrated. The residue was dissolved in DCM and successively washed with 1 M HCl, water and a mixture of NaHCO 3 solution and brine. Aqueous layers were back-extracted with DCM. The combined organic layer was dried and concentrated. The residue was adsorbed on to silica gel and purified by flash chromatography in petroleum ether/ethyl acetate 5-10-20 % to give 1.42 g (2.67 mmol, 92 %) of the target product as crystalline solid. The analytical sample was crystallized from toluene/petroleum ether. Melting

Figure S1 Biolayer interferometry sensograms
Binding profiles obtained and kinetic parameters calculated for interactions between hOGT and goblin1 and goblin2. Concentration series of each compound (3-fold serial dilution, top concentration 100 μM) were prepared in triplicate. Left-hand panels: binding profiles and global curve fitting from which the kinetic parameters were calculated. Right-hand panels: steady-state binding response from which the steady-state K d was calculated.

Compound 5a
To a stirred solution of 4a (1.39 g, 2.61 mmol) in a mixture of THF (tetrahydrofuran)/acetonitrile (1:1, v/v) (15 ml) a solution of ammonium cerium(IV) nitrate (3 g, 5.47 mmol) in water (7.5 ml) was added at room temperature. The reaction was stirred for 20 min, diluted with 10 % EDTA solution and ethyl acetate and the layers were separated. The organic layer was washed successively with 10 % aqueous Na 2 S 2 O 5 solution, water and a mixture of saturated NaHCO 3 solution and brine. The aqueous layers were back-extracted with ethyl acetate. The combined organic layer was dried and concentrated. The residue was adsorbed on to silica gel and purified by flash chromatography in (petroleum ether/DCM, 4:1, v/v)/ethyl acetate 5→45 % to give 1.1 g (2.58 mmol, ∼ 100 %) of the target product as tancoloured crystals. The analytical sample was crystallized from chloroform/petroleum ether.  The reaction was carried out in a 50 ml Falcon centrifuge tube. A solution of di-isopropylamine (3.5 ml, 25 mmol) in hexane (10 ml) was added dropwise to a stirred solution of phosphorus trichloride (1.1 ml, 12.6 mmol) in hexane (30 ml) at − 50 • C. The reaction was removed from the cooling bath, warmed to room temperature, and further vigorously stirred for 1 h and centrifuged at 3700 g for 15 min at 4 • C. Approximately 30 ml of the supernatant was transferred into a recovery flask; the residue was resuspended in methyl t-butyl ether (15 ml) and centrifuged as before. The supernatant was added to the recovery flask. The solution was concentrated to give 1.93 g (9.55 mmol, 76 %) of the crude target product (δ 169.9 p.p.m.) as a slightly yellowish liquid. A solution of the above residue (0.97 g, 4.8 mmol) in THF (5 ml) was added to a solution of p-methylbenzyl alcohol 1.17 g, 9.6 mmol) and di-isopropylethylamine (2.5 ml, 14.4 mmol) in THF (20 ml) at 0 • C (ice-bath) (reaction was carried out in a 50 ml Falcon centrifuge tube). The reaction was removed from the cooling bath and stirred further for 2 h at room temperature; the precipitate started to fall in 10-15 min. The reaction was diluted with methyl t-butyl ether (total volume 40 ml) and centrifuged as before. The supernatant was collected; the residue was suspended in methyl t-butyl ether (20 ml) and centrifuged again. The supernatants were pooled and concentrated. The residue was partitioned between DCM and 1 M phosphate (pH 7) buffer (pH of the aqueous phase 7 after shaking) and the layers were separated. The organic layer was washed with brine. The aqueous layers were successively extracted with the same portion of DCM. The organic layers were dried, concentrated and dried under vacuum overnight to give 1.7 g of the crude product as a yellowish freely running oil, which was shown to be sufficiently pure by NMR (δ 146 p.p.m.).

Compound 7a
To a solution of 5a (0.85 g, 2 mmol) and dimethylbenzyl N,Ndi-isopropylphosphoramidite 6 (0.97 g, 2.6 mmol) in acetonitrile (10 ml), 4,5-dicyanoimidazole (0.307 g, 2.6 mmol) was added at room temperature and reaction was stirred for 1 h. The reaction mixture was cooled down to 0 • C (ice-bath) and 3chloroperbenzoic acid (0.64 g, 2.6 mmol) was added in one portion. The reaction was stirred further for 1 h and quenched by the addition of a 20 % solution of Na 2 S 2 O 5 , stirred for 30 min, removed from the cooling bath and diluted with DCM. The layers were separated; the organic layer was washed successively with water and a mixture of concentrated NaHCO 3 solution and brine. The aqueous layers were back-extracted with DCM. The combined organic layer was dried and concentrated. The residue was adsorbed on to silica gel and purified by flash chromatography in (petroleum ether/DCM, 4:1, v/v)/ethyl acetate 10-40 % to give 1.21 g (1.7 mmol, 85 %) of the target product as a clear oil.  13 13 (17), or peptide and linker (12). To a solution of 5a (0.123 g; 0.29 mmol) in DCM (5 ml) and DIPEA (0.116 ml, 0.7 mmol), TBSCl (t-butyldimethylsilyl chloride) (0.075 g, 0.5 mmol) and DMAP (4-dimethylaminopyridine) (0.04 g, 0.035 mmol) were added sequentially. The reaction mixture was stirred for 16 h at room temperature. The reaction was quenched by the addition of methanol, stirred for 30 min and concentrated. The residue was partitioned between ethyl acetate and 10 % aqueous citric acid solution and the layers were separated. The organic layer was washed successively with water and a mixture of saturated NaHCO 3 solution and brine. The aqueous layer was additionally extracted with the same portion of ethyl acetate. The combined organic layer was dried and concentrated. The residue was purified by flash chromatography on silica; gradient elution petroleum ether/ether 10-30 % to give 0.14 g (0.26 mmol, 90 %) of the target product as a clear syrup.
[α] D = + 7.4 • (c 1.23 in chloroform) 1   were added sequentially. The reaction was stirred for 1.5 h; TLC (petroleum ether/DCM, 4:1, v/v)/ethyl acetate 40 % and DCM/methanol 10 % showed disappearance of the starting material and formation of a more polar new product. The reaction was quenched by addition of a suspension of Dowex 50WX8-100 H + resin in methanol and stirred for 10 min. The resin was filtered off, and the filtrate was concentrated to give 0.123 g (0.24 mmol, quant) of the target product as a yellow syrup. 1

Compound 11a
Peptide synthesis was performed on RinkAmide MBHA (4-methylbenzhydrylamine) low-load resin (0.38 mmol/g, 0.13 g, 0.05 mmol) (Novabiochem). Standard couplings were performed using 5 equiv. of amino acid, 5 equiv. of HCTU [O-(1H-6-chlorobenzotriazole-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate] and 10 equiv. of DIPEA in a microwaveheated reactor (CEM Liberty microwave peptide synthesizer) for 5 min at 70 • C. Deprotections were performed with 20 % piperidine in DMF initially for 0.5 min at 70 • C and then with a fresh portion of deprotection mixture for 3 min at 70 • C. Addition of the 'stretched serine' building block 8a was performed using 2.5 equiv. of the amino acid for 20 min at 60 • C. After this point, deprotections were performed initially for 5 min at room temperature and then with a fresh portion of deprotection mixture for 15 min at room temperature. The peptide was cleaved from the resin with a mixture of TFA/water/TIPS (tri-isopropylsilane) 92.5:5:2.5, by vol. (2 ml) for 3 h. The cleavage mixture was filtered into the centrifuge tube containing cold (0 • C) diethyl ether (40 ml). The resin was additionally twice washed with 1 ml of TFA for 5 min each and the washings were added to the same tube. The tube was kept in the freezer (−18 • C) for 16 h. The precipitated peptide was centrifuged at 3700 g at 4 • C for 15 min; the supernatant was discarded. The pellet was suspended in diethyl ether and centrifuged again. The procedure was repeated once more. The pellet was blow-dried with argon, dissolved in 20 % aqueous acetic acid and freeze-dried to give 0.025 g (0.03 mmol) of the crude peptide. HRMS (m/z), [ The peptide was synthesized using 'stretched serine' building block 10 (2.5 equiv.) according to the general procedure. 1