Structural modification of the tripeptide KPV by reductive “glycoalkylation” of the lysine residue

Peptides that exhibit enzymatic or hormonal activities are regulatory factors and desirable therapeutic drugs because of their high target specificity and minimal side effects. Unfortunately, these drugs are susceptible to enzymatic degradation, leading to their rapid elimination and thereby demanding frequent dosage. Structurally modified forms of some peptide drugs have shown enhanced pharmacokinetics, improving their oral bioavailability. Here, we discuss a novel glycomimetic approach to modify lysine residues in peptides. In a model system, the ε-amine of Ts-Lys-OMe was reductively alkylated with a glucose derivative to afford a dihydroxylated piperidine in place of the amine. A similar modification was applied to H-KPV-NH2, a tripeptide derived from the α-melanocyte stimulating hormone (α-MSH) reported to have antimicrobial and anti-inflammatory properties. Antimicrobial assays, under a variety of conditions, showed no activity for Ac-KPV-NH2 or the α- or ε-glycoalkylated analogs. Glycoalkylated peptides did, however, show stability toward proteolytic enzymes.


Introduction
In the recent past, there has been a significant increase in the market for therapeutic peptides and proteins [1]. This interest is attributed to peptides' high selectivity for their target, often with minimal side effects and toxicity [2]. Some problems that must be overcome for therapeutic peptides and proteins include proteolytic instability, immunogenicity, low oral bioavailability, and short half-life [3,4]. In order to enhance the pharmacokinetic properties of peptide drugs, various structural modifications have been effected. Examples of these modifications include N-methylation and the formation of cyclic peptides, which enhance membrane permeability and decrease susceptibility to enzymatic degradation [2,5]. Another strategy is to synthesize peptide analogs incorporating unnatural D-amino acids since they are less susceptible to proteolysis [6]. The half-life of a peptide can be increased using polymer conjugates, such as polyethyleneglycol (PEG) modified peptides. These PEGylated peptides have a larger hydrodynamic volume than their unmodified counterparts, which minimizes the elimination rate of a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 CaH 2. Methanol was dried and distilled from magnesium turnings. Silica gel for flash column chromatography was obtained from Sigma (particle size 40-63 µm). Glass TLC plates were coated with silica gel 60G F 254 manufactured by Merck Millipore. HPLC purification was performed on a Sorbent Purity C18 300Å 5 μm column (250 × 10.0 mm) with a flow rate of 1.0 mL/min and a gradient of 20-90% acetonitrile (+ 0.1% formic acid) over 20 min, monitoring UV-absorbance 218 and 254 nm. 1 H and 13 C NMR spectra were recorded on a Bruker AVIII-400-Nanobay spectrometer, AV500-Prodigy or Bruker AVIII-400-3. Chemical shifts are expressed in ppm downfield of TMS, in deuterated solvents, as specified. Optical rotations were measured on a JASCO-P2000 polarimeter. High resolution mass spectrometry (HRMS) was carried out using an ESI TOF 6210 (electrospray ionization time-of-flight) mass spectrometer (Agilent Technologies). Streptomyces griseus pronase was purchased from VWR, and specified to be !45,000 proteolytic units/g dry weight. A stock solution was prepared by dissolving 2 mg of the lyophilized powder in D 2 O (2 mL).
Sodium borohydride (40.0 mg, 1.06 mmol, 1.0 equiv) was added to a stirred solution of imine in dry methanol (15 mL) at 0˚C and stirred under N 2 for 4 h. The reaction was quenched by dropwise addition of 2M HCl (600 µL), the mixture concentrated, and the residue partitioned between EtOAc (40 mL) and water (10 mL). The aqueous layer was further extracted with EtOAc (2 x 20 mL) and the combined organic extracts were concentrated. The residue was purified by flash chromatography on silica gel, eluting with 95:5 CH 2 Cl 2 -MeOH to afford 6 as a brownish solid (240. 4

Nα-Tosyl-Nε-(2S,4R)-dihydroxypiperidine-L-lysine methyl ester (2).
A solution of Nαtosyl-Nε-1,2-O-isopropylidene-α-D-glucofuranose-L-lysine methyl ester (6) (105.0 mg, 0.24 mmol, 1.0 equiv) in TFA-water (2:1 v/v) solution was stirred for 3 h at rt. The TFA was coevaporated with toluene, and the residue was diluted with water and lyophilized. The dried sample was dissolved in dry MeOH (3 mL) and cooled to 0˚C. Sodium borohydride (30.6 mg, 0.49 mmol, 2.0 equiv) was added and stirring continued for 4 h under N 2 . The reaction was quenched by the dropwise addition of 2M HCl (0.5 mL). The mixture was concentrated, and the residue purified by flash column chromatography on silica gel eluting with 9:1 CH 2 Cl 2 -MeOH. A solution of the purified product 2 in MeOH (1 mL) was kept at 4˚C, which led to crystallization (42.0 mg, 42%). R f 0.37 ( Fmoc-K(Boc)-PV-NH 2 (10a). N-Hydroxysuccinimide (143.3 mg, 1.28 mmol, 1.0 equiv) and DCC (264.1 mg, 1.28 mmol, 1.0 equiv) were added to a solution of Fmoc-Lys(Boc)-OH (600.0 mg, 1.28 mmol, 1.0 equiv) in CH 2 Cl 2 (20 mL) at 0˚C. The mixture was stirred for 20 min, warmed to rt, stirred for 4 h and filtered through a plug of cotton in a Pasteur pipette. The filtrate was concentrated, placed in the freezer for 2 h, filtered a second time and the filtrate concentrated. The residue was dissolved in DMF (6 mL) and cooled in an ice bath. To the stirred mixture was added L-proline (147.4 mg, 1.28 mmol, 1.0 equiv) and diisopropylethylamine (268 µL, 199.0 mg, 1.54 mmol, 1.2 equiv). The mixture was stirred at 0˚C for 10 min, warmed to rt and stirred for 14 h. Dimethylformamide was removed by a stream of air. The residue was taken up in EtOAc (100 mL) and washed with 2M HCl (80 mL). The layers were separated, and the aqueous layer was further extracted with EtOAc (3 x 20 mL). The organic fractions were combined, filtered through anhydrous MgSO 4 and concentrated to afford the dipeptide acid that was used directly without purification R f 0.32 (9:1 CH 2 Cl 2 -MeOH).
Triethylamine (250 µL, 181.4 mg, 1.79 mmol, 3.0 equiv) and flame dried 4Å powdered molecular sieves (75.0 mg) were added to a solution of tripeptide amine (263.4 mg, 0.60 mmol, 1.0 equiv) in dry MeOH (3 mL). The mixture was stirred at rt and a solution of the aldehyde (328.3 mg, 1.91 mmol, 3.2 equiv) in dry MeOH (3 mL) was added. The mixture was left to stir at rt for 24 h. The reaction was filtered through a pad of Celite 1 that was washed well with MeOH. The filtrate was cooled to 0˚C, NaBH 4 (73.2 mg, 1.93 mmol, 3.2 equiv) was added, and the mixture was stirred for 4 h under N 2 . The reaction was quenched by dropwise addition of 2M HCl (250 µL). The mixture was concentrated, and the residue purified by flash column chromatography, eluting with 9:1 CH 2 Cl 2 -MeOH to afford the tripeptide 11a (194 mg, 27%) R f 0.54 (9:1 CH 2 Cl 2 -MeOH).  Ã Reported for the major conformation only; two species were observed that were presumed to be rotamers about the prolyl amide bond.
Boc-K(εG')PV-NH 2 (11b). Following the same series of reactions in the conversion of 10a to 11a above, compound 10b (125.0 mg, 0.19 mmol) was converted to 11b (41 mg, 60%). H-K(εG Ã )PV-NH 2 (12b). By analogy to the procedure described for conversion of 11a to 12a, compound 11b (139.0 mg, 0.23 mmol) was converted to 12b. The crude product was purified by HPLC to afford ε-glycoalkylated 12b ( Determination of stability of tripeptides to pronase. To a solution of each tripeptide Ã in D 2 O (300 µL) was added 1M NH 4 HCO 3 (20 µL) and 50 mM CaCl 2 (40 µL). The pH of the resulting solution was adjusted to 7.0 by the addition of 3.7% HCl (10-12 µL). The volume was adjusted to 395 µL and the 1 H-NMR spectrum recorded at 500 MHz. An aliquot (2 µL) of the 2 mg/mL pronase stock solution was added to the solution of tripeptide and the 1 H-NMR spectrum recorded at 15 min intervals for 1 h at RT. The solution was warmed to 37˚C using the NMR spectrometer's variable temperature controller, and spectra recorded, at 15 min intervals, for 2 h. The reaction was then incubated in an Imperial III incubator (LabLine) at 37˚C and transferred briefly to the NMR probe at room temperature periodically to monitor the reaction.
In the current context, we sought to perform two sequential glycoalkylations in a controlled fashion. Aldehyde 4 was condensed with the ε-amino group of lysine derivative 5. Evidence for imine formation was afforded by 1 H NMR: there was no residual aldehyde signal (δ 9.68 ppm, RCH = O, d, J = 1.9 Hz) and the imine gave rise to a distinct new signal (δ 7.59 ppm, RCH = N, d, J = 4.5 Hz). Following verification of imine formation, reduction was performed under standard conditions to give the secondary amine 6.
The next step in the synthesis of the 3,5-piperidinediol involved liberation of the masked aldehyde followed by an intramolecular reductive amination. Acid hydrolysis of the remaining acetal led to an equilibrium mixture of compounds: the two anomers of hemiacetal 7 and the open chain aldehyde 8. Reduction of the cyclic iminium ion led to formation of piperidine 2.
From the crystal structure of compound 2, shown in Fig 3A, the piperidine-2,4-diol ring is symmetric along the ring plane passing through N and C4. Each hydroxyl group of the diol adopts an equatorial orientation. 1 H NMR analysis of compound 2 confirmed the symmetry of the piperidine, showing three pairs of equivalent protons, Fig 3B: Hx (H2e and H6e); Hy (H2a and H6a); and Hz (H3 and H5). A doublet of doublet peak was observed at δ 2.94 corresponding to H2e, H6e with a large geminal coupling constant (J 2e,2a and J 6e,6a = 10.4 Hz) and a small vicinal coupling constant (J 2e,3a and J 6e,5a = 3.3 Hz). This small vicinal coupling constant places H3 and H5 in axial positions, consistent with the equatorial orientation of the hydroxyl groups in the crystal structure.
Having confirmed the structure and determined reaction conditions for "glycoalkylation," similar conditions were utilized to modify the α-or ε-amino groups of the lysine residue in the tripeptide H-KPV-NH 2 . For site-specific modification, the lysine building block in the tripeptide synthesis had orthogonal protecting groups. For α-modification, the protecting groups were Boc at the ε-position and Fmoc at the α-position. The protecting groups were switched for the ε-modification (Fig 4).
Three derivatives of H-KPV-NH 2 were prepared to test for activity against S. aureus and stability toward proteases: αG Ã -KPV-NH 2 (12a), H-K(εG Ã )PV-NH 2 (12b) and Ac-KPV-NH 2 (12c). The abbreviation G Ã represents the dihydroxylated piperidine in place of the α-NH 2 or ε-NH 2 group in compounds 12a and 12b respectively. The end-capped tripeptide Ac-KPV-NH 2 (12c) was intended as a positive control. For both the α-and ε-modification, Fmoc deprotection of the tripeptide (10a or 10b) led to the free amine at the α-or ε-position, respectively. Each free amine was condensed with aldehyde 4 by reductive alkylation to afford tripeptides 11a and 11b (Fig 5, with the sugar being designated as G' in the furan form). The 1,2-acetonide functionality in compounds 11a and 11b was cleaved in TFA-water, liberating an aldehyde that underwent reductive aminocyclization to form the 3,5-dihydroxypiperidine ring at the α and ε-positions, respectively. Ac-KPV-NH 2 (12c) was synthesized from compound 10a, in order to compare the activities of the two derivatives 12a and 12b with the activity of 12c as previously reported in the literature. Fmoc deprotection of 12a, acetylation of the resulting amine with acetic anhydride, and Boc deprotection with TFA afforded 12c.

Biological assays
The sensitivity of various bacterial strains was tested using the agar diffusion method [33][34][35][36] with the compounds 12a-c that we had synthesized. Details are provided in S1 File. Whilst the positive control, ampicillin, showed inhibition of bacterial growth, no inhibition zones were observed for the negative control, water, and compounds 12a-c.  To verify the activity of Ac-KPV-NH 2 (12c), the peptide was purchased from Bachem (Bubendorf, Switzerland), the same supplier as was used by Charnley et al. [19], following protocols similar to those reported by Cutuli et al. [12] and Charnley et al. [19]. Details are provided in S1 File. Again, no inhibition of bacterial growth was observed.
These results were surprising and disappointing because Ac-KPV-NH 2 (12c) has been reported as an anti-microbial agent [12,19,37]. The original report by Catania and coworkers in 2000 described activity against both Staphylococcus aureus and Candida albicans, with effects over a broad range of concentrations, including "the physiological (picomolar) range [12]." In 2009, there was debate over the original report of antifungal activity [38,39]. Singh and Mukhopadhyay independently described the 90% staphylocidal activity of Ac-KPV-NH 2 (12c) at micromolar concentrations and 50% activity in the nanomolar concentration range [37]. Charnley et al. reported broad range activity against both Gram-positive and Gram-negative bacteria [19]. On the other hand, without further discussion, Grieco et al. stated that "these molecules have weak activity in standard microbiology conditions and this hampers a realistic clinical use [40]." Lau et al. recently performed direct comparisons of 30 ultra-short antimicrobial peptides against Staphylococcus aureus, Pseudomonas aeruginosa, and Candida albicans [41]. Their study included five tripeptides, Ac-KPV-NH 2 (12c) amongst them; none of the tripeptides were active against the panel of skin pathogens, indicating MICs greater than 100 μM. While the compounds did not show any antimicrobial activity under the variety of conditions tested, the impact of glycoalkylation could be assessed vis-à-vis improved stability to proteolytic enzymes. Pronase is a commercially-available cocktail of enzymes used routinely to digest proteins to their constituent amino acids [42]. Each of the three peptides (12a-c) was treated with pronase, and the composition of the mixture monitored by 1 H NMR spectroscopy (see S1 File). The "parent" peptide, Ac-KPV-NH 2 (12c) was degraded to its three constituent amino acids within 24 hours. The signal attributable to Hα of the proline (P) residue shifted upfield by about 0.2 ppm, with a concomitant change from an apparent triplet (in the tripeptide) to a doublet of doublets in the free amino acid, consistent with a change in conformation of the pyrrolidine ring. The signal attributable to Hα of the valine (V) residue shifted upfield by nearly 0.5 ppm. These upfield shifts are in accordance with removal of the electron-withdrawing N-acyl group in each case. The α-glycoalkylated tripeptide (12a) was completely stable under the conditions of the pronase experiment. Less clear-cut was the behavior of the εglycoalkylated tripeptide (12b). The peptide appears to be stable, with Hα signals of both P and V remaining well-defined and with the same chemical shift and the molecular ion was still evident in the mass spectrum. The broad signals assigned to Hε and the protons of the piperidine ring reflect the dynamic nature of the Lys side chain. Upon prolonged incubation with the mixture of proteolytic enzymes, perhaps undergoing autoproteolysis, these signals generally moved upfield and became broader.

Conclusions
We have developed the reaction chemistry to produce regioselectively glycoalkylated peptides. Specifically, reductive amination of D-glucose-derived aldehyde 4 with either the α-or εamino group of lysine residues gave a secondary amine. Upon liberation of the aldehyde derived from the anomeric carbon of glucose, an intramolecular reductive amination could be induced to afford a dihydroxylated piperidine moiety. Acknowledging that the impact of such a modification on biological activity is unlikely to be generalizable to peptides of assorted classes, we sought to study the effect glycoalkylation on the antibacterial activity of Ac-KPV-NH 2 (12c). Unfortunately, during the course of our work, controversy arose in the literature surrounding its alleged antimicrobial activity. Like others, we were unable to reproduce the results under a number of assay conditions. Nevertheless, we have shown that the internal peptide bonds of the glycoalkylated tripeptides, 12a and 12b, are stable over several days to pronase. Future work will involve application of the glycoalkylation concept to other sequences and we trust that this approach will appeal to others interested in improving the bioavailability, solubility and half-life of lysine-containing peptides.
Supporting information S1 File. Procedures and NMR spectra. Experimental procedures for the synthesis of aldehyde 4, 1 H and 13 C NMR spectra for the compounds involved in the synthesis of aldehyde 4, and 1 H and 13 C NMR spectra for the compounds in reaction schemes 1 (Fig 2), 2 (Fig 4), and 3 (Fig 5), computing details, atom coordinates, bond lengths and angles from the X-ray structure determination of compound 2; NMR spectra over the timecourse of the pronase-stability experiments. (PDF) S2 File. Crystallographic information file. Crystallographic information file for the hydrate of compound 2 as determined by X-ray crystallography. Data has been deposited at the CCDC with deposition number 1825648. (CIF)