Letter Cite This: Org. Lett. XXXX, XXX, XXX−XXX
pubs.acs.org/OrgLett
Improving the Cell Permeability of Polar Cyclic Peptides by Replacing Residues with Alkylated Amino Acids, Asparagines, and D‑Amino Acids Laura K. Buckton and Shelli R. McAlpine* School of Chemistry, University of New South Wales, Sydney NSW 2051, Australia S Supporting Information *
ABSTRACT: The design, synthesis, and cell permeability of 19 hydrophilic macrocyclic peptides is presented. By systematically analyzing the impact of three different approaches (alkylated amino acids, asparagines, and D-amino acids) on the permeability of polar peptides, a well-defined strategy for optimizing cell permeability is provided. These three new methods can be used individually or in combination to effectively convert polar peptides into cell permeable molecules, and the results can be applied to the rapidly expanding peptide therapeutic industry.
R
oughly 140 peptide therapeutics are evaluated in clinical trials each year1 where the global peptide drug market is predicted to increase from $21 billion in 2015 to $46.6 billion in 2024.2 However, the most challenging issue facing peptide drug development is the optimization of their physical properties, including cell membrane permeability.3 Improving the permeability of peptide drugs is difficult, and as such the focus has been on producing permeable molecules before evaluating their biological activity. Pioneering research has been accomplished by leaders in this field including Kessler,4,5 Lokey,6−9 Yudin,10,11 and Fairlie12,13 (Figure 1). Four strategies have been investigated: backbone N-methylation, inclusion of asparagine-like side chains, side-chain modifications, and Damino acids. The challenge of increasing permeability is so significant that strategies have primarily focused on modifying hydrophobic scaffolds. The agreed upon threshold for considering peptides cell permeable are compounds with apparent permeability (Papp × 10−6 cm/s) ≥ 1.6,8 However, our goal is to produce polar peptides with Papp values of ∼8, which would place them in the same range as small molecules (marketed drugs Papp median = 16, average = 4).14 Several examples of macrocycles containing one polar side chain have been reported as cell permeable (Papp > 1), but there are no reports of cyclic peptides containing more than one polar side chain that can penetrate membranes. Therapeutically relevant molecules often contain polar and structurally diverse side chains. Understanding how strategies can be applied to compounds with more than one polar side chain is critical when developing new molecular scaffolds. Two of the most commonly used criteria to predict membrane permeability are calculated partition coefficients (cLogP) and polar surface area (PSA). These calculations © XXXX American Chemical Society
Figure 1. Structures and permeability values of selected macrocyclic peptides studied by leading groups in the field: Fairlie,12 Lokey,8,9 Kessler,4 and Yudin.11 Permeability is represented as (Papp × 10−6 cm/ s) where the superscript indicates values from (a) Caco-2 cells, (b) PAMPA, or (c) RRCK cells.
provide useful predictions for small, 2-D molecules15 but often fail to provide insight into the permeability of complex, 3-D molecules like macrocyclic peptides. As such, recent work has expanded the criteria for cyclic peptides, where cLogP is Received: October 30, 2017
A
DOI: 10.1021/acs.orglett.7b03363 Org. Lett. XXXX, XXX, XXX−XXX
Letter
Organic Letters recommended to be between 0 and 5 and PSA < 200 Å2.16 Given that cyclic peptides invariably violate at least one criterion while maintaining their permeability, it has been proposed that the cell permeability of these molecules is related to their “chameleonic” behavior, whereby the molecules change their conformation and hydrophobicity depending on their surrounding environment.16,17 However, predicting “chameleonic” behaviour is currently not possible and therefore cell permeability of complex molecules must be determined through experimentation. Described herein is the design, synthesis, and permeability of 19 macrocyclic peptides that are based on two therapeutically interesting, polar structures, LB51 and LB59.18,19 Compounds were designed systematically in order to evaluate the impact of three strategies on the permeability of cyclic pentapeptides that contained up to five polar side chains. The new molecules incorporated: (a) alkylated amino acids to mask the polar side chains, (b) asparagine residues, and (c) D-amino acids. The synthesis of these new compounds is described, and their cell permeability properties are evaluated. CLogP and PSA were calculated in ChemDraw v16 and the applicability of utilizing these values to predict permeability is evaluated. Together these data offer insight into the most effective approaches for converting biologically relevant structures into cell permeable molecules. The two lead structures, LB59 and LB51 are relatively polar, both containing a free lysine, serine, and tyrosine (Figures 2 and 3). In order to improve cell permeability of LB59, which is membrane impermeable (Papp = 0.83), series 1 was produced (Figure 2). These molecules contain alkylated side chains, where their roles in cell permeability are examined for the first time on polar cyclic peptides. All compounds were synthesized utilizing Fmoc solid-
Figure 3. Series 2: Structure of LB51 analogues incorporating methyl groups on side chains. Papp determined in Caco-2 assays. Average of four replicates ± SEM.
phase peptide chemistry to produce linear peptides (Scheme 1). Scheme 1. Synthesis of Compound 2 on Solid Phase
Cleaving these peptides from the resin and cyclizing them in solution under dilute conditions (0.001 M) with a cocktail of coupling agents20 afforded the cyclic peptides with alkylated side chains. For molecules 2, 4, and 5, the side-chain protecting groups were removed from the macrocycles using TFA in dichloromethane (50−90% v/v), which upon purification by reversed-phase HPLC produced the final product. For compounds 1 and 3, selective removal of protecting groups was required. All molecules were characterized using LC/MS and NMR. Compounds were evaluated in Caco-2 assays in order to determine their cell permeability and the values were compared
Figure 2. Series 1: Structure of LB59 analogues incorporating alkylated amino acids. Papp determined in Caco-2 assays. Average of four replicates ± SEM. B
DOI: 10.1021/acs.orglett.7b03363 Org. Lett. XXXX, XXX, XXX−XXX
Letter
Organic Letters to the commonly used control compounds phenol red21 and atenolol.22 Placing an isopropyl moiety on lysine’s side chain produced compound 1 (Papp = 1.57, Figure 2), which had slightly improved permeability over LB59. Maintaining the isopropyl group on the lysine and masking the serine with methyl or tert-butyl groups produced compounds 2 and 3, respectively. It was hypothesized that both molecules would have increased permeability given that they had higher cLogP values and lower PSA than 1. However, both compounds were less permeable than 1, with Papp values of 0.58 and 0.11 for 2 and 3, respectively. Conversion of the isopropyl (2) to a carboxybenzyl moiety produced compound 4, which was insoluble and could not be tested for permeability. Inclusion of a tert-butyl group on the serine and a dimethyl group on the lysine produced compound 5, which had similar permeability to 2 and 3. Surprisingly, the hydrophobicity of the tert-butyl group did not improve permeability. The most significant finding from series 1 (compounds 1−5) is that despite the molecules being within the recommended ranges for cLogP and PSA,16 they could not penetrate cells. Thus, calculated properties are not adequate predictors of permeability and experimental strategies are required in order to evaluate cell permeability. Synthesis of analogues of LB5118 (Papp = 3.27) produced series 2. All four molecules were produced utilizing a similar strategy to series 1 (Scheme 1). The dimethyl lysine analogue 6 was significantly less cell permeable than LB51, with a Papp = 1.05 versus 3.27 (Figure 3) despite the cLogP within the recommended range. Similarly, the cLogP increased and PSA decreased with compound 7 compared to LB51 (Figure 3) and these changes did not reflect the compounds’ poorer cell permeability (Papp = 0.58). Removal of a single methyl group on the lysine of 7 produced 8, which had a slightly higher permeability (Papp = 1.89) than 7. In contrast to the trend observed in series 1, inclusion of a methyl group on serine, coupled with a methyl on tyrosine and dimethyl on lysine, produced the most permeable compound of the series, 9 (Papp = 5.55). Analysis of molecules 1−9 indicated that alkylating the serine, tyrosine, and lysine with a methyl improved the permeability: compare 9 versus LB51 (Papp = 5.55 versus 3.27). The most significant finding of series 2 is that the correct combination of alkylated side chains was required. These data indicated that large hydrophobic groups should be avoided and methyl groups are the most effective at masking the polar side chains. Compound LB59 contains an alanine at the position where LB51 contains an asparagine, and the improved permeability of LB51 over LB59 (Papp = 3.27 versus 0.83, respectively) supports Yudin’s theory that asparagine’s may facilitate cell permeability.11 Therefore, 7 and 9 were used as a design starting points in series 3 (Figure 4). Compounds were synthesized to include two or three asparagines, 10, and 11, respectively (Figure 4) using the conditions described for the synthesis of molecules 1−9. Compound 10, with two asparagines, was relatively permeable (Papp = 4.35). In contrast, 11, which contained three asparagines had a lower Papp = 2.58 than LB51. Combining the favorable properties of 9 and 10 led to the design and synthesis of 12, which contained methyl groups on lysine, tyrosine, and serine, as well as two asparagines. The methyl and asparagine effects were not additive, where the permeability of 12 was lower than 9 or 10 (Papp = 2.34 versus 5.55 or 4.35, respectively). These data prove that macrocyclic conformations are a major contributor to the cell permeability (e.g., 9 versus 12), and the impact of
Figure 4. Series 3: Structures incorporating asparagine and methyl groups on the polar side chains. Papp determined in Caco-2 assays. Average of four replicates ± SEM.
modifying side chains is best identified through experimental methods (e.g., free serine in 7 versus methylated serine in 9, and phenylalanine in 9 versus asparagine in 12). Diastereomers of LB51 were also tested for their permeability as shown in Table 1 (series 4) (synthesis reported Table 1. Series 4: Diastereomers of LB51
Papp determined in Caco-2 assays. Average of four replicates ± SEM.
in ref 19). These molecules would confirm whether the sidechain position and macrocyclic conformation were critical for permeability. It was anticipated that inverting the stereochemistry would influence the conformation of the molecules and impact permeability. Four out of the five LB51 diastereomers showed improved permeability over LB51. The most effective modification was inversion of lysine from L- to D-, yielding compound 14 with a Papp = 4.62. Thus, despite these molecules having identical cLogP and PSA values, they have different permeability properties, proving that the orientation of the peptide backbone plays a significant role in membrane permeability. In order to ensure the observed trends in Caco-2 assays were a result of passive diffusion and not due to active transport, we evaluated lead compound LB51 and the most permeable molecule (9) in a parallel artificial membrane permeability assay (PAMPA). PAMPA measures passive diffusion only, whereas C
DOI: 10.1021/acs.orglett.7b03363 Org. Lett. XXXX, XXX, XXX−XXX
Letter
Organic Letters
(2) Peptide Therapeutics Market.http://www. transparencymarketresearch.com/peptide-therapeutics-market.html (3) Craik, D. J.; Fairlie, D. P.; Liras, S.; et al. Chem. Biol. Drug Des. 2013, 81, 136. (4) Biron, E.; Chatterjee, J.; Ovadia, O.; et al. Angew. Chem., Int. Ed. 2008, 47, 2595. (5) (a) Ovadia, O.; Greenberg, S.; Chatterjee, J.; et al. Mol. Pharm. 2011, 8, 479. (b) Marelli, U. K.; Bezençon, J.; Puig, E.; et al. Chem. Eur. J. 2015, 21, 8023. (6) Furukawa, A.; Townsend, C. E.; Schwochert, J.; et al. J. Med. Chem. 2016, 59, 9503. (7) (a) Rand, A. C.; Leung, S. S. F.; Eng, H.; et al. MedChemComm 2012, 3, 1282. (b) Rezai, T.; Bock, J. E.; Zhou, M. V.; et al. J. Am. Chem. Soc. 2006, 128, 14073. (c) Rezai, T.; Yu, B.; Millhauser, G. L.; et al. J. Am. Chem. Soc. 2006, 128, 2510. (d) Schwochert, J.; Lao, Y.; Pye, C. R.; et al. ACS Med. Chem. Lett. 2016, 7, 757. (8) Bockus, A. T.; Schwochert, J. A.; Pye, C. R.; et al. J. Med. Chem. 2015, 58, 7409. (9) White, T. R.; Renzelman, C. M.; Rand, A. C.; et al. Nat. Chem. Biol. 2011, 7, 810. (10) Zaretsky, S.; Scully, C. C. G.; Lough, A. J.; et al. Chem. - Eur. J. 2013, 19, 17668. (11) Hickey, J. L.; Zaretsky, S.; St. Denis, M. A.; et al. J. Med. Chem. 2016, 59, 5368. (12) (a) Hill, T. A.; Lohman, R.-J.; Hoang, H. N.; et al. ACS Med. Chem. Lett. 2014, 5, 1148. (b) Nielsen, D. S.; Lohman, R.-J.; Hoang, H. N.; et al. ChemBioChem 2015, 16, 2289. (13) Nielsen, D. S.; Hoang, H. N.; Lohman, R.-J.; et al. Angew. Chem., Int. Ed. 2014, 53, 12059. (14) O'Hagan, S.; Kell, D. B. PeerJ 2015, 3, e1405. (15) (a) Lipinski, C. A.; Lombardo, F.; Dominy, B. W.; et al. Adv. Drug Deliv. Rev. 1997, 23, 3. (b) Veber, D. F.; Johnson, S. R.; Cheng, H.-Y.; et al. J. Med. Chem. 2002, 45, 2615. (16) Santos, G. B.; Ganesan, A.; Emery, F. S. ChemMedChem 2016, 11, 2245. (17) Whitty, A.; Zhong, M.; Viarengo, L.; Beglov, D.; Hall, D. R.; Vajda, S. Drug Discov. Today 2016, 21, 712. (18) Buckton, L. K.; Wahyudi, H.; McAlpine, S. R. Chem. Commun. 2016, 52, 501. (19) Rahimi, M. N.; Buckton, L. K.; Zaiter, S. S.; Kho, J.; Chan, V.; Guo, A.; Konesan, J.; Kwon, S.; Lam, L.; Lawler, M. F.; et al. ACS Med. Chem. Lett. 2018, DOI: 10.1021/acsmedchemlett.7b00310. (20) Styers, T. J.; Rodriguez, R.; Pan, P.-S.; et al. Tetrahedron Lett. 2006, 47, 515. (21) Ferruzza, S.; Scarino, M. L.; Gambling, L.; et al. Cell Mol. Biol. 2003, 49, 89. (22) Lennernaäs, H. J. Pharm. Sci. 1998, 87, 403.
Caco-2 assays evaluate both passive and active transport. LB51 had Papp values of 2.97 ± 0.03 and 3.27 ± 0.20 measured in PAMPA and Caco-2, respectively. Compound 9 had Papp values of 5.41 ± 0.11 and 5.55 ± 0.20 measured in PAMPA and Caco2, respectively. The agreement between the values calculated in both assays indicates that these molecules passively diffuse across membranes and the established structure-permeability relationships are due to changes in passive diffusion and not the result of active transport. In conclusion, we report the synthesis of 12 new compounds and the permeability of 19 molecules. Three molecules (9, 10, and 14) showed reasonable cell permeability, with Papp values on par with optimized examples reported by others (Figure 1). While those previously reported compounds included up to one polar group, our molecules contained four to five polar groups, demonstrating that the strategies presented here can be successfully applied to improve the permeability of polar cyclic peptides. This work also demonstrates that the theoretical predictors cLogP and PSA are unreliable and it supports current evidence that cyclic peptides can be cell permeable despite violating proposed permeability criteria. Given that compounds 9, 10, and 11 all have PSA > 200 Å2 and compounds 10 and 11 have cLogP values
