Ternary nylon-3 copolymers as host-defense peptide mimics: beyond hydrophobic and cationic subunits.

Host-defense peptides (HDPs) are produced by eukaryotes to defend against bacterial infection, and diverse synthetic polymers have recently been explored as mimics of these natural peptides. HDPs are rich in both hydrophobic and cationic amino acid residues, and most HDP-mimetic polymers have therefore contained binary combinations of hydrophobic and cationic subunits. However, HDP-mimetic polymers rarely duplicate the hydrophobic surface and cationic charge density found among HDPs ( Hu , K. ; et al. Macromolecules 2013 , 46 , 1908 ); the charge and hydrophobicity are generally higher among the polymers. Statistical analysis of HDP sequences ( Wang , G. ; et al. Nucleic Acids Res. 2009 , 37 , D933 ) has revealed that serine (polar but uncharged) is a very common HDP constituent and that glycine is more prevalent among HDPs than among proteins in general. These observations prompted us to prepare and evaluate ternary nylon-3 copolymers that contain a modestly polar but uncharged subunit, either serine-like or glycine-like, along with a hydrophobic subunit and a cationic subunit. Starting from binary hydrophobic-cationic copolymers that were previously shown to be highly active against bacteria but also highly hemolytic, we found that replacing a small proportion of the hydrophobic subunit with either of the polar, uncharged subunits can diminish the hemolytic activity with minimal impact on the antibacterial activity. These results indicate that the incorporation of polar, uncharged subunits may be generally useful for optimizing the biological activity profiles of antimicrobial polymers. In the context of HDP evolution, our findings suggest that there is a selective advantage to retaining polar, uncharged residues in natural antimicrobial peptides.

Introduction

The innate immune response to bacterial infection includes the release of relatively short peptides that attack prokaryotic cells in preference to eukaryotic cells.[1] These “host-defense peptides” (HDPs) often have a net positive charge near neutral pH; Coulombic forces attract HDPs to bacterial cell surfaces, which generally bear a substantial net negative charge.[2−4] Once HDPs are localized on a cell surface, they are thought to disrupt membrane barrier function.[2,3] These disruptions appear to be associated with induction of membrane curvature, and the type of curvature involved requires certain proportions of lysines, arginines, and hydrophobic residues, which sets a specific range of cationic charge density and hydrophobicity for HDPs.[5] Many HDPs adopt α-helical conformations in interfacial environments, while others form small β-sheets or discrete tertiary structures enforced by internal disulfides.[4]

The variation in conformation, composition, and sequence evident among HDPs has led a number of groups to explore sequence-random copolymers as functional HDP mimics.[6−8] Such efforts are appealing from a practical perspective because the high cost of synthesizing peptides or other oligomers with discrete sequences would make it very difficult to use this type of molecule in real-world applications.[1b] HDP-mimetic copolymers are interesting from a fundamental perspective because their properties pose a question: why would evolution have elicited sequence-specific polypeptides for a task that can be accomplished in the absence of sequence control? One possible answer to this question is that HDPs perform functions in addition to bacterial membrane disruption.

The HDP-mimetic copolymers studied to date have been largely binary, containing one subunit that displays a positive charge on a side chain and another subunit that displays a hydrophobic side chain.[6−9] In addition, antimicrobial homopolymers have been constructed from subunits that contain both cationic and hydrophobic moieties.[10] This focus on hydrophobic and cationic components, however, is somewhat at odds with the considerable sequence variation among HDPs. Indeed, HDP-mimetic polymers typically display higher charge density and more extensive hydrophobic surface than do HDPs.[11] A high-fidelity mimic must simultaneously have lower cationic charge density and lower levels of hydrophobicity, which is difficult to achieve with most binary cationic and hydrophobic subunit combinations. These considerations prompted us to examine three-component polymers containing cationic, hydrophobic, and “neutral” subunits as potential HDP mimics. Glycine is the single most common residue among known HDPs, even though Gly ranks behind Leu and Ala among known proteins.[12] Serine is quite common among HDPs, with a prevalence comparable to that found among proteins.[12] We synthesized and evaluated the biological activity profiles of ternary nylon-3 copolymers in which previously described hydrophobic and cationic subunits are supplemented by either a glycine-like or serine-like subunit.

Results and Discussion

Polymer Synthesis and Characterization

The ternary nylon-3 polymers discussed here contain subunits from one of two β-lactams that have not previously been employed to generate antimicrobial materials (Scheme 1). One β-lactam bears no substituent and gives rise to a glycine-like subunit in polymer chains. This simplest of β-lactams is a derivative of β-alanine, which can also be designated β-homoglycine.[13] We designate this β-lactam HGβ and the resulting subunit HG. The other β-lactam, designated HSβ, gives rise to a β3-homoserine subunit (HS) in polymer chains. The hydroxymethyl β-lactam precursor to HSβ has previously been synthesized in enantiopure form,[14] and we followed the reported route to generate racemic material, which was then trityl-protected to generate HSβ. Each ternary polymer contains a hydrophobic subunit, CO or CH, and a cationic subunit, MM or DM. The latter subunits bear a peripheral amino group that should be protonated and therefore positively charged at neutral pH. The necessary β-lactams, COβ, CHβ, MMβ, and DMβ, have been used previously to generate biologically active binary nylon-3 copolymers.[8] All of the chiral β-lactams were used in racemic form, and all of the ternary copolymers discussed here are therefore heterochiral mixtures.

Scheme 1

β-Lactams and Corresponding Subunits within the Nylon-3 Polymer Chain

Nylon-3 copolymers were synthesized via base-initiated anionic ring-opening polymerization (AROP) of β-lactams.[15] Reactions involving HGβ were conducted in N,N-dimethylacetamide (DMAc), and reactions involving HSβ were conducted in tetrahydrofuran (THF). p-tert-Butylbenzoyl chloride was included in all of the polymerizations; this acid chloride presumably reacts rapidly under AROP conditions to generate a mixture of N-acyl-β-lactams.[8,16,17] These imides then serve as co-initiators for the ring-opening polymerization process, leaving a p-tert-butylbenzoyl group at the N-terminus of each chain.[16,17] An imide derived from one of the three β-lactam starting materials remains at the C-terminus of each chain.

Acid chloride:β-lactam ratios were chosen to generate average chain lengths of 18–35 subunits. The resulting materials were analyzed via gel-permeation chromatography (GPC) prior to removal of side-chain protecting groups. Polydispersity index (PDI) values were in the range 1.05–1.43 according to data from multiangle light scattering (MALS) and a refractive index detector. Boc and trityl side-chain deprotection was accomplished by treatment with trifluoroacetic acid (Figure 1). The structures of the ternary polymers are shown in Figure 2.

Figure 1

Synthesis of DM + CH + HS ternary copolymers. All of the other ternary copolymers discussed here were synthesized analogously.

Figure 2

Structures of the ternary nylon-3 copolymers examined in this work. R′ represents the side chain from one of the starting β-lactams. All of the polymers are sequence-random and heterochiral.

Biological Activities

A panel of four bacteria, including one Gram-negative species (Escherichia coli[18]) and three Gram-positive species (Bacillus subtilis,[19]Staphylococcus aureus,[20] and Enterococcus faecium(21)), was used to gauge the antibiotic properties of the ternary copolymers. Antibacterial activity was measured in terms of minimum inhibitory concentration (MIC), i.e., the lowest concentration of a polymer that fully inhibits bacterial growth. Evaluation of HDPs and synthetic analogues typically involves an assessment of prokaryotic cell versus eukaryotic cell selectivity based on a comparison of MIC values with the extent to which an agent disrupts red blood cell (RBC) membranes. The latter property was measured by monitoring the release of hemoglobin from RBCs as a function of agent concentration. We used HC10, the concentration of a polymer that causes release of 10% of the hemoglobin from human RBCs, as our index of hemolytic activity.

The studies reported here are based largely on two previously reported binary nylon-3 copolymers, MMCO and DMCH.[8c] Both polymers are active against the bacteria in our test panel (low MIC values), but both are also quite hemolytic (low HC10 values). These two binary polymers provide an excellent opportunity to determine whether incorporation of a third nylon-3 subunit, one that is neither cationic nor hydrophobic, can enable decoupling of the antibacterial activity from the hemolytic activity. In all previous studies by our group and others involving binary cationic–hydrophobic copolymers, diminishing the hydrophobic content necessarily meant increasing the cationic content and vice versa. Use of an HG or HS subunit, however, allows the hydrophobic content to be changed without affecting the overall cationic content or the overall charge to be altered without changing the net hydrophobicity.

Table 1 compares the biological activity profile of MMCO with the profiles of six new ternary copolymers, three in which varying proportions of the hydrophobic CO subunits have been replaced with HG subunits and three in which varying proportions of the CO subunits have been replaced with HS subunits. The HC10 values listed in Table 1 are supplemented by the full hemolysis data sets presented in Figure 3. The proportions indicated in the polymer designation are based on the β-lactam proportions used for the synthesis of that polymer (i.e., the material designated MMCOHS was prepared by copolymerization of a β-lactam mixture containing 60 mol % MMβ, 30 mol % COβ, and 10 mol % HSβ). In general, the β-lactam starting materials were fully consumed in the polymerization reactions. The data in Table 1 show that replacing a small proportion of the hydrophobic CO subunits with “neutral” HG units has relatively little effect on the biological activity profile: the ternary copolymers MMCOHG and MMCOHG are very similar to the binary polymer MMCO in terms of MIC and HC10 values. Replacing most of the hydrophobic CO units with HG, as in MMCOHG, causes a stark decline in the hemolytic activity, but the antibacterial activity declines as well (i.e., the MIC value becomes higher). This observation is consistent with the general view that a minimum level of hydrophobicity is necessary for membrane-disruption activity.

Table 1

Biological Activities of MM + CO + HG and MM + CO + HS Ternary Copolymers versus MMCO

				MIC (μg/mL)d
polymer	Mna	DPb	PDIc	E. coli	B. subtilis	S. aureus	E. faecium	HC10 (μg/mL)e
MM60CO40	4402	22	1.06	25	0.78	3.13	3.13	6.25–12.5
MM60CO30HG10	5352	29	1.29	25	3.13	6.25	3.13	3.13
MM60CO25HG15	5704	31	1.31	25	3.13	12.5	6.25	12.5
MM60CO10HG30	5670	33	1.43	200	3.13	25	50	>400
MM60CO30HS10	5441	25	1.07	25	1.6	6.25	3.13	50–100
MM60CO25HS15	6296	28	1.27	200	1.6	6.25	6.25	100–200
MM60CO10HS30	8873	35	1.05	>200	6.25	100	100	>400

Number-average molecular weight.

Average degree of polymerization.

Polydispersity index based on GPC data obtained for polymers bearing Boc protecting groups on MM unit side chains and trityl protecting groups on HS unit side chains.

Minimum inhibitory concentration for bacterial growth.

Polymer concentration for 10% lysis of hRBCs.

Figure 3

Hemolytic profiles of (left) MM + CO + HG and (right) MM + CO + HS ternary copolymers vs MMCO.

The HS subunit is expected to be more hydrophilic than the HG subunit because of the pendant hydroxyl group in the former, and replacement of CO with HS exerts more interesting effects on the biological activity profile relative to replacement with HG. Thus, the ternary copolymer MMCOHS is generally similar to MMCO in terms of antibacterial activity, but the ternary copolymer is considerably less hemolytic than the binary prototype. The antibacterial activity erodes, however, with further replacement of CO subunits by HS subunits.

Table 2 presents the biological activity profiles for ternary copolymers related to the binary copolymer DMCH; hemolytic assay results are shown in Figure 4. Replacing a small proportion of the hydrophobic CH units with either HG or HS has a similar impact: relative to the binary starting point, DMCHHG and DMCHHS display comparable antibacterial activities but significantly diminished hemolytic activities. Increasing the extent of hydrophobic unit replacement to generate DMCHHG or DMCHHS causes a substantial decline in antibacterial activity. In this series, we explored ternary copolymers in which a subset of the cationic units is replaced with either HG or HS. This effort was motivated by the prior observation that the DM homopolymer is quite hemolytic.[8c] For these modifications, HS provides a more interesting outcome than HG. Both DMCHHG and DMCHHS are relatively similar to DMCH in terms of antibacterial profile, but DMCHHS is significantly less hemolytic than the binary copolymer, while DMCHHG is comparable to the binary copolymer in hemolytic activity.

Table 2

Biological Activities of DM + CH + HG and DM + CH + HS Ternary Copolymers versus DMCH

				MIC (μg/mL)d
polymer	Mna	DPb	PDIc	E. coli	B. subtilis	S. aureus	E. faecium	HC10 (μg/mL)e
DM50CH50	4766	26	1.05	12.5	≤1.6	6.25	6.25	6.25
DM50CH40HG10	4341	24	1.25	12.5	3.13	12.5	12.5	25–50
DM50CH25HG25	5539	33	1.25	50	3.13	25	50	200
DM50CH40HS10	5731	28	1.09	25	1.6	12.5	6.25	50
DM50CH25HS25	7482	32	1.08	100	3.13	50	25	200
DM40CH50HG10	4383	26	1.21	6.25	3.13	12.5	12.5	12.5
DM40CH50HS10	5443	28	1.09	12.5	3.13	12.5	12.5	100

Number-average molecular weight.

Average degree of polymerization.

Polydispersity index based on GPC data obtained for polymers bearing Boc protecting groups on DM unit side chains and trityl protecting groups on HS unit side chains.

Minimum inhibitory concentration for bacterial growth.

Polymer concentration for 10% lysis of hRBCs.

Figure 4

Hemolytic profiles of (left) DM + CH + HG and (right) DM + CH + HS ternary copolymers vs DMCH.

The results summarized in Table 2 suggest that improvements in the biological activity profile (i.e., increases in HC10 without major changes in MIC) can be achieved by replacing small amounts of either the hydrophobic or the cationic subunit in DMCH with neutral HG or HS subunits; therefore, we examined another set of ternary copolymers in which small amounts of bothDM and CH subunits were replaced (Table 3 and Figure 5). This approach proved to be effective; the polymer DMCHHG appears to have the most favorable profile among all of the ternary polymers examined in this study.

Table 3

Biological Activities of Representative Binary versus Ternary Nylon-3 Copolymers Having the Same Hydrophobic:Cationic Molar Ratio and Varied Content of the Third (Hydrophilic) Subunit HG or HS

				MIC (μg/mL)d
polymer	Mna	DPb	PDIc	E. coli	B. subtilis	S. aureus	E. faecium	HC10 (μg/mL)e
DM50CH50	4766	26	1.05	12.5	≤1.6	6.25	6.25	6.25
DM47.5CH47.5HG5	3558	20	1.20	12.5	≤1.6	12.5	12.5	400
DM45CH45HG10	3174	18	1.24	25	≤1.6	12.5	25	400
DM45CH45HS10	4943	25	1.12	25	1.6	12.5	12.5	200–400
DM40CH40HS20	5539	26	1.05	50–100	3.13	25	50	200–400

Number-average molecular weight.

Average degree of polymerization.

Polydispersity index based on GPC data obtained for polymers bearing Boc protecting groups on DM unit side chains and trityl protecting groups on HS unit side chains.

Minimum inhibitory concentration for bacterial growth.

Polymer concentration for 10% lysis of hRBCs.

Figure 5

Hemolytic profiles of (left) DM + CH + HS and (right) DM + CH + HG ternary copolymers vs DMCH.

Conclusions

We have explored a new, biologically inspired approach to creating sequence-random copolymers that mimic the antibacterial/hemolytic activity profile characteristic of host-defense peptides. In contrast to previous efforts focused on binary copolymers in which one subunit is hydrophobic and the other cationic, we have examined ternary copolymers in which hydrophobic and cationic components are augmented by “neutral” components, i.e., components that are neither charged nor hydrophobic. On the basis of amino acid residue prevalence among HDPs, we selected two new subunits, one glycine-like (HG) and one serine-like (HS). Starting from previously known binary hydrophobic–cationic nylon-3 copolymers that display strong antibacterial activities but are also highly hemolytic, we have shown that partial replacement of hydrophobic subunits, cationic subunits, or both can lead to a decline in an unfavorable property (hemolysis) while a desirable property (antibacterial activity) is maintained. These findings suggest that the exploration of ternary copolymers may be useful for other efforts to optimize synthetic biomaterial performance.