Sequence-Defined Backbone Modifications Regulate Antibacterial

Jan 9, 2017 - In response to the urgent need for new antibiotic development strategies, antimicrobial peptides (AMPs) and other synthetic polymers are...
4 downloads 5 Views 2MB Size
Articles pubs.acs.org/acschemicalbiology

Sequence-Defined Backbone Modifications Regulate Antibacterial Activity of OligoTEAs Mintu Porel,†,‡ Dana N. Thornlow,†,‡ Christine M. Artim,† and Christopher A. Alabi*,† †

Robert Frederick Smith School of Chemical and Biomolecular Engineering, Cornell University, Ithaca, New York 14853, United States S Supporting Information *

ABSTRACT: In response to the urgent need for new antibiotic development strategies, antimicrobial peptides (AMPs) and other synthetic polymers are being actively investigated as promising alternatives to traditional antibiotics. Although most AMPs display lytic activity against several types of bacteria, they have poor toxicology profiles and are susceptible to proteolysis in vivo. While many synthetic variants have been created to mimic AMPs by tuning the hydrophobic to cationic ratio of the side-chain groups, few have decoupled the effects of charge from hydrophobicity in discrete systems, and none have investigated the effect of backbone hydrophobicity. We recently developed a rapid and efficient approach for the assembly of synthetic sequencedefined oligothioetheramides (oligoTEAs) that are resistant to protease activity. Our oligoTEA assembly scheme allows direct access to the oligomer backbone, which enables precise tuning of oligoTEA hydrophobicity while keeping charge constant. In this study, we synthesized a new class of antibacterial oligoTEAs (AOTs) with precise control over backbone hydrophobicity and composition. Our studies suggest that AOTs lyse cells via membrane permeabilization and that hydrophobicity and macromolecular conformation are key properties that regulate AOT activity. Some of our AOTs show highly promising antibacterial activity (MIC ∼ 0.5−5 μM) against clinically relevant pathogens in the presence of serum, with little to no toxicity against RBCs and HEK293 cells. Taken together, our data identify design parameters and criteria that may be useful for assembling the next generation of potent and selective AOTs.

T

hydrophobic and cationic residues.3−5 AMPs are natural components of the innate host defense system and are capable of modulating host immune response as well as disrupting the lipid bilayer in the bacterial cell membrane.4−6 Bilayer disruption is the primary mode of cell death and occurs via nonspecific interactions with the bacterial membrane, leading to membrane permeabilization and cell death.5 Additional events following bilayer disruption that may contribute to cell death include membrane depolarization and binding to cytoplasmic components.7,8 AMP specificity for bacteria over mammalian cells has been attributed to differences in surface charge and the large amount of cholesterol in the mammalian cell membrane, which has been shown to suppress bilayer disruption.9,10 Furthermore, since membrane permeabilization is driven by nonspecific AMP interaction with several essential lipids in the

he natural ability of bacteria to evolve and develop resistance to the current suite of antibiotics is occurring at an alarming rate and posing a significant threat to public health and national security. The recent decline in the rate of new antibiotic development can be attributed to difficulties associated with their mode of discovery, structural complexity, and ability to access targets. Soil organisms, the traditional source of antibiotics, are mostly depleted due to over-mining,1 and synthetic medicinal approaches have been unable to keep pace with antibiotic need in part due to the structural complexity of these natural products. Finally, antibiotics that derive their specificity by targeting bacteria-specific intracellular proteins and nucleic acids are difficult to develop due to poor intracellular penetration of organic compounds.2 As a result, combating the decline in the rate of antibiotic development requires new modes of antibiotic discovery, preparation, and extracellular targets. Antimicrobial peptides (AMPs) are small amphiphilic peptides (12−50 residues) composed of spatially segregated © 2017 American Chemical Society

Received: September 23, 2016 Accepted: January 9, 2017 Published: January 9, 2017 715

DOI: 10.1021/acschembio.6b00837 ACS Chem. Biol. 2017, 12, 715−723

Articles

ACS Chemical Biology

Figure 1. (A) Assembly scheme for antibacterial oligoTEAs (AOTs). Fluorous-tagged allyl amine40 is subjected to a thiol−ene reaction (ia) with the desired thiol and purified via FSPE (ib; general methods in Supporting Information, page S4). The resulting purified product then undergoes a Michael addition reaction with a Boc-protected N,N-amino allylacrylamide monomer (iia), followed by purification via FSPE (iib). This process continues until the desired length is attained. The final construct is then capped via a thiol−ene reaction with Boc-protected aminopropyl-2(mercapto propane)sulfide (Supporting Information, Scheme 2), deprotected with a TFA/CH2Cl2 mixture, and purified via FSPE and HPLC. (B) Antibacterial activity against B. subtilis. This was measured as the minimal inhibitory concentration (MIC) required to kill 50% of the bacteria in a liquid culture. The backbone was held constant (propanedithiol), while the repeat unit was varied from n = 1−4 (*** indicates a p < 0.0002 relative to AOT-1mer).

Recently, our group developed a new class of sequencedefined synthetic oligomers called oligothioetheramides (oligoTEAs) based on the sequential addition of Nallylacrylamides to symmetric dithiols.40,41 The versatility of this assembly process facilitated the design and synthesis of oligoTEAs with a wide variety of sequences and compositions.42 We recently reported that cyclic oligoTEAs bearing pendant guanidinium groups have potent antibacterial activities against both Gram-positive and negative cells.42 The cyclic oligoTEA backbone was mainly composed of alkyl thioethers interspersed with tertiary amides with no intramolecular hydrogen bonding capabilities. Linear and small ring cyclic guanidinium oligoTEA constructs had the best balance between antibacterial activity and hemolysis, although the latter remained fairly high. Herein, we report the synthesis of linear amine-based antibacterial oligoTEAs (AOTs) and evaluate the relationship between AOT composition, sequence, and their ability to lyse pathogenic and nonpathogenic bacteria strains. By tuning the overall charge (pendant group modification) and hydrophobicity (backbone modifications), we created noncytotoxic AOTs with potent antibacterial activity in the presence of serum. We also demonstrate that AOTs lyse cells via membrane permeabilization and that both hydrophobicity

cell membrane, development of target resistance against this mechanism is expected to be very slow.11 On the basis of these promising attributes, including their broad-spectrum activity, AMPs have emerged as viable alternatives to conventional antibiotics. However, unlike traditional antibiotics, their potency and promise is attenuated due to their susceptibility to proteolytic degradation, toxicology profile, and high preparation costs.4,11 Toward this end, peptidomimetics with modified peptide backbones that are resistant to proteolysis such as βpeptides,12−15 α-AApeptides,16−18 γ-AApeptides,19−21 oligoacyl-lysines,22 peptoids,23 and others have been designed with helical, hydrophobic and amphiphilic properties requisite for antimicrobial activity. Furthermore, a wide variety of nondegradable synthetic polymers including polynorborenes,24,25 lipid modified cyclobutene copolymers,26 Nylon-3 copolymers,27−30 poly(methacrylates),31−33 poly(arylamides),34 and other polymers,35,36 as well as small molecular variants,37−39 have been created to mimic the properties and antibacterial activities of AMPs. Although these synthetic constructs are simpler and inexpensive to synthesize, they lack the precise sequence definition of peptides and peptidomimetics. 716

DOI: 10.1021/acschembio.6b00837 ACS Chem. Biol. 2017, 12, 715−723

Articles

ACS Chemical Biology

Figure 2. (A) General structure of the AOTs used in this work. The three central dithiol R groups represent different substitution patterns with functional groups 1−6. aThe relative retention time of dithiols 1−6 after a single modification with N,N-dimethylallylamine (see Supporting Information Figure 6 for details). (B) Correlation between the relative hydrophobicity predicted from the summation of the individual dithiol comonomer retention times and the actual AOT retention time measured via HPLC for all 14 AOTs (AOT111-AOT666). The orange band represents the 95% confidence interval.

varying degrees of hydrophobicity (Figure 1, green). All AOTs in this manuscript were designed to be symmetric. To determine the length required for potent antibacterial activity, we synthesized and characterized four AOTs with increasing oligomer lengths from a 1-mer to a 4-mer (Figure 1, Supporting Information Table 1) and evaluated their antibacterial potency against B. subtilis. All AOTs in this series had a propyl backbone (R = (CH2)3) and differed only in the overall number of amine and backbone residues. The data in Figure 1b clearly show increasing activity with increasing oligoTEA repeat units. We settled on the AOT-4mer (8-unit oligomer) for further evaluation based on its potency and reasonable assembly yield. The potency of longer oligoTEAs, synthesized with large fluorous tags, will be explored in future studies. The AOT-4mer framework was used to create a library of 14 AOTs with a variety of dithiol co-monomers. As shown in Figure 2a, six different dithiol comonomers were substituted into the three central positions along the oligoTEA backbone and used to vary the overall hydrophobicity of the AOT. The AOTs are identified by their dithiol sequence. For example, AOT626 represents an AOT with a 1,6-hexanedithiol in the second dithiol position, a 1,3-propanedithiol in the third position, and another 1,6-hexanedithiol in the fourth dithiol position. Following purification, all 14 AOTs were characterized and confirmed via LCMS (Supporting Information Table 2). The AOT structure was confirmed to be stable to proteolysis (after incubation in serum) due to the lack of primary amides in the backbone (Supporting Information Figure 4). The hydrophobicity of each synthesized AOT was inferred from its retention time measured on a HPLC using a C18 reverse-phase column (Supporting Information Figure 5). Since all synthesized AOTs are fully symmetric, we sought to determine the predictability of the measured retention time based on the individual hydrophobicity of the dithiols. To do this, we measured the retention time of the six dithiol

and macromolecular conformation are key properties that regulate AOT activity. Taken together, our data highlight the promise and potential of future AOT designs as potent clinical antibacterial candidates.



RESULTS AND DISCUSSION Synthetic Design of AOTs. We recently invented a rapid and efficient synthetic method for the assembly of precise sequence-defined oligoTEAs with a tunable and flexible thioether backbone.40,41 This unique methodology utilizes reaction orthogonality and a soluble fluorous support to achieve precise sequence control in the liquid phase. As shown in Figure 1a, the N-allylacrylamide monomer framework (synthesis and characterization in Supporting Information Scheme 1, Figures 1 and 2) includes a reactive acrylamide group that can undergo phosphine-catalyzed Michael addition with thiols, the desired tert-butyloxycarbonyl (Boc) protected amine pendant functional group, and a reactive allyl group that can undergo photoinitiated thiol−ene “click” addition with another thiol. The step-by-step iterative synthetic scheme is illustrated in Figure 1a. Following the reaction of each monomer to the fluorous-tagged construct (Figure 1a, ia and iia), a fluorous solid-phase extraction (FSPE) technology43,44 is used to purify the fluorous tagged oligomeric product (Figure 1a, ib and iib). Fast solution-phase kinetics of both reactions, coupled with the FSPE technology, makes this platform a unique and efficient approach for synthesizing sequence-defined oligoTEAs. The final oligoTEA construct is obtained following acid deprotection of the fluorous tag and Boc groups (Figure 1a, iii) and HPLC purification (Supporting Information Scheme 2). Almost all AMPs have a net cationic charge and significant hydrophobic domains. In an effort to mimic these chemical features, we designed our AOTs to contain cationic pendant groups using an amino-N-allylacrylamide monomer (Figure 1, red) and a tunable hydrophobic backbone using a dithiols with 717

DOI: 10.1021/acschembio.6b00837 ACS Chem. Biol. 2017, 12, 715−723

Articles

ACS Chemical Biology

Figure 3. MIC and hemolysis data for AOTs. Heat map representation of antibacterial activity of AOT111−AOT666 against four bacterial strains: B. subtilis, S. epidermidis, MRSA, and VRE. The deeper red color indicates higher antibacterial potency (lower MIC value). The numbers in the heat map are the respective MIC values (in μM). Hemolytic activity (% hemolysis relative to a 0.1% triton X control) of AOT111−AOT666 is shown in gray scale (right). The numbers are the % hemolytic values relative to a 0.1% triton-X control.

comonomers after modification with N,N-dimethylallylamine (used to enhance ionization in the MS) via LCMS (Supporting Information Figure 6). The retention times were normalized to L-dithiothreitol, and these values were used to compute and infer a relative AOT hydrophobicity by simple addition of the dithiol comonomer retention time. For example, AOT-111 (L) will have a calculated relative retention time value of 3.00 (1.00 + 1.00 + 1.00, Figure 2), while AOT525 will have a calculated relative retention time of 7.07 (2.54 + 1.99 + 2.54, Figure 2). As shown in Figure 2b, the calculated values show an excellent linear correlation with the measured retention times. We hope to use this as a calibration curve in the future to design and predict the relative hydrophobicity of symmetrical amine-based oligoTEAs. To examine the effect of chirality on the chemical properties of AOTs, AOT212 was prepared with a racemic DLdithiothreitol and a chiral L-dithiothreitol (Figure 2, compound 1 DL and L). AOT212 prepared with L-dithiothreitol retained its chirality (Supporting Information Figure 7) and had almost an identical retention time relative to its racemic form. AOTs Are Bactericidal and Selective. In previous studies, we showed that cationic linear and cyclic oligoTEAs display

higher activity against Gram-positive vs Gram-negative bacteria.42 Since Gram-positive cocci are the leading cause of bacterial infections in humans, we decided to focus our studies on Gram-positive bacteria. AOTs were tested on two pathogenic strains, methicillin-resistant Staphylococcus aureus (MRSA) and vancomycin-resistant Enterococcus (VRE), estimated to have caused 12 000 deaths in 2013 alone,45 and two nonpathogenic strains, B. subtilis and S. epidermidis. The antibacterial activities, reported as the minimum inhibitory concentration (MIC) of all 14 AOTs against the four bacterial strains are shown in Figure 3 (Supporting Information Figures 8−11). To place these activities in perspective, we also tested a negative control, a peptide mimic with six lysine units (Figure 3, (Lys)6), and three positive controls, melittin, vancomycin, and rifampicin. The heat map in Figure 3 reveals a number of interesting trends. First, the relative intensity of the heat map indicates that the pathogenic bacteria strains, MRSA and VRE, are much more difficult to kill than the nonpathogenic strains, B. subtilis and S. epidermidis. Second, we observe that relatively hydrophilic compounds, characterized by a retention time of 9 min or less, including AOT121, AOT111, and the (Lys)6 control, have poor activity across all strains (Supporting 718

DOI: 10.1021/acschembio.6b00837 ACS Chem. Biol. 2017, 12, 715−723

Articles

ACS Chemical Biology

Figure 4. Effect of AOTs on growth of intact cells. (A) Cell growth kinetic measurements (kill curves) of B. subtilis exposed to AOTs, vancomycin, rifampicin, and melittin. Fifteen times the MIC of the respective compounds was used. (B) The growth curves in part A were fit to a general logistic equation. The death rate extracted from this fit is plotted against the AOT retention time measured via reverse-phase HPLC. Error bars represent the standard error of death rate values from the fit of the logistic equation to the growth curve.

contrast to the 1 h incubation period for the hemolysis assay. AOT212, AOT222, AOT151, AOT161, AOT252, and AOT262 were selected as promising candidates based on their high antibacterial potency and low to mid hemolytic activity. Results from the MTS experiment, also performed at 100 μg/mL, showed that non-hemolytic AOT212 and AOT222 and slightly hemolytic AOT151 and AOT161 had no significant cytotoxicity to HEK293 cells relative to the blank control, while the moderately hemolytic AOT252 and AOT262 were severely toxic to HEK293 cells (Supporting Information Figure 13). These collective data suggest that the backbone hydrophobicity of AOTs plays a large role in their antibacterial activity and cytotoxicity. Hydrophobicity beyond a certain threshold is needed for antibacterial activity; however, further increases beyond this threshold erodes AOT selectivity for bacteria cells due to increasing hemolytic activity. The data also indicate that AOT backbone modifications can alter AOT conformation and have a significant effect on antibacterial activity. Bacterial Cell Death Kinetics Correlates with Hydrophobicity. To investigate the mode of killing, we performed growth/kill kinetics measurements with B. subtilis in log phase (OD600 = 0.5) exposed to AOTs and the antibiotics vancomycin, melittin, and rifampicin as mechanistic controls. Melittin is a traditional membrane permeabilization agent that undergoes rapid pore formation when dosed at a concentration above its MIC value.46 Vancomycin on the other hand is a cellwall interfering agent known to bind to the pentapeptide sequence of lipid II.47 Rifampicin is a nonlytic bacteriostatic antibiotic that primarily targets replication.48 All three antibiotics display distinct differences in their kinetic behavior as shown in Figure 4a. Kill-kinetics studies were performed at 15 × MIC with all “active” AOTs, i.e., those with MIC values less than 5 μM (Figure 4a). The kinetic curves in Figure 4 indicate that AOTs with increasing hydrophobic character display faster kill kinetics. To quantify this qualitative observation, we fit the kill-kinetic curves to a general logistic equation (Supporting Information Figure 14) and performed a nonlinear regression (GraphPad Prism 7.0a) to extract the steepness parameter,

Information Figure 12). Above this 9 min threshold, antibacterial activity against B. subtilis, S. epidermidis, and MRSA appears to be largely insensitive to AOT hydrophobicity. However, AOT activity against VRE does appear to increase with increasing hydrophobicity. Two oligomers, AOT232 and AOT242, are exceptions to these general trends. These AOT structures differ from AOT222 by a single substitution at the central dithiol comonomer: AOT232 has an ethylene oxide group and AOT242 has a meta-substituted phenyl group. The 6−10-fold difference in antibacterial activity between AOT242 and AOT252 (meta vs para substituted) highly suggests that conformation plays a significant role in their mode of action. We hypothesize that AOT232 and AOT242 adopt a structural motif that prevents their interaction with the bacterial cell membrane. To examine the effect of chirality on the biological properties of AOTs, we compared the antibacterial activity of the racemic AOT212 (DL) with its chiral form, AOT212 (L). The data in Figures 3 and 4b show that the chiral and racemic versions of AOT212 have similar antimicrobial activities and kill kinetics, thus indicating that the chirality of AOT212 does not appear to play a significant role in its mode of action. Finally, the extent of AOT cytotoxicity against mammalian cells was measured via a hemolysis assay at an elevated dose of 100 μg/mL (Figure 3). The data show that AOTs containing any combination of the relatively hydrophilic dithiol comonomers, 1DL, 1L, 2, and 3 were non-hemolytic (