Mimetics of the AMPs
Given their broad specificity, amphiphilic antimicrobial peptides appear to be ideal therapeutic agents. Over the last decade, there has been considerable interest in developing AMPs as intravenously administered antibiotics. However, despite extensive efforts in the pharmaceutical and biotechnology industry, it has proven difficult to accomplish this goal.[2
] As a first step in this direction, various workers designed antimicrobial peptides by idealizing the amphiphilic alpha-helical arrangement of sidechains[19
] observed in the natural structures, leading to a large number of potent and selective antimicrobial compounds.[3
] Also, cyclic peptides,[22
] turn-forming, and hairpin forming peptides[19
] have been extensively studied, but are beyond the scope of this review [see [25
] for a recent review].
Peptidomimetic approaches that mimic the peptide primary structure have been pursued using amide bond isosteres or modifications of the peptide backbone by chain extension or heteroatom incorporation.[26
] Several groups designed β-peptides (peptides built from β-amino acids with an additional methylene group in their backbone relative to α-amino acids), which maintained this amphiphilic architecture.[27
] A systematic analysis of β-peptide structure and activity revealed that facial amphiphilicity as well as a precise ratio of charged to hydrophobic residues was essential for potent and selective activity, similar to what has been described for the antimicrobial α-peptides.[30
] Both diastereomeric (D-amino acid substitutions) peptide sequences and α/β-peptides that did not adopt globally amphiphilic helices showed antimicrobial activity suggesting a well-defined amphiphilic secondary structure was not essential.[29
] Although these peptides may be unable to adopt regular α-helix structures, their potent antibacterial activity may well require the presence of highly flexible backbones to permit formation of irregular but globally amphiphilic conformations upon interaction with membranes. Helical, cationic, facially-amphiphilic peptoid mimics (poly-N-substituted glycine) of magainin-2 have also been produced that show selective and potent antibacterial activity against Gram positive and Gram negative bacteria.[33
] Significantly, the β-peptides and peptoids are resistant to proteases, providing important advantages over alpha-peptides. However, β-peptides and peptoids are relatively difficult and expensive to synthesize in large quantities.
Shai and coworkers have explored the use of short basic peptides with an N-terminal fatty acyl chain as potential antimicrobials[34
] (for a review see: [35
]). Linear oligomers containing alternating acyl chains and cationic amino acid (lysine) residues have been shown to be active against Gram-negative bacteria. The addition of an acyl group to the N-terminus enhanced potency, leading to a molecule active in vivo in a mouse peritonitis models.[53
]. Recently, this group described a series of palmitoylated di, tri, and tetrapeptides[36
] with varying specificities for a variety of bacteria and fungi. These lipopeptides assemble into nanostructures that might be related to their biological activities.
Non-peptidic compounds that are facially amphiphilic have also been produced from a steroid scaffold.[38
] These compounds, termed ceragenins, have antibacterial activities similar to AMPs, but unsatisfactory safety profiles have limited their use to topical and medical devise applications. [40
] Polymeric molecules have attracted significant attention recently as progress toward biomimetic macromolecules continues. Several polymers have been synthesized that likely adopt amphiphilic structures in membrane environments and further confirm that well-defined amphiphilic secondary structures are not essential. These include phenylene ethynylene,[41
] short polymethacrylate,[44
] and polynorbornene backbones[47
] that are active against Gram-positive and Gram-negative bacteria and have lower hemolytic activities against human erythrocytes. Such compounds are not preferred for systemic therapeutic use but their ease and economical cost of synthesis make them highly suitable for many material applications to create antimicrobial surfaces and products.[51
Design of aryl and heteroaryl-amide antimicrobial peptide mimetics
The goal of our synthetic approach was to capture the structural and biological properties of antimicrobial peptides within the framework of inexpensive oligomers.[55
] While it has been possible to solve some relatively simple problems such as susceptibility to proteolysis, more severe problems have included the expense of the materials, toxicity, limited efficacy, and limited tissue distribution. We reasoned that small synthetic oligomers that adopt amphiphilic secondary structures while exhibiting potent and selective antimicrobial activity would be less expensive to produce, they might have better tissue distribution, and that it would be much easier to fine-tune their structures and activities to minimize toxicity. Therefore, an active program to de novo
design synthetic mimics of antimicrobial peptides (SMAMPs) from inexpensive synthetic oligomers was undertaken (see ). This effort has led to the identification of a class of SMAMPs that possess efficacy, safety and pharmaceutical qualities suitable for development as intravenous antibiotics.
In general, de novo
design of oligomers was accomplished in several steps. We first defined a three-dimensional framework, or backbone, using molecular dynamics and quantum force field calculations to assure it was able to assume a semi-rigid conformation.[55
] Next, side groups were added computationally to maximize diversity and maintain drug-like properties. The best combinations of functional groups were then computationally selected to produce a cationic, amphiphilic structures.[59
] Representative compounds were synthesized from this selected library to evaluate their activities. In our initial studies, we prepared compounds related to the core structure 1
, which contained alternating 1,3 phenylene diamine units connected by a isophthalic acid.[55
Once an active series had been established, the activity was optimized by increasing the rigidity of the backbone through hydrogen-bonding and/or introduction of new substituents as shown with the arylamides in . Hydrogen bonding in the backbone does not appear to be essential as the phenylene ethynylene was highly active.[61
Representative SMAMPs highlighting design principles leading to potent in vivo activity.
Mechanism of Membrane Activity
Understanding the interactions between SMAMPs and lipid bilayers are a complicated, multi-component problem yet scientifically important. Beyond understanding the structure-function correlation of SMAMPs, learning how these unique molecules interact with lipid membranes will contribute to our fundamental understanding of molecular-membrane self-assemblies which are critically important for numerous biological processes. Bacterial membranes and mammalian cell membranes are very distinct with respect to their phospholipid compositions.[64
] It was shown that the exact lipid type influenced the leakage of calcein loaded vesicles and that specifically the presence of ‘bacterial’ type lipids was essential. For example, the phenylene ethynylene SMAMP () exhibited ~75% dye leakage from PE/PG vesicles versus only 10 - 15% leakage against PC/PS and PC/PG vesicles, despite the presence of the same negatively charged lipid mol%.[65
] Very similar results were obtained with SMAMPS in the arylamide series where leakage activity was again dependent on the volume fraction of negative intrinsic curvature PE. This dependence on negative intrinsic curvature lipids is consistent with the observed biological activity of this SMAMP as bacteria contain much higher volume fractions of negative intrinsic curvature lipid than mammalian cells. Synchrotron small-angle X-ray scattering (SAXS) studied further confirmed these findings and showed how this SMAMP interacts with membranes of four different lipid compositions.[66
] This SMAMP induced an inverted hexagonal phase (HII
) in PE/PG vesicles, but only when the PE content was above a minimum threshold concentration of 64%. It was concluded that the ability of the SMAMP to restructure membranes critically depends on the concentration of negative intrinsic curvature PE present in the membrane. These results strongly support the hypothesis that exact lipid type may be essential for the SMAMPs ability to distinguish microorganisms from mammalian cells.
Antimicrobial activities of lead oligomers
Structure activity relationships in the arylamide series identified lead compounds that were optimized for Gram-positive activity. illustrates results from a susceptibility screen with two of these lead compounds, PMX30016 and PMX10129, against Gram-positive and Gram-negative human clinical isolates. Both compounds exhibited broad coverage against the Gram-positive pathogens with S. aureus and coagulase-negative Staphyloccal species showing the lowest MICs (0.5 ug/ml). S. pneumoniae and S. viridians were the least susceptible with MICs of 8 and 8 to 16 ug/ml, respectively. For the Gram-negative pathogens, PMX30016 had more extensive coverage than PMX10129 but overall coverage was less than for the Gram-positive organisms.
Table 1 Clinical isolate screen vs. Gram-positive and Gram-negative pathogens. The in vitro susceptibility assays were performed in accordance with defined CLSI guidelines for the specific organisms tested. PMX30016 and PMX10129 are members of the arylamide series. (more ...)
These SMAMPs are very stable in plasma and MICs are not affected by the addition of pooled human plasma at 40 to 60% concentrations. Full antimicrobial activity is retained under physiological salt concentrations. The arylamides are bactericidal and reductions in viable bacteria of greater than 3 logs (2× to 4× MIC concentrations) occur in 30 minutes to 6 hours upon exposure, depending on the compound and organism. The time-kill kinetics for the oligomers are concentration-dependent and appear to be somewhat slower than those for many of the AMPs which are reported to be ≤ 30 minutes. The reasons for this are unclear but may be related to the smaller size of the SMAMPs (MWs = 520 to 960 D) since polymeric mimetics with MWs > 2,000 D show time-kill kinetics of less than 30 minutes, despite having higher MICs (6 to 25 ug/ml).
Cytotoxicity of PMX30016, PMX10129 and PMX30006, a progenitor of PMX30016, were evaluated using mammalian cell lines (). Although antimicrobial activities between the three SMAMPs are similar, PMX30016 and PMX10129 are significantly less cytotoxic versus mammalian cells and have superior selectivity indices (ratios of EC50 values for cytotoxicity to MIC for MRSA). Factors that have been associated with enhanced selectivity for bacterial cells over mammalian cells are the number of positive charges, identity of the positively charged endgroup, rigidity across the backbone, hydrophobicity and stiffness of the sidechains.
Table 2 Selectivity Indices between MRSA and mammalian cells. Cytotoxicity was evaluated in a colorimetric assay using a transformed human liver cell line (HepG2 cells) and an embryonic mouse cell line (NIH/3T3 cells). This assay measures the bioreduction of (more ...)
To investigate the potential for bacteria to develop resistance against the antimicrobial oligomers, serial passage experiments in the presence of sub-lethal SMAMP concentrations were performed with S. aureus
, MRSA and P. aeruginosa
. shows results for SMAMPs PMX10066, PMX10072 and PMX70004 up to 17 passages with S. aureus
ATCC 29213. MIC values were determined at each passage and the culture at 0.5× the MIC was selected for further passage. As a control, parallel cultures were also exposed to 0.5× MIC concentrations of ciprofloxacin or norfloxacin, two broad spectrum fluoroquinolones for which resistance has been reported in this experimental format.[67
] There is no change in the MICs for PMX70004, PMX10072 or PMX10066 over the entire 17 passage time course. Conversely, an increase in the MIC is readily observed by passage seven for both ciprofloxacin and norfloxacin. These data are very similar to results from resistance studies with the AMPs and together demonstrate the difficulty bacteria have in building resistance to this mechanism of cell killing.
Figure 3 The development of bacterial resistance with three different classes of SMAMPs and two fluoroquinolones against S. aureus ATCC 29213. Serial passage and MICs were performed in microtiter panels containing antimicrobials, each over a range of doubling (more ...)
Animal efficacy studies
PMX30016 and PMX10129 have been evaluated for in vivo
efficacy versus S. aureus
in a mouse thigh burden model.[68
] Robust efficacy is observed with both PMX30016 and PMX10129. Efficacies for PMX30016 exceed 5 log10
and 4 log10
reductions in viable bacterial cell counts (colony forming units, cfu) with total doses of 10 mg/kg (5 mg/kg/dose) and 4 mg/kg (2 mg/kg/dose), respectively (see ). PMX10129 is slightly less efficacious in this dosing format. Comparable efficacy was achieved with PMX30016 against methicillin-resistant S. aureus (MRSA ATCC33591) in the mouse thigh burden model done under similar conditions (not shown). The activity of the lead SMAMPs in the mouse thigh burden when administered by iv injection distinguishes their efficacies from many of the AMPs that lack efficacy when administered parentally in different compartments from the infection. [32
] This robust efficacy, likely due to their smaller size and abiotic structures which enhance bioavailability and stability in vivo
, illustrates the power of this SMAMP approach.
Figure 4 In vivo efficacy in the mouse thigh burden model. Neutropenic mice were inoculated in their thigh muscles with S. aureus ATCC 13709 and at one and seven hours after infection the mice were administered compound by iv injection (n = 4/group). At 25 hours (more ...)
Animal safety studies
Single dose toxicity studies were conducted to define the maximum tolerated dose (MTD) where no adverse signs are evident over a 4 to 7 day period following SMAMP administration (). Compounds were administered by either IV bolus injection in the tail vein of mice and rats or by IV infusion via femoral vein catheter in rats. For PMX10129, the MTD in rats is 10 mg/kg by IV bolus injection, 20 to 30 mg/kg by 1 hour IV infusion and 45 mg/kg by 4 hour IV infusion. For PMX30016, the MTD is 15 mg/kg by IV bolus injection and 20 to 30 mg/kg by 1 hour IV infusion. The sequentially higher MTDs observed for these SMAMPs when administered by IV bolus or IV infusions indicate that the MTD-related adverse events are primarily dependent on maximum concentrations attained in the blood (Cmax) rather than the total dose of drug (AUC).
Table 3 Maximum Tolerated Doses in mice and rats. Mice or rats where administered a single dose of compound and clinical signs were recorded over a 4 to 7 day period following compound administration. Gross necropsy was also performed at the conclusion of the (more ...)