Designing New Hybrid Antibiotics: Proline-Rich Antimicrobial Peptides Conjugated to the Aminoglycoside Tobramycin

Resistance to aminoglycoside antibiotics is a serious problem, typically arising from inactivating enzymes, reduced uptake, or increased efflux in the important pathogens for which they are used as treatment. Conjugating aminoglycosides to proline-rich antimicrobial peptides (PrAMPs), which also target ribosomes and have a distinct bacterial uptake mechanism, might mutually benefit their individual activities. To this aim we have developed a strategy for noninvasively modifying tobramycin to link it to a Cys residue and through this covalently link it to a Cys-modified PrAMP by formation of a disulfide bond. Reduction of this bridge in the bacterial cytosol should release the individual antimicrobial moieties. We found that the conjugation of tobramycin to the well-characterized N-terminal PrAMP fragment Bac7(1–35) resulted in a potent antimicrobial capable of inactivating not only tobramycin-resistant bacterial strains but also those less susceptible to the PrAMP. To a certain extent, this activity also extends to the shorter and otherwise poorly active fragment Bac7(1–15). Although the mechanism that allows the conjugate to act when its individual components do not is as yet unclear, results are very promising and suggest this may be a way of resensitizing pathogens that have developed resistance to the antibiotic.


■ INTRODUCTION
Aminoglycosides (AGs) are a group of structurally different amino-modified sugars and represent a clinically important class of antibiotics for their broad spectrum of activity against a broad range of pathogenic bacteria. 1−3 Their mechanism of action is based on inhibition of bacterial protein synthesis by binding to the 16S rRNA of the bacterial minor ribosomal unit. 4−6 Despite their efficacy, antibacterial resistance (AMR) to AGs has dramatically increased due to enzymatic inactivation, target modification, reduced uptake and/or drug efflux. 7−9 AMR represents a serious problem when AGs are used to treat chronic pathologies such as cystic fibrosis (CF), 10 which is the case with tobramycin. 11 This antibiotic is part of the standard-of-care in CF patients for dealing with pulmonary infections caused by Pseudomonas aeruginosa. 12 Maintaining the efficacy of tobramycin despite growing bacterial antibiotic resistance would therefore be very important for several infectious diseases.
Antimicrobial peptides (AMPs) are immune effectors that can prevent or combat microbial infections as important components of the innate host defense in multicellular organisms. 13,14 Most AMPs interact with the microbial surface and act by compromising the integrity of cellular membranes and/or interfering with cell wall synthesis, 15,16 which is generally reflected in a broad spectrum of antimicrobial activity. 17 Deshayes et al. conjugated redesigned membraneactive AMPs displaying different sequences, hydrophobicity and helical amphiphilicity, with tobramycin. 18,19 The purpose of this kind of research was to obtain a unimolecular but multifunctional drug with a broad-spectrum antibiotic activity against antibiotic-resistant strains, using different and synergistic mechanisms to kill them.
In this scenario, Proline-rich AMPs (PrAMPs) may be of interest for conjugate development, since they are characterized by a nonlytic mode of action and are selective for some species of Gram-negative bacteria. 20−22 Two distinct mechanisms of action have been observed for PrAMPs, both involving protein synthesis: (i) type I PrAMPs allow initiation of protein synthesis but prevent the transition into the elongation phase by hindering the accommodation of tRNA in the A-site, 23−27 whereas (ii) type II PrAMPs allow initiation and elongation of protein synthesis but hinder termination of translation by trapping release factors in the ribosome. 28,29 Bac7, a 60-residue, linear peptide isolated from bovine neutrophils, and especially its fragments, are among the beststudied PrAMPs. 20,30 The derivatives Bac7(1−16) and Bac7 (1−35), corresponding to the 16 and 35 N-terminal residues, respectively, showed comparable antimicrobial activity to native, full-length Bac7. 31−33 On the other hand, removal of only one further C-terminal residue (Arg) from Bac7(1−16) significantly reduces antimicrobial activity. 31,33 Furthermore, it has been found that to interact with the ribosome and inhibit protein synthesis, Bac7 functional fragments cross the bacterial inner membrane via the SbmA transporter, without permeabilizing it at active concentrations. For this reason, they may be termed bacteria-penetrating peptides (BPPs), as well as AMPs.
We anticipated that the BPP and AMP properties of Bac7 derivatives could be exploited to obtain new bifunctional hybrids by covalently linking them to tobramycin. This would allow targeting bacterial ribosomes via two different inhibitory mechanisms, while likely utilizing different cell internalization mechanisms. To this end we synthesized peptide chimeras derived from both Bac7(1−15) and Bac7(1−35) conjugated with tobramycin, and subjected them to preliminary testing. We linked them by introducing a disulfide bond between tobramycin and PrAMP to develop a conjugate that could be active as such, but with an easily cleavable covalent linker that would also allow release of the active components in the reductive intracellular environment. Should the system work, the aim is to eventually design a system that could do this while avoiding premature release in the extracellular medium or in the blood. 34 ■ RESULTS AND DISCUSSION Synthesis of the Tobramycin-PrAMP Conjugates. The primary hydroxyl group at the 6′ position in tobramycin was chosen as the conjugation point between the peptide and tobramycin (Tob) because of its expected higher relative reactivity (compared with the other secondary hydroxyls present) and because functionalization is unlikely to compromise antimicrobial activity. 18 The primary hydroxyl of tobramycin and other aminoglycosides is in fact not essential for RNA binding. 6,35 On the other hand, the amino groups are important for antibiotic activity, due to H-bond formation and electrostatic interaction with the ribosome, so they were protected with the tert-butyloxycarbonyl (Boc) group to give (Boc) 5 Tob (1). 36,37 The primary 6′-hydroxyl was then selectively functionalized using succinic anhydride to introduce a terminal carboxyl function (2) 19 that allows for coupling with the α-amine of an amidated Cys residue (3) (Scheme 1).
The reaction between succinic anhydride and compound 1 was carried out in anhydrous toluene with 4-(dimethylamino)- pyridine (DMAP) at 85°C overnight and was monitored by MS. After purification, the yield of the modified Tobramycin (mTob) (2) was ∼60%, in agreement with the literature. 19 The presence of byproducts was revealed by mass spectroscopic analysis, which indicated the presence of both single (M 1068.5) and double hemiacylation (M 1168.6). The modification and protection of Tob allowed its direct use in solid-phase peptide synthesis (SPPS), so that compound 2 was coupled to the α-amine of a Cys residue bound to Rink amide resin, adding 2 equiv, with PyBop (0.98 equiv) and DIPEA (2 equiv) as coupling reagents and monitoring the reaction with the Kaiser Test. The product (3, mTob-Cys-NH 2 ) was then cleaved from the resin using a standard cleavage mixture (Scheme 1c) and the crude material was precipitated with 20 mL of cold tert-butyl methyl ether and collected in good crude yield (>90%), indicating efficient coupling under these conditions. ESI-MS (single peak at M 670.5) corresponded to compound 3, and analytical RP-HPLC revealed a sufficient purity for it to be used in subsequent synthetic steps without purification.
For direct coupling of the antibiotic to the selected PrAMPs, the peptides Bac7 (1−35) 16 ]}. These were synthesized in SPPS using the Trityl and Rink Amide resins respectively, resulting in a peptide amide for the shorter peptide (see Table 1). After purification by RP-HPLC, the correctness of the peptides was confirmed by ESI-MS analysis at M 4310.8 and at M 2022.3, respectively. A small aliquot of both peptides was alkylated with 2iodoacetamide and served as controls during biological assays. These are designated as Bac7 (1−35)[Cys 36 ALK] and Bac7 (1− 15)[Cys 16 ALK], respectively (see Table 1). Alkylation was confirmed by ESI-MS analysis (M 4368.0 and 2079.5, respectively). To promote the formation of the heterodimer formation during this step, the sulfhydryl group of the Cys residue linked to Tob was preactivated with 2,2′-dithiodipyridine 38 (Scheme 2).
This reaction was relatively straightforward, and the ESI-MS spectrum, with a major peak at M = 779.5, confirmed the presence of compound 4 in 70% yield. This was then conjugated to Bac7(1−35)[Cys 36 ]-OH or to the shorter  Figure) to yield (6), 12.5%  16 ]-NH 2 in a manner similar to that previously described for linking these peptides to fluorescent dyes. 33 The reaction was carried out at 1:1 molar ratio of antibiotic and peptide and at a relatively high dilution to limit peptide homodimerization, and was monitored to completion by analytical RP-HPLC. The products were then purified by preparative RP-HPLC and corresponded to the desired heterodimers as confirmed by ESI-MS: M 4978.5 and 2689.9 for mTob-Bac7(1−35)[Cys 36 ]-OH (5) and mTob-Bac7(1− 15)[Cys 16 ]-NH 2 (6) respectively (see Table 1). Antimicrobial Activity. The antimicrobial activity (MIC) of the synthesized conjugates was evaluated against the wellcharacterized reference strain E. coli BW25113. Unmodified tobramycin and the Cys-alkylated form of the peptides Bac7(1−35)[Cys 36 ALK]-OH and Bac7(1−15)[Cys 36 ALK]-NH 2 were used as the free antibiotic and peptide controls, respectively (the use of compound 3 with a free thiol administered in the bacterial growth medium was inappropriate and thiol-alkylated 3 lost activity, see Supporting Information S1  36 ]-OH retained the same activity of the aminoglycoside. To rule out the possibility that the antibacterial effects observed with the conjugated compounds were actually due to synergistic effects of tobramycin and PrAMPs released in the medium, we performed a checkerboard assay with Bac7(1− 15)[Cys 16 ALK]-NH 2 and tobramycin, on the E. coli BW25113ΔSbmA strain, where a significant increase in antimicrobial activity was observed on conjugates with respect to the single compounds. The results showed that the MIC of tobramycin against E. coli did not change with increasing concentration of Bac7(1−15)[Cys 16 ALK]-NH 2 , indicating that there was no synergistic effect between the two compounds (see Supporting Information S2).
Transport across the inner membrane of some Gramnegative species (e.g., E. coli and S. enteriditis) via the protein SbmA plays a central role in the mode of action of PrAMPs, 39 giving them access to bacterial ribosomes. We therefore tested whether the PrAMP component confers this internalization mechanism to Tob conjugates. For this purpose, we used a mutant bacterial strain with deletion of the gene encoding the transporter SbmA (E. coli BW25113ΔsbmA). As expected, Bac7(1−15)[Cys 16 ALK]-NH 2 proved to be inactive in the absence of SbmA and sensitivity to Bac7(1−35)[Cys 36 ALK]-OH also decreased significantly (Table 1). Interestingly, deletion of SbmA also reduced sensitivity toward tobramycin 4-fold. This is consistent with reports that mutation of the sbmA gene is common in strains adapted to amikacin, an aminoglycoside related to tobramycin. 40,41 The role of SbmA in tobramycin's mode of action therefore deserves further scrutiny. However, it was surprising that both mTob-Bac7(1− 35)[Cys 36 ]-OH and mTob-Bac7(1−15)[Cys 16 ]-NH 2 essentially retained their antibacterial activity in the absence of the SbmA transporter (Table 2). This would indicate that the conjugate can access an as yet undefined internalization route, possibly through an acquired capacity to perturb the bacterial membrane.
This possibility was investigated by propidium iodide uptake experiments on E. coli BW25113 cells treated with each of the compounds at their MIC values. The cytofluorimetric profiles (Supporting Information S3) are distinctly different to those caused by the known lytic peptide colistin, suggesting that the Tob conjugates do not acquire the capability to permeabilize the bacterial membrane. Nonetheless, they were also different to those of the unconjugated species, which were nonlytic and appeared like untreated controls. The profiles were visibly different with respect to both morphology (in terms of side scatter, SSC) and the PI fluorescence. This suggests that the conjugates have acquired the capacity to somewhat perturb the membrane, while not fully permeabilizing it. Should this be related to an increased cell-penetration capacity, it would help explain their activity toward the E. coli BW25113ΔsbmA.
In any case, an SbmA-independent mode of action indicates an advantage of conjugates over ordinary PrAMPs. First, it may be more difficult for bacteria to develop resistance to these compounds, by mutating this nonessential gene. Second, this appears to be a robust property shared by the shorter and less active 1−15 peptide component, when linked to tobramycin. Third, it would broaden the activity spectrum to bacteria that do not normally carry a gene for this transporter. This is the case with Pseudomonas aeruginosa, and indeed most strains of this pathogen are weakly susceptible to PrAMPs. 42 Until the spread of resistance to this antibiotic, the bacterium was instead efficiently controlled by tobramycin, especially in CF and other chronic diseases. 43,44 It was therefore of interest to assess if mTob-Bac7(1−15)[Cys 16 ]-NH 2 and mTob-Bac7(1− 35)[Cys 36 ]-OH were active against a panel of clinically isolated P. aeruginosa strains, some of which multidrug resistant (e.g., PA10, PA21, and PA-35, see Table 2), and comparing the activity to unconjugated antibiotic and peptide components. The ATCC 27853 strain was included as the reference strain in the screening.
The susceptibility of the different P. aeruginosa strains to unconjugated tobramycin and the Bac7 fragments was found to be quite variable. As expected, the reference ATCC 27853 strain was quite susceptible to tobramycin, but not to Bac7 Overall, the tobramycin-PrAMP conjugates effectively inhibited E. coli both in the presence and the absence of the SbmA transporter and all seven P. aeruginosa strains, with MIC values ranging from 1 to 4 μM, indicating that the conjugates are generally as potent and broad-spectrum antimicrobials as the constituent molecules, more effective in a few interesting cases.

Bioconjugate Chemistry pubs.acs.org/bc Article
In summary, by covalently binding tobramycin to active or inactive Bac7 fragments using versatile synthetic routes, we were able to prepare interesting new antibiotic compounds, and showed that they can overcome the insensitivity of bacterial strains to individual molecules. This favorable effect was observed for reference as well as clinically isolated Gramnegative bacterial strains (E. coli and P. aeruginosa) and extends to other Gram-negative species (A. baumanii and S. enteritidis). Strategies to enhance antibacterial activity by linking dualacting antimicrobials, resulting in so-called antimicrobial hybrids, have been widely demonstrated. 45−47 This could be due to reduced susceptibility to degradation by bacterial enzymes or to efflux systems, and/or to improved cell penetration properties of the hybrid. These hybrids could therefore be used as an alternative to combined treatment with the unlinked antimicrobials. The combined use of unlinked antibiotics has also shown promising results in the treatment of drug-resistant nosocomial bacterial strains, 48−52 but results in complex dual efficacy, pharmacokinetic, and toxicity profiles. Conjugated hybrid agents would avoid this in drug development.
Further studies are needed to better understand the mechanism of action of our conjugated molecules: specifically, whether they enter the cell and/or act on their intracellular target as a conjugated hybrid or as individual components. Tobramycin and Bac7 fragments were linked by a disulfide bond, so it is expected that the reducing environment in bacterial cells 53,54 would cause the release of the individual components in the cytosol. Indeed, the disulfide bonds are cellular redox switches. In addition, the Cys residue is linked to tobramycin via a succinic acid bound to the antibiotic by an ester bond, which could also be susceptible to cleavage by bacterial esterases. Therefore, the individual peptides and the modified or unmodified tobramycin could act separately upon entering the bacterial cell, and independently reach their respective target sites in the bacterial ribosome. On the other hand, if conjugate molecules do not separate, they would act as a single, multimodal antibacterial compound that simultaneously binds to and affects the ribosome, with dynamics that do not necessarily overlap with those of the individual components. 6,27 Future tests using conjugates linked by irreversible bonds and/or the addition of fluorophores to the hybrid molecules may help clarify these mechanistic aspects.
Alternatively, or in addition to this, the advantage conferred by conjugation over the individual components could be due to improved transit to the bacterial surface and/or uptake into the bacterial cells. The conjugated molecules may have gained an increased capacity to cross the external barriers and/or bacterial membrane, allowing both the peptides and tobramycin to reach and inactivate their cytosolic targets more rapidly. This may explain the results obtained with P. aeruginosa PA10 and PA21 strains, in which only the hybrid molecules inhibited bacterial growth. Our conjugates are therefore promising model systems that point to the usefulness of linking antibiotics to PrAMPs. Clearly, they are unsuitable for systemic use as such, given the reducing properties of plasma, 34  (dimethylamino)methylene]-1H-benzotriazolium 3-oxide tetrafluoroborate 2-(1H-benzotriazole-1-yl)-1,1,3,3-tetramethyluronium tetrafluoroborate (TBTU) were purchased from Iris Biotech. 2-Chlorotrityl chloride resin (∼1 mmol/g equiv) and Rink Amide resin (0.35 mmol/g equiv) were obtained from Novabiochem. Muller-Hinton growth medium was from Difco, and 96-well round-bottom microtiter plates were from Sarstedt. Bacterial reference strains were provided by the American Type Culture Collection (ATCC) and by the Deutsche Sammlung von Mikroorganismen and Zellkulturen (DSMZ). Clinically isolated strains were isolated from CF patients as reported previously. 55 Solid-phase syntheses were performed using an Initiator+ Alstra microwave peptide synthesizer (Biotage). ESI-MS analyses were performed using an Esquire 4000 instrument (Bruker Daltonics). The measured mass was compared to the average mass calculated using the Peptide Mass Calculator (PeptideWeb.com). 1 H and 13 C NMR spectra were recorded respectively at 500 and 101 MHz for 1 H and 13 C NMR, on a Jeol EX-400 instrument (400 MHz).
(Boc) 5 Tob (compound 1, see Scheme 1). (Boc) 5tobramycin was synthesized as described by Michael et al., 1999. 56 Briefly, tobramycin (0.25 g, 0.53 mmol, 1 equiv) was dissolved in a DMSO/H 2 O mixture (15 mL, 6:1) and warmed at 60°C while stirring. Di-tert-butyl dicarbonate (1.15 g, 5.5 mmol, 10 equiv) was then added to the mixture. The solution was stirred o.n. at 60°C, then cooled to RT and 5 mL of 30% aqueous ammonia added to stop the reaction. The white precipitate was collected, washed several times with water and dried to yield 0. (Boc) 5 Tob-hemisuccinate (mTob) (compound 2, see Scheme 3). (Boc) 5 -Tob (0.37 g, 0.38 mmol, 1 equiv) was dissolved in 20 mL anhydrous toluene and treated with succinic anhydride (0.058 g, 0.58 mmol, 1.5 equiv) and 4-(dimethylamino)pyridine (DMAP) (0.23 g, 1.9 mmol, 5 equiv). The solution was heated at 85°C for 20 h in a paraffin oil bath under argon flux until completion of the reaction (monitored by analytical RP-HPLC and ESI-MS). After cooling to room temperature, 20 mL of dichloromethane and 40 mL of aqueous HCl (pH 2.5) were added to separate organic and aqueous phases. The combined organic layer was washed with brine, dried over Na 2 SO 4 , and the solvent removed to yield 0. , when using the Rink Amide resin but kept at 50°C with the 2-chlorotrityl chloride resin to prevent premature detachment. Cleavage was performed with a TFA/TIS/DODT/H 2 O (82:5:8:5 v/v) mixture for 3 h. The products were precipitated with cold tert-butyl methyl ether, collected, and dried overnight under vacuum. Analysis of the crude peptides by RP-HPLC showed them to be relatively pure so they were used in the next steps without further purification. Analytical and preparative RP-HPLC were respectively carried out using Waters Symmetry C18, 100 Å, 3 16 ]-NH 2 ) were dissolved in 300 μL of 10 mM HCl, and divided into five aliquots of 60 μL. The reaction was performed in the dark and under nitrogen. Twenty-five μL of iodoacetamide (Iaa) (stock solution 2 mM Iaa in ethanol) was diluted in 125 mL of 0.5 M Tris-acetate and 2 mM EDTA at pH 8. Then 5 mL of 0.1 mM ascorbic acid (Sigma) was added at room temperature and the reaction was maintained under gentle agitation for 30 min. Another four aliquots were added every 30 min under the same conditions, and the fourth one was immediately followed by 8 mL of 10 mM Iaa dissolved in Tris, pH 8, and 2 mL of 1 mM ascorbic acid to scavenge traces of iodine. The reaction mixture was then diluted with 0.05% trifluoroacetic acid in water to a final pH of 2.5 and the peptides were directly purified by RP-HPLC. Purity was confirmed by analytical RP-HPLC and ESI-MS (