In Vitro Activity of a Novel Siderophore-Cephalosporin, GT-1 and Serine-Type β-Lactamase Inhibitor, GT-055, against Escherichia coli, Klebsiella pneumoniae and Acinetobacter spp. Panel Strains

This study investigates GT-1 (also known as LCB10-0200), a novel-siderophore cephalosporin, inhibited multidrug-resistant (MDR) Gram-negative pathogen, via a Trojan horse strategy exploiting iron-uptake systems. We investigated GT-1 activity and the role of siderophore uptake systems, and the combination of GT-1 and a non-β-lactam β-lactamase inhibitor (BLI) of diazabicyclooctane, GT-055, (also referred to as LCB18-055) against molecularly characterised resistant Escherichia coli, Klebsiella pneumoniae and Acinetobacter spp. isolates. GT-1 and GT-1/GT-055 were tested in vitro against comparators among three different characterised panel strain sets. Bacterial resistome and siderophore uptake systems were characterised to elucidate the genetic basis for GT-1 minimum inhibitory concentrations (MICs). GT-1 exhibited in vitro activity (≤2 μg/mL MICs) against many MDR isolates, including extended-spectrum β-lactamase (ESBL)- and carbapenemase-producing E. coli and K. pneumoniae and oxacillinase (OXA)-producing Acinetobacter spp. GT-1 also inhibited strains with mutated siderophore transporters and porins. Although BLI GT-055 exhibited intrinsic activity (MIC 2–8 μg/mL) against most E. coli and K. pneumoniae isolates, GT-055 enhanced the activity of GT-1 against many GT-1–resistant strains. Compared with CAZ-AVI, GT-1/GT-055 exhibited lower MICs against E. coli and K. pneumoniae isolates. GT-1 demonstrated potent in vitro activity against clinical panel strains of E. coli, K. pneumoniae and Acinetobacter spp. GT-055 enhanced the in vitro activity of GT-1 against many GT-1–resistant strains.


Introduction
The World Health Organisation has prioritised third-generation cephalosporin and carbapenem-resistant Enterobacteriaceae and Acinetobacter baumannii as critical pathogens for research and development of new antibiotic drugs [1]. More importantly, the emergence of strains co-harbouring carbapnemases, such as bla KPC-2 or bla NDM-9 , and colistin resistance gene, such as mcr-1, have made have made treatment more challenging [2,3]. This has highlighted the importance of developing new antibiotics to address the problem of antibiotic resistance. However, in contrast to the rapid and continuous escalation in the emergence of multidrug-resistant (MDR) Gram-negative bacteria and the diminishing efficacy of the current antibiotic arsenal, there has been a lag in investment in new research and drug development [1]. A bottleneck in current approaches for designing novel antibiotics is the limited number of drug targets, such as components of nucleic acid synthesis and cell wall assembly, or bacterial ribosomes [4].
Since the first antibiotics were discovered, Gram-negative bacteria have developed various resistance mechanisms, such as hydrolysis or modification of antibiotics, reducing antibiotic uptake via loss of porins, reducing the affinity of antibiotic-target binding and increasing the expression of efflux pumps [5,6]. Moreover, the outer membrane barrier is a crucial resistance determinant in Gramnegative bacteria, as reduced permeability of the cell wall potentiates other antibiotic resistance mechanisms.
The novel siderophore-cephalosporin GT-1 (previously known as LCB10-0200) is in development by LegoChem Biosciences (Daejeon, Korea) and Geom Therapeutics (San Francisco, CA, USA). This compound is a conjugate of a novel siderophore-dihydroxypyridone and a modified aminothiazolylglycyl cephalosporin and exploits bacterial iron-uptake systems to enhance entry into Gram-negative pathogens using a "Trojan-horse" strategy [7] (Figure 1a). First, this approach opens a more efficient route for the antibiotic to enter the bacterial periplasm via siderophore-uptake systems. Second, the conjugated structure resists the hydrolytic activity of many current extended-spectrum βlactamases and carbapenemases. As a result, GT-1 exhibits activity against many MDR pathogens.
One well-described strategy to enhance the activity of β-lactam antimicrobials in the presence of β-lactamases is to combine them with β-lactam inhibitors (BLIs) [8]. In the last 5 years, new generations of BLIs have been approved for clinical use, including vaborbactam (cyclic boronic acid BLI) and avibactam (diazabicyclooctane BLI) [9,10]. GT-055 (also referred to as LCB18-055), in development by LegoChem Biosciences and Geom Therapeutics, is a novel non-β-lactam BLI of the diazabicyclooctane class [11] (Figure 1b). In a previous study, Oh et al. characterised the in vitro activity of GT-1, focusing primarily on Pseudomonas aeruginosa [7]. Only 17 P. aeruginosa strains with identified β-lactamases (OXA-2, OXA-10, OXA-17, VIM and IMP) were investigated. Even though the in vitro activity of GT-1 was also determined in other species, such as Escherichia coli, Klebsiella pneumoniae and A. baumannii, the resistome profile of the tested strains has not been described. In this study, the in vitro activity of GT-1 and other antibiotics was investigated against panels of characterised isolates of E. coli, K. pneumoniae and Acinetobacter spp., which exhibit a variety of β-lactam resistance mechanisms. Additionally, this is the first study to examine the synergistic combination of GT-1 and GT-055 and compare its potency to other combinations, such as azithromycin/avibactam and ceftazidime/avibactam. Until now, there has been no systematic investigation of the correlation between GT-1 potency and the different iron uptake systems of E. coli, K. pneumoniae and Acinetobacter spp. clinical isolates. Moreover, there has been no report describing other mechanisms that can contribute to resistance to GT-1, such as GT-1 hydrolysis In a previous study, Oh et al. characterised the in vitro activity of GT-1, focusing primarily on Pseudomonas aeruginosa [7]. Only 17 P. aeruginosa strains with identified β-lactamases (OXA-2, OXA-10, OXA-17, VIM and IMP) were investigated. Even though the in vitro activity of GT-1 was also determined in other species, such as Escherichia coli, Klebsiella pneumoniae and A. baumannii, the resistome profile of the tested strains has not been described. In this study, the in vitro activity of GT-1 and other antibiotics was investigated against panels of characterised isolates of E. coli, K. pneumoniae and Acinetobacter spp., which exhibit a variety of β-lactam resistance mechanisms. Additionally, this is the first study to examine the synergistic combination of GT-1 and GT-055 and compare its potency to other combinations, such as azithromycin/avibactam and ceftazidime/avibactam. Until now, there has been no systematic investigation of the correlation between GT-1 potency and the different iron uptake systems of E. coli, K. pneumoniae and Acinetobacter spp. clinical isolates. Moreover, there has been no report describing other mechanisms that can contribute to resistance to GT-1, such as GT-1 hydrolysis or porin loss. Thus, this study was undertaken to elucidate the genomic characteristics of the panel strains and associated GT-1 minimal inhibitory concentrations (MICs).
In combination with the β-lactamase inhibitor GT-055 (at 4 µg/mL), GT-1/GT-055 MICs for all strains decreased to ≤0.12 µg/mL, similar to or lower than the MICs for the combination of CAZ-AVI (Table 1). GT-055 alone exhibited activity against the E. coli isolates with MICs ranging from 2 to 8 µg/mL. The siderophore uptake system and ompC and ompF integrity were investigated to evaluate the contribution of the siderophore transporters and porin loss to the MIC of GT-1 ( Table 2). There were no deletions or early terminations in the tonB, exbB, exbD, fiu, or fepA genes in any of the isolates. The GT-1 MIC was ≤0.5 µg/mL with the triple deletion of fecA, fhuA and iroN in strain YMC2016/06/N138 or the deletion of fecA, iutA, fyuA and iroN in strains YMC2017/07/B11634, YMC2016/01/N95, YMC2016/10/N189, as well as the deletion of iutA, fhuA and iroN in strain YMC2016/06/N255. Furthermore, the deletion of both ompC and ompF in strain YMC2016/06/N138 and deletion of ompF in strain YMC2016/10/N189 did not appear to be associated with increases in GT-1 MIC. Although strain YMC2016/02/N176 exhibited a high GT-1 MIC of 8 µg/mL, there was one observed deletion of fyuA without porin loss. In this case, the β-lactamases present were likely responsible for the high MIC, as the combination with GT-055 reduced the MIC to ≤0.12 µg/mL. Abbreviations: AZT, aztreonam; AZT-AVI: aztreonam/avibactam; CAZ, ceftazidime; CAZ-AVI, ceftazidime/avibactam; MEM, meropenem; IPM, imipenem; CL: colistin, ESBL, extended-spectrum β-lactamase; Pnase, penicillinase; NS-OXAse, narrow-spectrum oxacillinase; MLST, multilocus sequence type; MIC, minimum inhibitory concentration. Antibiotic susceptibility testing was performed using the agar dilution method. Interpretation followed the Clinical and Laboratory Standards Institute guidelines M100, 28th ed., with the exception of colistin, in which the European Committee on Antimicrobial Susceptibility Testing (EUCAST) guidelines v9.0 were applied.
GT-1 MICs ranged from 0.5 to 1 µg/mL against KPC-2-and IMP-1-producing K. pneumoniae. These MICs were ≥16-fold lower compared with carbapenems, for which the MICs ranged from 16 to 64 µg/mL. GT-1 MICs against the ESBL-producing strains SHV-12, SHV-83 and CTX-M-15 ranged from 0.25 to 4 µg/mL, 8-to 64-fold lower than the ceftazidime MICs and 2-to 8-fold lower than the carbapenem MICs. Among five isolates expressing the DHA-1 β-lactamase, four isolates exhibited high GT-1 MICs (4-64 µg/mL). Although strain YMC2016/04/N62 was resistant to colistin (MIC = 32 µg/mL), it exhibited a GT-1 MIC of 0.5 µg/mL. To directly assess the activity of GT-1 in the presence of DHA-1 β-lactamase, bla DHA-1 was cloned into ZpUC19 and transformed into E. coli DH5α, which resulted in a 16-fold increase in the GT-1 MIC compared to the control carrying the empty ZpUC19 vector (Table 3). However, in combination with the β-lactamase inhibitor GT-055, the GT-1/GT-055 MICs decreased to ≤0.12 µg/mL, except for strains YMC2010/10/P347 and YMC2012/08/C631. In YMC2010/10/P347, the GT-1 MIC decreased 4-fold to 16 µg/mL when combined with GT-055, which was similar to the MIC for GT-055 alone. No change in GT-1 MIC (0.5 µg/mL) was observed against strain YMC2012/08/C631 (IMP-1) in combination with GT-055. The siderophore uptake systems and integrity of ompK35 and ompK36 were also investigated to elucidate the role of the siderophore transporters and porin loss on GT-1 MICs against K. pneumoniae ( Table 2). No deletion or early termination of tonB or exbD was observed, but iroN was absent in all of the isolates. The deletion of exbB was observed in strain YMC2011/8/B10311, but the GT-1 MIC was low at 0.5 µg/mL. The GT-1 MIC was ≤1 µg/mL in strain YMC2011/07/B7207, in which fecA, fiu and fyuA were deleted; in strain YMC2013/7/B3993 in the absence of fepD, fiu, cirA and fyuA, in strain YMC2011/7/B774 with the deletion of fecA, fiu, fhuA and fyuA, and in strain YMC2012/8/C631 with the deletion of fecA, fiu, cirA, fhuA and fyuA. Additionally, the single deletion of ompK35 in strain YMC2013/7/B3993 and double deletion of ompK35 and ompK36 in strain YMC2010/10/R2142 did not result in high GT-1 MICs. This suggested that mutations in the siderophore transporter genes fecA, fiu, cirA, fhuA, or fyuA and the porin genes ompK35 or ompK36 do not markedly affect GT-1 MICs. Early termination of fiu in strain YMC2011/7/B36 and deletion of fecA in strain YMC2011/11/B7578 were associated with GT-1 MICs of 16 and 64 µg/mL, respectively. However, similar mutations were also observed in strains exhibiting low GT-1 MICs, so the importance, if any, of these mutations in determining GT-1 MICs in strains YMC2011/7/B36 and YMC2011/11/B7578 remains unclear. Although there was only one deletion of iroN in the siderophore uptake system in strain YMC2010/10/P347, this strain exhibited a high GT-1 MIC of 64 µg/mL. The low MIC for the combination of GT-1/GT-055 with this isolate (relative to GT-1 alone or GT-055 alone) was consistent with the high GT-1 MICs reflecting the activity of β-lactamases inhibited by GT-055.

Discussion
Resistance among Gram-negative species to existing classes of antimicrobials, including carbapenems, has limited treatment options for clinicians. In some cases, the only effective option is colistin, a drug once abandoned due to its serious side effect profile. However, the effectiveness of colistin has decreased as a result of the rapid spread of mcr-like genes [13]. The next generation of antibiotics should include new agents that would impede the development of antibiotic resistance in MDR bacteria. Siderophore-antibiotic conjugants may partially fulfil this requirement. Because ferric ion is indispensable for bacterial growth and virulence [14,15], any mutations in siderophore uptake systems during antibiotic treatment could lead to the loss of bacterial virulence, thus ultimately reducing pathogenicity in host cells. In addition, siderophore uptake may involve more than one siderophore transporter system [16]; hence, a single mutation in or deletion of one transporter may not completely prevent siderophore-mediated influx. Consequently, in most cases, multiple mutations in different bacterial siderophore uptake transporters are required to effectively limit the entry of a siderophore-antibiotic conjugate [16]. More importantly, loss of function of tonB or tonB3 (in Acinetobacter spp.), which encodes the primary proton motive force provider for active transport in Gram-negative bacteria, can completely retard bacterial growth in iron-depleted environments, such as occurs during an infection [12,17].
In this study, the novel siderophore cephalosporin, GT-1, exhibited potent activity against many E. coli, K. pneumoniae and Acinetobacter spp. MDR strains. However, high GT-1 MICs were observed for some strains. Non-susceptibility to GT-1 exhibited by some K. pneumoniae strains was often associated with the presence of AmpC β-lactamase DHA-1, for which the GT-1 MICs ranged from 4 to 64 µg/mL, with the exception of strain YMC2011/11/B1440. As we found no sequence differences over nearly 1000 bp upstream of the bla DHA-1 gene (data not shown), reduced bla DHA-1 expression seems an unlikely explanation; thus, the reason for the lower GT-1 MIC for this isolate remains to be determined. As expected, an increase in GT-1 MIC was observed in strain DH5α, which was transformed with ZpUC19::bla DHA-1 , confirming the sensitivity of GT-1 to β-lactamase DHA-1.
Another important aspect of our study was the systemic evaluation of siderophore uptake systems and porin loss in E. coli and K. pneumoniae panel strains and the resulting effect on GT-1 MICs. Previous reports demonstrated that single and double knock-out mutations of cir and fiu in E. coli decrease inhibition zones of catecholate siderophore-conjugated antibiotics [16]. However, our data suggest that GT-1 is efficiently transported into E. coli and K. pneumoniae cells with multiple defects in siderophore transporters. This could be explained by the structure of the GT-1 siderophore, which is a fusion of hydroxamate and catecholate that can be transported via hydroxamate or catecholate receptors. In addition, GT-1 did not exhibit a decrease in potency against isolates with single or double deletions of the porin genes ompC and/or ompF in E. coli or of ompK35 and/or ompK36 in K. pneumoniae. These results were in accordance with previous findings regarding siderophore-conjugated antibiotics [16] and suggest that the novel siderophore dihydroxypyridone is a suitable conjugant for the development of novel antibiotics. We observed similar findings for GT-1 with Acinetobacter spp. strains. In our analyses, GT-1 MICs were not high in double or triple bauABCDE mutants in strains YMC2012/09/R2209, YMC2012/01/R79, YMC2011/07/R812 and YMC2003/01/R306. This supports the hypothesis that GT-1 can be taken up via a transport system other than that involved in acinetobactin uptake and underscores the need for additional studies of GT-1 in this species.

Isolates
The clinical isolates examined in this study were obtained from a university-affiliated hospital in South Korea and collected from 2013 to 2017. For E. coli, approximately 2000 isolates were screened to select for specific antimicrobial-resistant phenotypes. Acinetobacter spp. and K. pneumoniae isolates were selected from our previously characterised panel strains [21,22]. Three K. pneumoniae strains (YMC2016/01/R859, YMC2016/02/N207 and YMC2016/04/N62) exhibiting carbapenem resistance were also added. Finally, 13 E. coli strains, 14 K. pneumoniae strains and 11 Acinetobacter spp. strains expressing different β-lactamases (narrow-spectrum and ESBL, KPC-and OXA-carbapenemases, metallo-and AmpC) were selected. MDR strains were defined as exhibiting resistance to at least three different antibiotic classes [23].

Antimicrobial Susceptibility Tests
MICs for bacterial strains were determined using the Mueller-Hinton agar dilution technique, according to CLSI M07-A10 10th edition and M100 28th edition guidelines (2018) [24,25]. For colistin, interpretations were based on the European Committee on Antimicrobial Susceptibility Testing guidelines version 9.0 (2019). Antibiotic agents used as comparators included aztreonam, ceftazidime, meropenem, imipenem and colistin. The β-lactamase inhibitor avibactam was tested in combination with aztreonam and ceftazidime. GT-055 was included to assess the synergistic effect with GT-1. Antibiotic concentrations ranged from 0.12 to 256 µg/mL. GT-055 was tested at 4 µg/mL, similar to the avibactam concentration recommended by the CLSI guidelines. A previous study reported no change in GT-1 activity against cells grown on iron-depleted Muller-Hinton medium [26]. Therefore, the in vitro activity of GT-1 and GT-1/GT-055 under iron-depleted conditions was not assessed in this study.

DNA Extraction and Whole-Genome Sequencing
Bacteria were cultured overnight in Luria-Bertani broth at 37 • C. One millilitre of the overnight bacterial culture was used, and genomic DNA was extracted using a Wizard genomic DNA purification kit (Promega, WI, USA) according to the manufacturer's instructions. Whole-genome library DNA was prepared using a Miseq reagent kit v3 and sequenced using Miseq v3.2 × 300-bp paired-end read cartridges (Illumina, CA, USA).

Cloning
Following PCR amplification with the primers listed in Table S1, bla DHA-1 with HindIII and EcoRI restriction site ends and bla PER-1 with BamHI and EcoRI restriction site ends were cloned into ZpUC19.
The ZpUC19 constructs were then transformed into E. coli DH5α, and GT-1 MICs were determined. Zeocin (50 µg/mL) was used for colony selection.

Analysis of Siderophore Uptake System and Porin Loss
Well-characterised siderophore transporters and porins from previous studies were included in the analysis ( Table 4). The GenBank accession numbers of reference sequences are listed in Table S2. The selected target genes were mapped against the whole genome sequences of the panel strains using Bowtie alignment [31]. The mapped DNA sequences were translated into protein sequences to identify mutation-associated alterations affecting protein coding.