In Vitro Activity of a Novel Siderophore-Cephalosporin LCB10-0200 (GT-1), and LCB10-0200/Avibactam, against Carbapenem-Resistant Escherichia coli, Klebsiella pneumoniae, Acinetobacter baumannii, and Pseudomonas aeruginosa Strains at a Tertiary Hospital in Korea

The siderophore–antibiotic conjugate LCB10-0200 (a.k.a. GT-1) has been developed to combat multidrug-resistant Gram-negative bacteria. In this study, the in vitro activity of LCB10-0200 and LCB10-0200/avibactam (AVI) has been investigated against carbapenem-resistant Escherichia coli, Klebsiella pneumoniae, Acinetobacter baumannii, and Pseudomonas aeruginosa. Minimal inhibitory concentrations (MICs) of LCB10-0200, LCB10-0200/AVI, aztreonam, aztreonam/AVI, ceftazidime, ceftazidime/AVI, and meropenem were measured using the agar dilution method. Whole genome sequencing was performed using Illumina and the resistome was analyzed. LCB10-0200 displayed stronger activity than the comparator drugs in meropenem-resistant E. coli and K. pneumoniae, and the addition of AVI enhanced the LCB10-0200 activity to MIC ≤ 0.12 mg/L for 90.5% of isolates. In contrast, whereas LCB10-0200 alone showed potent activity against meropenem-resistant A. baumannii and P. aeruginosa at MIC ≤ 4 mg/L for 84.3% of isolates, the combination with AVI did not improve its activity. LCB10-0200/AVI was active against CTX-M-, SHV-, CMY-, and KPC- producing E. coli and K. pneumoniae, while LCB10-0200 alone was active against ADC-, OXA-, and VIM- producing A. baumannii and P. aeruginosa. Both LCB10-0200 and LCB10-0200/AVI displayed low activity against IMP- and NDM- producing strains. LCB10-0200 alone exhibited strong activity against selected strains. The addition of AVI significantly increased LCB10-0200 activity against carbapenem-resistant E. coli, K. pneumoniae.


Introduction
Carbapenem-resistant Escherichia coli, Klebsiella pneumoniae, Acinetobacter baumannii, and Pseudomonas aeruginosa have been recognized as being of "critical priority" to the research and development of new antibiotics according to the World Health Organization [1]. Even though colistin has been used as a last resort treatment of carbapenem-resistant bacteria, the emergence of its resistance has been reported worldwide [2][3][4]. A similar plight has been observed with ceftazidime/avibactam (CAZ-AVI), an antibiotic approved by the US Food and Drug Administration (FDA) in 2015 [5]. Recently, the emergence of NDM-, KPC-and/or MCR-1 co-producing E. coli and K. pneumoniae strains have been discovered [6][7][8][9]. These strains are resistant to both carbapenems and colistin, limiting treatment choices in clinical settings. Therefore, the need for developing new antibiotics that are active against carbapenem-resistant strains is highly critical and urgent. Recently, LegoChem Biosciences (Daejeon, Korea) and Geom Therapeutics (San Francisco, CA, USA) have developed the novel siderophore-cephalosporin LCB10-0200 (a.k.a. GT-1), which increases the antibiotics' influx into bacteria via the siderophore uptake system, and could potentially treat carbapenem-resistant bacterial infections [10].
In a previous study, our group evaluated the in vitro activity of LCB10-0200 against panels of well-characterised E. coli, K. pneumoniae, and Acinetobacter spp. strains showing diverse antibiograms [11][12][13]. Panel strains of these three species were classified into different groups, including Non-Extended Spectrum β-lactamase Acinetobacter spp. In the current study, we focused on the investigation of LCB10-0200 s activity against carbapenem-resistant E. coli, K. pneumoniae, A. baumannii, and P. aeruginosa and compared its activity with aztreonam (ATM), ceftazidime (CAZ), and meropenem (MEM). Moreover, avibactam (AVI), a second generation β-lactamase inhibitor, was also included in this study. AVI covalently binds and inhibits Ambler class A, class C, and some class D β-lactamases [14][15][16]. As a result, AVI can reverse the activity of CAZ in CAZ-resistant strains [17,18]. The combination of AVI and CAZ was approved by the US FDA as treatment for complicated intra-abdominal infections (cIAI), complicated urinary tract infections, hospital acquired bacterial pneumonia, and ventilator-associated bacterial pneumonia [19]. In addition, the combination of AVI with ATM also demonstrated good activity against Ambler class A/C and class B β-lactamase-coproducing strains [20]. This combination was studied in a phase IIa clinical trial for the treatment of cIAI [21]. Until now, the in vitro activity of a siderophore-cephalosporin and AVI combination has not been well studied. For that reason, this study also investigated the synergistic activity of AVI and LCB10-0200 in comparison with CAZ-AVI and ATM-AVI in vitro. The higher MICs of LCB10-0200 and LCB10-0200/AVI were correlated with the corresponding resistome profiles to explain the underlying resistance mechanisms.

Discussion
Antimicrobial resistance has been recognized as a global public health issue. Currently, the annual number of deaths caused by bacterial infection is approximately 700,000 in the entire world. This number is predicted to be around 10,000,000 deaths with the cost of around 100 trillion dollars by 2050 [23]. According to USA CDC, the number of new cases per year increased by approximately 29% from 2 million in 2013 [24] to 2.8 million in 2019 [25]. Furthermore, the number of deaths increased by 20% from 28,000 in 2013 to 35,000 in 2019. However, this status may be worsened due to the emergence of the coronavirus disease 2019 (COVID-19) pandemic, which is caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Indeed, the infection caused by viruses creates opportunistic chances for co-infection with bacteria. Since the beginning of COVID-19, there have been extensive studies and systematic reviews in terms of the co-infection between SARS-CoV-2 and secondary bacterial infection around the world. The ratio of bacterial superinfection in COVID-19 patients ranged from 3.2% to 15% in a UK secondary-care setting and Wuhan hospitals [26,27]. According to a cohort study conducted in a hospital in Barcelona, Spain, 4.7% of COVID-19 patients were co-infected with P. aeruginosa or E. coli with an average time from admission to bacterial infection diagnosis of 10.6 days [28]. Another systematic analysis of postmortem studies conducted by Clancy et al. identified that 8% of patients were infected with SARS-CoV-2 and bacteria and 24% of patients who died by SARS-CoV-2 were possibly co-infected with bacteria [29]. The highest ratio of secondary bacterial infection in this study was A. baumannii, followed by Staphylococcus aureus, P. aeruginosa, and K. pneumoniae [29]. Moreover, it was reported that 71% of patients admitted by SARS-CoV-2 in hospitals in China were treated by broad-spectrum antibiotics without the confirmation of secondary bacterial infection to save patient lives and to reduce the additional complications [30]. Consequently, this fact may have driven the antibiotic resistance rate in the COVID-19 hotspots. On the other hand, there are some good signs that can affect the worldwide antibiotic resistance climate. Firstly, the reduction of travelers all over the world during COVID-19 pandemic can subsequently reduce the spread of different types of antimicrobial-resistant bacteria from regions to regions. Secondly, stringent hand hygiene, self-quarantine, and social distancing in the community and health facilities can decrease not only the spread of SARS-CoV-2 but also the cross-infection of antibioticresistant bacteria [31]. However, the concerns about the higher ratios of antibiotic resistance in the COVID-19 aftermath should be considered and research & development of new antibiotics should be conducted in more efficient ways in parallel with enhanced antimicrobial stewardship programs. Even though antimicrobial resistance is one of the greatest threats in the mid-twenty-first century, financial investment in antimicrobial development has reduced in recent years due to the low rate of success and revenue as compared to its high investment cost [32]. Payne et al. indicated that approximately 3.5% of candidates from high throughput screening can reach to phase I of clinical trials. According to the European observatory on health systems and policies, the success rates for an antibiotic candidate in phase I → II, II → III, and III → IV are 33%, 59.3%, and 75.8%, respectively, and it takes around 13-21 years for a candidate to be available on market [33]. Once a new antibiotic is approved, it is used as a last resort and therefore limits the profitability. Another difficulty in the development of new antibiotics is the limitation of traditional drug discovery platforms, which usually results in quite similar drug structures or previously identified targets [34,35]. Low permeability on the bacterial membranes, especially in Gram-negative bacteria is also the cause of the failure in the early stage of antibiotic development [33]. Other factors such as variations in drug targets, drug hydrolyses, overexpression of efflux pumps, and porin losses are also the barriers in the later stages of novel antibiotic development [33]. However, some strategies have been applied to tackle these challenges. Firstly, various non-profit and government-based programs such as European Gram-negative antibacterial engine (ENABLE), combating bacterial resistance in Europe (COMBACTE), US Biomedical Advanced Research and Development Authority (BARDA), Global Antibiotic Research and Development Partnership (GARDP), and Combating Antibiotic-Resistant Bacteria Biopharmaceutical Accelerator (CARB-X) have been implemented to foster novel antibiotic development [36][37][38][39][40]. Secondly, novel approaches including inhaled delivery and liposomal delivery have been developed to increase antibiotic concentration in lung infection and to overcome the low drug permeability [41,42]. One way to improve the drug influx into the bacterial membrane is the conjugation between antibiotic and siderophore, a.k.a "Trojan horse" strategy, which was applied in the development of LCB10-0200 [11]. Importantly, recent advances in bioinformatics, machine learning, and deep learning have been applied in prediction of antimicrobial molecules [43,44]. Recently, Stokes et al. applied different neural network algorithms including Chemprop and ensembling to learn and predict antimicrobial properties from theirs chemical structures. A set of 2335 molecules from a FDA-approved drug library and a modest natural product library were used as a training set for growth inhibition against E. coli BW25113. The trained model was then applied to predict antimicrobial molecules from a set of 6111 molecules from the Drug Repurposing Hub and identified a broad-spectrum antimicrobial molecule, Hacilin [44]. Importantly, the structure of Haicilin is structurally divergent from current antibiotics [44]. This approach can overcome one of the shortcomings of conventional drug screening, in which the candidate structures are quite similar to known antibiotics. Also, this approach can reduce time and cost for drug library screening and development.
To cope with the predicted and potential scenarios of antibiotic resistance, our group explored the in vitro activity of LCB10-0200 alone and in combination with AVI against multiple carbapenem-resistant Gram-negative clinical isolates with various carbapenem resistance determinants. The predominant resistance mechanisms observed in this study belonged to KPC-and OXA-producing strains. Against KPC-producing strains, LCB10-0200 had a high activity with the MIC range of ≤0.12-16 mg/L, and a MIC 50 of 1 mg/L. The MIC 90 in KPC-producing strains was also 8 mg/L, being similar with the previous report [11]. LCB10-0200 activity was significantly enhanced in combination with AVI, (i.e., LCB10-0200/AVI MIC 90 was at least 16-fold lower than CAZ-AVI MIC 90 ). Of note, CAZ-AVI resistant K. pneumoniae strains have increased prevalence in many parts of the world in recent years due to the spread of a mutation in the omega loop of KPC-2 and KPC-3 [45][46][47][48]. This has prompted an urgent need to develop new antimicrobial agents against these resistant strains. Even though, there was no CAZ-AVI resistant strain detected in this study, the potent activity of LCB10-0200/AVI against KPC-producing strains has shown promising results, and further studies need to be carried out to determine the activity of LCB10-0200/AVI against CAZ-AVI resistant KPC-producing K. pneumoniae.
In addition, carbapenem-resistant E. coli, K. pneumoniae, and P. aeruginosa strains were selected. Of interest, LCB10-0200 was active against GES-4, or VIM-2 producing strains, but inactive against NDM-1, NDM-5, and NDM-9 producing strains. Addition of AVI did not enhance the activity of LCB10-0200. This was consistent with the fact that AVI has limited, or no activity against metallo-β-lactamase-producing strains [53]. The LCB10-0200 MIC against IMP-1 producing P. aeruginosa strain (YMC2017/08/U4581) was 32mg/L, which was 64-fold higher than the LCB10-0200 MIC of IMP-1 producing K. pneumoniae YMC2012/08/C631 in the previous study (0.5 mg/L) [11]. The discrepancy may be due to the reduced background of β-lactamases in the K. pneumoniae YMC2012/08/C631. Studies using more IMP-producing strains should be performed to get better insights.

Specimen Collection and Antibiotics
A total of 93 non-duplicate clinical isolates, including 15 E. coli, 27 K. pneumoniae, 25 A. baumannii, and 26 P. aeruginosa strains were collected during 2015-2018 in a Universityaffiliated hospital in Korea. Species identification was confirmed using matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS) (ASTA, Suwon, Korea) according to the manufacturer instructions. In brief, the single bacterial colony was smeared on the target plate, followed by 1-2 µL of 70% formic acid (Sigma, St. Louis, MO, USA). After 3-5 min for air-drying, 1-2 µL of matrix solution (α-cyano-4-hydroxycinnamic acid was overlaid on the same spot followed by an additional air-dry step. Finally, the peptide profile was obtained using ASTA MicroIDSys with the coreDB v1.26 and the mass spectra ranging from 2000 to 20,000 daltons. E. coli protein (YbdYbiotech, Seoul, Korea) was used as calibrator. Antibiotics used in this study include ATM (Dong-A Biotech Co., Seoul, Korea), CAZ (CJ Health Care, Seoul, Korea), and MEM (Yuhan Co., Seoul, Korea). AVI was kindly provided by LegoChem Biosciences. LCB10-0200 was manufactured by LegoChem Biosciences.

Susceptibility Tests and MIC Determinations
Minimum inhibitory concentrations (MICs) were determined using the agar dilution method and interpreted according to the CLSI guidelines [22,54]. Antibiotic concentrations used ranged from 0.12 mg/L to 256 mg/L. MIC interpretation for LCB10-0200, LCB10-0200/AVI, and ATM-AVI MIC was not available at the time of this study. A previous study reported that LCB10-0200 MICs against bacteria grown on Muller Hinton medium did not vary significantly as compared to iron-depleted Muller Hinton medium [55]. Therefore, the in vitro activity of LCB10-0200 in iron-depleted medium was not investigated in this study.

DNA Extraction and Whole Genome Sequencing
Bacterial genomic DNA (gDNA) extraction was performed using the Wizard genomic DNA purification kit (Promega, WI, USA). The quantity and quality of gDNA was measured using a NanoDrop spectrophotometer (ND-2000 Thermo Fisher Scientific, Waltham, MA, USA) and agarose gel-electrophoresis. Whole genome sequencing was performed at different centers (Supplementary Table S1). E. coli strains and some K. pneumoniae strains were sequenced at Korea Research Institute of Bioscience & Biotechnology (KRIBB, Daejeon, Korea) and Life's Art of Science (LAS, Gimpo, Korea). The libraries were prepared using TruSeq Nano DNA Library Preparation Kit and sequencing was performed on Illumina MiSeq platform (Illumina, CA, USA) using MiSeq reagent Kit v3 (600 cycles-2 × 300). Sequencing of select K. pneumoniae, A. baumannii, and P. aeruginosa isolates was performed at the J. Craig Venter Institute (JCVI, CA, USA) where the libraries were prepared using the Nextra XT library kit and sequencing was performed on the Illumina NextSeq 500 instrument using the NextSeq 500 High Output Kit (300 cycles-2 × 150).
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/ph14040370/s1, Table S1: GenBank Accession numbers and the sequencing sites for all the tested strains.