Antibacterial Activity of Peptide Derivatives of Phosphinothricin against Multidrug-Resistant Klebsiella pneumoniae

The fast spread of bacteria that are resistant to many classes of antibiotics (multidrug resistant) is a global threat to human and animal health with a worrisome scenario ahead. Novel therapeutical strategies are of crucial importance to combat this phenomenon. For this purpose, we investigated the antimicrobial properties of the naturally occurring tripeptide Bialaphos and a dipeptide L-leucyl-L-phosphinoithricin, the synthesis and diastereomers separation of which are herein described. We demonstrate that these compounds are effective on clinical isolates of the human pathogen Klebsiella pneumoniae, causing hospital-acquired and community-acquired infections. The tested isolates were remarkable for their resistance to more than 20 commercial antibiotics of different classes. Based on previous literature data and our experiments consisting of glutamine supplementation, we suggest that both compounds release phosphinothricin—a well-known nanomolar inhibitor of glutamine synthetase—after their penetration in the bacterial cells; and, in this way, exert their antibacterial effect by negatively affecting nitrogen assimilation in this pathogen.


Introduction
An important achievement of medicine in the 20th century was the discovery of natural compounds exhibiting antimicrobial activity, and the development of new drugs based on them, which made it possible to significantly reduce mortality from infectious diseases. The widespread use of antibiotics in medical practice along with their overuse, however, led to the emergence of antibiotic-resistant pathogens and even multidrug-resistant (MDR) bugs, i.e., microorganisms that carry the genes or mutations for inactivating or making ineffective several types of antibiotics. This problem has become frightening. In 2019, more than 1.2 million people worldwide-and potentially many more-died as a direct result of the inefficacy of the antibiotics in current use to treat antibiotic-resistant bacterial infections [1].

Synthesis of L-Leu-L-PT
L-Leu-L-PT and L-Leu-D-PT were synthesized by the condensation of N-hydroxysuccinimide ester of N-Cbz-L-Leu (N-Cbz-L-Leu-OSu) with D,L-PT in water/1,2-dimethoxyethane mixture, followed by one-pot removal of Cbz-protecting group (Scheme 1). Unexpectedly, it was possible to separate the diastereomers by ion-exchange chromatography on sulfocationite Dowex-50WX8 (H+-form), by eluting the resin with a large volume of water. 31 P-NMR analysis of the eluted fractions is depicted in Figure 2. L-Leu-D-PT eluted first, but was contaminated with some L-Leu-L-PT (approximately 10% in fraction n. 7 (700 mL, see Figure 2; for NMR spectra- Figures S6, S7 and S8). Then, a mixture of both diastereomers was eluted, and finally L-Leu-L-PT was eluted. It should be noted here that NMR chemical shifts ( 1 H, 13 C and 31 P) of L,Dand L,L-diastereomers were different (original spectra are provided in the Supplementary information). Notably, a preliminary screening showed that only the last-eluted diastereomer (L-Leu-L-PT) possessed antibacterial activity. This is in line with the fact that only the derivatives of L-Glu are effective inhibitors of GS [16].

Synthesis of L-Leu-L-PT
L-Leu-L-PT and L-Leu-D-PT were synthesized by the condensation of N-hydroxysuccinimide ester of N-Cbz-L-Leu (N-Cbz-L-Leu-OSu) with D,L-PT in water/1,2-dimethoxyethane mixture, followed by one-pot removal of Cbz-protecting group (Scheme 1). Unexpectedly, it was possible to separate the diastereomers by ion-exchange chromatography on sulfocationite Dowex-50WX8 (H+-form), by eluting the resin with a large volume of water. 31 P-NMR analysis of the eluted fractions is depicted in Figure 2. L-Leu-D-PT eluted first, but was contaminated with some L-Leu-L-PT (approximately 10% in fraction n. 7 (700 mL, see Figure 2; for NMR spectra- Figures S6, S7 and S8). Then, a mixture of both diastereomers was eluted, and finally L-Leu-L-PT was eluted. It should be noted here that NMR chemical shifts ( 1 H, 13 C and 31 P) of L,D-and L,L-diastereomers were different (original spectra are provided in the Supplementary information). Notably, a preliminary screening showed that only the last-eluted diastereomer (L-Leu-L-PT) possessed antibacterial activity. This is in line with the fact that only the derivatives of L-Glu are effective inhibitors of GS [16]. Scheme 1. i-Cbz-L-Leu-OSu/DME/H2O/Na2CO3/NaHCO3; ii-HBr/AcOH; iii-Dowex 50WX8 (H+), elution with H2O.

Bialaphos and L-Leu-L-PT Inhibit the Growth of E. coli K12 in Minimal Medium
It was shown that aminophosphonic acids penetrate poorly into bacteria [20], and D,L-PT is not an exception. In fact, even in minimal medium E (EG), it has poor antimicrobial activity towards E. coli (MIC90 266-304 µg/mL, Table 1). In striking contrast, the dipeptide L-Leu-L-PT was approximately 6000-fold more active (MIC90 0.04 µg/mL, Table  1), while the tripeptide Bialaphos inhibited the growth of E. coli in the same medium at a

Bialaphos and L-Leu-L-PT Inhibit the Growth of E. coli K12 in Minimal Medium
It was shown that aminophosphonic acids penetrate poorly into bacteria [20], and D,L-PT is not an exception. In fact, even in minimal medium E (EG), it has poor antimicrobial activity towards E. coli (MIC 90 266-304 µg/mL, Table 1). In striking contrast, the dipeptide L-Leu-L-PT was approximately 6000-fold more active (MIC 90 0.04 µg/mL, Table 1), while the tripeptide Bialaphos inhibited the growth of E. coli in the same medium at a MIC 90 <0.001 µg/mL (Table 1). These differences in the activity of Bialaphos and L-Leu-L-PT are likely because Bialaphos actively penetrates E. coli using both oligopeptide transporter Opp and dipeptide permease Dpp [21], while the dipeptide most likely uses only the latter. It should be noted that both Bialaphos and L-Leu-L-PT were poorly effective when a rich medium (Mueller-Hinton) was used. In this case, the MIC 90 values for Bialaphos and L-Leu-L-PT were 500 and 1600 µg/mL, respectively (Table 1), which is in agreement with the known literature data disclosing poor antibacterial activity of phosphonopeptides on rich medium [20,22]. A similar relationship between the activities on minimal and rich media (Table 1) was observed in our experiments with Alaphosphin, a dipeptide containing a phosphonic moiety on the α-carbon, known to target the biosynthesis of bacteria cell wall.  As a next step, we examined the effects of Bialaphos and L-Leu-L-PT on the growth of the reference strain of K. pneumoniae ATCC 13883 using the agar-diffusion test. Both compounds inhibited the growth of K. pneumoniae in a dose-dependent manner, starting from 0.2 µg/disk, with approximately the same efficacy (Figures 3A and S1A and Table 2). It should be noted that, in the case of the inhibition of E. coli growth, Bialaphos was at least 40-fold more active than L-Leu-L-PT (Table 1). These differences might be ascribed to the differences in the transport rate and/or the activity of peptidyl permeases in E. coli and K. pneumoniae.
The cytoplasmic cleavage of Bialaphos and L-Leu-L-PT likely releases L-PT ( Figure S2), a very effective and specific inhibitor of GS [14,15]. Accordingly, the inhibitory activity of Bialaphos and L-Leu-L-PT was attenuated when the medium was supplemented with 0.5 mM L-glutamine ( Figures 3B and S1B). This suggests that the metabolic target of Bialaphos and L-Leu-L-PT in K. pneumoniae is very likely GS. Moreover, the protective effect of glutamine turned out to be dose-dependent for both peptides, i.e., Bialaphos ( Figure S3) and L-Leu-L-PT ( Figure S4).
The growth of the reference strain of K. pneumoniae ATCC 13883 is inhibited by many antibiotics, such as aminoglycosides, cephalosporins, fluoroquinolones, and tetracyclines. Amongst the antibiotics used in this work, the most effective were Ciprofloxacin, Cefotaxime, and Levofloxacin ( Figure 4-disks 4, 7, 8, respectively; and Table S1), while the activity of the other tested antibiotics was approximately the same as that of Bialaphos and L-Leu-L-PT ( Figure 4, Table S1). It is worth point out that Ampicillin (Figure 4, disk 2 and Table S1) was active only when used in combination with Sulbactam, the β-lactamase inhibitor (Figure 4, disk 12 and Table S1) and even in this case it exerted a limited effect as it can be deduced from the small inhibition halo. All the antibiotics on disks were ap-plied using the doses typically tested in antibiotics susceptibility testing, ranging from 5 to 30 µg/disk [23]. Bialaphos and L-Leu-L-PT were applied to disks at the lowest concentration in this range (5 µg/disk), while the zones of inhibition were distinct and ranged from 18 to 23 mm in diameter. When analyzed in vitro, on average, the dose tested is more closely resembling the concentration attained when these antibiotics are administrated in patients as medicines. ATCC 13,883 Bialaphos, L-Leu-L-PT and Antibiotics of Different Classes As a next step, we examined the effects of Bialaphos and L-Leu-L-PT on the of the reference strain of K. pneumoniae ATCC 13,883 using the agar-diffusion t compounds inhibited the growth of K. pneumoniae in a dose-dependent manner, from 0.2 µg/disk, with approximately the same efficacy ( Figure 3A, Figure S1A a 2). It should be noted that, in the case of the inhibition of E. coli growth, Bialapho least 40-fold more active than L-Leu-L-PT (Table 1). These differences might be to the differences in the transport rate and/or the activity of peptidyl permeases and K. pneumoniae. The cytoplasmic cleavage of Bialaphos and L-Leu-L-PT likely releases L-PT S2), a very effective and specific inhibitor of GS [14,15]. Accordingly, the inhibito ity of Bialaphos and L-Leu-L-PT was attenuated when the medium was suppl as it can be deduced from the small inhibition halo. All the antibiotics on disks were applied using the doses typically tested in antibiotics susceptibility testing, ranging from 5 to 30 µg/disk [23]. Bialaphos and L-Leu-L-PT were applied to disks at the lowest concentration in this range (5 µg/disk), while the zones of inhibition were distinct and ranged from 18 to 23 mm in diameter. When analyzed in vitro, on average, the dose tested is more closely resembling the concentration attained when these antibiotics are administrated in patients as medicines.

Growth Inhibition of Multidrug-Resistant Klebsiella pneumoniae Clinical Isolates
MDR bacteria are threatening, especially when isolated in health care settings (hospitals, nursing homes, etc.), amongst immunocompromised patients. Carbapenems-resistant strains of K. pneumoniae are classified by the WHO as the first and most dangerous group of drug-resistant bacteria [4] since carbapenems are the last resort of β-lactam antibiotics to treat MDR infections.
Here, we tested Bialaphos and L-Leu-L-PT on three MDR strains that are clinical isolates of K. pneumoniae (Table S2, strains 1161, 1158 and 1133) from the Shumakov Centre of Transplantology (Moscow, Russia). First, we showed that Bialaphos and L-Leu-L-PT inhibited the growth of all three MDR strains in a dose-dependent manner at 0.5-10 µg/disk (Table 2, Figure S5). It is worth noting that under the assay conditions used in this study L-Leu-L-PT is slightly more potent than Bialaphos in the clinical isolates (Table 2,

Growth Inhibition of Multidrug-Resistant Klebsiella pneumoniae Clinical Isolates
MDR bacteria are threatening, especially when isolated in health care settings (hospitals, nursing homes, etc.), amongst immunocompromised patients. Carbapenems-resistant strains of K. pneumoniae are classified by the WHO as the first and most dangerous group of drug-resistant bacteria [4] since carbapenems are the last resort of β-lactam antibiotics to treat MDR infections.
Here, we tested Bialaphos and L-Leu-L-PT on three MDR strains that are clinical isolates of K. pneumoniae (Table S2, strains 1161, 1158 and 1133) from the Shumakov Centre of Transplantology (Moscow, Russia). First, we showed that Bialaphos and L-Leu-L-PT inhibited the growth of all three MDR strains in a dose-dependent manner at 0.5-10 µg/disk (Table 2, Figure S5). It is worth noting that under the assay conditions used in this study L-Leu-L-PT is slightly more potent than Bialaphos in the clinical isolates (Table 2, Figure S5). At present, we cannot provide an explanation for this, but cannot exclude that this difference may be due to mutations that affect the transport rate of dipeptide and oligopeptide transporters in the MDR strains respect to the reference strain ATCC 13883.
Finally, we compared the inhibitory activity of Bialaphos and L-Leu-L-PT against the MDR strains 1161, 1158 and 1133 of K. pneumoniae with that of twelve commercially available antibiotics belonging to different classes, i.e., aminoglycosides, β-lactams, cephalosporins, fluoroquinolones, tetracyclines, etc. In these experiments, 5.0 µg/disk of Bialaphos and L-Leu-L-PT were used. Notably, this amount is much lower than the amount needed to test most of the other antibiotics. In agreement with the previous results (Table 2, Figure S5), Bialaphos and L-Leu-L-PT were quite potent on all the three MDR strains ( Figure 5, Table S1) whereas, among the tested antibiotics, only Polymyxin B inhibited the growth in all these strains. The growth of strain 1161 was also inhibited by Tetracycline, while that of strain 1133 was inhibited by Gentamicin ( Figure 5, Table S1). Although the molecular mechanisms responsible for the multidrug resistance in these three strains of K. pneumoniae has not yet been analyzed in detail, Bialaphos and L-Leu-L-PT turned out to be effective inhibitors of the growth of all these MDR strains. According to the current knowledge [16][17][18], based on studies carried out in other microorganisms using Bialaphos, the inhibition of the growth is a consequence of GS targeting, thus eventually affecting Gln metabolism and nitrogen assimilation [24]. while that of strain 1133 was inhibited by Gentamicin ( Figure 5, Table S1). Although th molecular mechanisms responsible for the multidrug resistance in these three strains o K. pneumoniae has not yet been analyzed in detail, Bialaphos and L-Leu-L-PT turned ou to be effective inhibitors of the growth of all these MDR strains. According to the curren knowledge [16][17][18], based on studies carried out in other microorganisms using Bialapho the inhibition of the growth is a consequence of GS targeting, thus eventually affectin Gln metabolism and nitrogen assimilation [24].

Discussion
After the discovery of the penicillin by Fleming, microorganisms attracted researchers' interest as the most important source of biologically active compounds of different classes. Many of these substances are used as antibiotics per se, acting on the biosynthesis of the bacterial cell wall, protein biosynthesis, bacterial DNA replication, etc.; or serve as lead compounds for the targeted synthesis of new compounds with stronger antibacterial activity at lower doses.
Microorganisms also produce a wide range of substances with an unusual phospho-

Discussion
After the discovery of the penicillin by Fleming, microorganisms attracted researchers' interest as the most important source of biologically active compounds of different classes. Many of these substances are used as antibiotics per se, acting on the biosynthesis of the bacterial cell wall, protein biosynthesis, bacterial DNA replication, etc.; or serve as lead compounds for the targeted synthesis of new compounds with stronger antibacterial activity at lower doses.

Discussion
After the discovery of the penicillin by Fleming, microorganisms attracted researchers' interest as the most important source of biologically active compounds of different classes. Many of these substances are used as antibiotics per se, acting on the biosynthesis of the bacterial cell wall, protein biosynthesis, bacterial DNA replication, etc.; or serve as lead compounds for the targeted synthesis of new compounds with stronger antibacterial activity at lower doses.
Microorganisms also produce a wide range of substances with an unusual phosphorus-carbon bond (P-C bond), which is biochemically highly stable. Antibacterial compounds with P-C bond have found practical application in medicine and agriculture, since the phosphorus-containing group can mimic a phosphate monoester or a tetrahedral intermediate (or reaction transition state) of the carboxyl group transformations [25,26]. Notable examples (Figure 6) include the antibiotic Fosfomycin (an irreversible inhibitor of muramyl ligase A [27], the first enzyme of peptidoglycan synthesis), which is used for the treatment of cystitis; the antimalarial Fosmidomycin (an inhibitor of 1-deoxy-d-xylulose-5-phosphate reductoisomerase [28,29], an essential enzyme of the non-mevalonate pathway of isoprenoid biosynthesis), which is also active against different enterobacteria, but not against Gram-positive microorganisms or anaerobes; the already mentioned Alaphosphin (Section 2.2); and the commercial herbicide Bialaphos (Figure 1), a naturally occurring tripeptide, which, upon cleavage in the cell, gives rise to L-PT ( Figure 1)-a very efficient inhibitor of glutamine synthetase [14,15]. Microorganism genome mining is the modern approach to discover phosphorus-containing antibacterials. It is based on the exploitation of the genomic information to gain  Microorganism genome mining is the modern approach to discover phosphoruscontaining antibacterials. It is based on the exploitation of the genomic information to gain knowledge on the biosynthesis of compounds with P-C bond. Most of the known phosphonates are derived from phosphoenolpyruvate by isomerization to phosphonopyruvate (universal building block used for the biosynthesis of compounds with P-C-bond), which is catalyzed by the enzyme phosphoenolpyruvate mutase, followed by subsequent enzymatic decarboxylation, catalyzed by phosphonopyruvate decarboxylase, which gives rise to phosphonoacetaldehyde [25,26,30]. Phosphonoacetaldehyde can then undergo a large spectrum of transformations serving as another universal biochemical building block. The analysis in genome databases of the genes coding for these two enzymes, as well as of their homologues, enabled to conclude that up to 10-15% of bacterial species can produce phosphonates. Therefore, the genome mining of 10,000 actinomycetes led to rediscovering of many old phosphonates, as well as 19 new compounds, including those with antibacterial activity [31].
The enzymes involved in the biosynthesis of peptidoglycan (murein), an important constituent of the bacterial cell wall, are common targets of different amino acid-derived compounds with antibacterial activity [32]. Aminophosphonic acids, with a phosphoruscontaining group replacing the carboxyl-one (-COOH) are amongst these amino acid analogues and derivatives. However, aminophosphonates, as such, poorly penetrate in the cells of eukaryotes and prokaryotes; thereby to deliver these compounds some modifications of the molecule are required. This could explain why many of the naturally produced aminophosphonates are found in nature as short peptides, which are taken up by bacteria and fungi via peptidyl permeases [33]. Subsequent intracellular cleavage of these penetrated phosphonate-containing peptides by cellular peptidases releases the active compound, the aminophosphonic acid. Among the first examples of such a «Troyan-horse» (prodrug) strategy for the design of the antibacterial agent was the synthesis of Alaphosphin ( Figure 6). Following intracellular cleavage, Alaphosphin releases the phosphonic analogue of alanine, a highly effective inhibitor of alanine racemase, thus leading to the inhibition of the biosynthesis of the bacterial cell wall and, consequently, the inhibition of bacteria growth [22]. Dehydrophos ( Figure 6) may be considered as a "double prodrug" because it penetrates the bacterial cell using peptidyl permeases and, after the cleavage of the peptide bond, provides the phosphonic analogue of dehydroalanine. This analogue undergoes spontaneous rearrangement into methyl acetylphosphonate [34], which is an analogue of pyruvic acid and a very strong inhibitor of pyruvate dehydrogenase [35] and ref. within.
Herein, we used L-PT (Figure 1) as a phosphorus-containing antibacterial to inhibit the growth of the reference E. coli K12 strain MG1655, the reference strain of K. pneumoniae ATCC 13883 as well as three clinical MDR strains of K. pneumoniae available in the collection of the Shumakov Federal Research Centre of Transplantology and Artificial Organs. L-PT is known to irreversibly inhibit GS, a key intracellular enzyme required for the synthesis of glutamine from glutamate and ammonia, and in that respect, it is fundamental in microbial nitrogen assimilation. The inhibition mechanism has been elegantly elucidated by demonstrating that the C-P group mimics the phosphorylated intermediate of glutamate formed during the enzymatic reaction, and that this intermediate does not allow the completion of the enzymatic reaction [14,15]. To deliver the poorly penetrating L-PT in E. coli and K. pneumoniae, we incorporated it into a synthetic dipeptide, i.e., L-Leu-L-PT, as well as used it in the form of the commercially available herbicide Bialaphos (Figure 1), which is a naturally occurring L-PT-containing tripeptide with a well-characterized biosynthesis [36] and low toxicity to vertebrates. In fact, the acute oral LD 50 values of Bialaphos for male and female rats are 268 and 404 mg/kg, respectively; and the acute oral LD 50 value for chicken is greater than 5000 mg/kg. Bialaphos was also shown not to be mutagenic in the Ames assay [37]. Furthermore, Phosalacine, a tripeptide consisting of L-PT-L-Ala-L-Leu, was also shown to lack toxicity in mice when administered at 500 mg/kg [38].
Once taken up by bacteria, both Bialaphos and L-Leu-L-PT are proposed to be cleaved by peptidases thus giving rise to L-PT ( Figure S2). To the best of our knowledge, the dipeptide L-Leu-L-PT does not exist in nature. Therefore, herein, we show for the first time its synthesis and purification and tested its antimicrobial potential on the E. coli K12 reference strain MG1655. Furthermore, we observed that L-Leu-L-PT and Bialaphos were also effective at inhibiting the growth of the reference strain K. pneumoniae ATTC 13883 and, more importantly, the growth of MDR K. pneumoniae isolates. We observed that a few µg/disk of Bialaphos and L-Leu-L-PT inhibit the growth of three clinical isolates of K. pneumoniae (MDR strains 1161, 1158 and 1133), otherwise resistant to more than twenty commercially available antibiotics of different classes, including carbapenems Imipenem and Meropenem (Table S2). This is a finding that we regard as remarkable.

Materials
Sodium salts of Bialaphos and D,L-phosphinothricin (Glufosinate-ammonium) were obtained from Santa Cruz Biotechnology; Alaphosphin was from Fluka. Agar agar powder No 1 for bacteriology was from LobaChemie. N-(Benzyloxycarbonyl)-L-leucine Nhydroxysuccinimide ester (Z-L-Leu-OSu) was synthetized according to [39] and was freshly recrystallized from i-PrOH before use. All other reagents, salts and solvents were of highest purity and used as supplied by Aldrich and Acros.
Ion-exchange chromatography was carried out on Dowex 50WX8, H+-form, 100-200 mesh (BioRad) using water for elution. NMR spectra were recorded on a Bruker AM-300 (300.13 MHz for 1 H, 75.43 MHz for 13 C, and 121.44 MHz for 31 P) using D 2 O as a solvent with sodium 3-trimethyl-1propanesulfonate (DSS) as internal standard, or 85% H 3 PO 4 as external standard. Chemical shifts are given in parts per million (ppm), the letter "J" indicates spin-spin coupling constants which are given in Hertz (Hz).

The Microdilution Method to Determine the Antimicrobial Activity of Tested Compounds against Escherichia coli
The minimum inhibitory concentration of 90% colony-forming units (MIC 90 ) of the test strain E. coli K12 MG1655 was determined by the broth microdilution method in the minimal medium EG containing MgSO4•7H 2 O (0.2 g), citric acid•H 2 O (2.0 g), anhydrous K 2 HPO 4 (10.0 g), NaNH 4 HPO 4 •H 2 O (3.5 g), and glucose (4.0 g), milliQ water (1.0 L) final pH 7, prepared as described earlier [40]. Briefly, overnight cultures (2 mL) of E. coli K12 strain MG1655 were centrifuged at 3500 rpm for 15 min at 15 • C and the cell pellets resuspended in an equivalent volume of physiological solution (9 g/L NaCl). The OD 600 was then brought to 1.0. The resuspended cells were inoculated (1:25) in minimal medium EG (2 mL) and grown at 37 • C up to OD 600 = 0.5, then diluted (1:25) in the same minimal medium and dispensed in a 96-well microplate previously set up with the appropriate serial dilutions of the compounds to be tested (L-Leu-L-PT, Bialaphos, D,L-PT, Alaphosphin). The number of colony-forming units (CFU)/mL at time zero was between 0.5-1.0 × 10 6 /well and the final volume in each well was 200 µL. The microplate was incubated at 37 • C for 24 h in the microplate reader Varioskan Lux (Thermo Scientific) and every hour the OD 600 was automatically recorded. MIC 90 was calculated at 22 h from the time of the inoculum using the equation: % inhibition = [1 − (OD 600 treated/OD 600 untreated)] × 100.

The Agar Diffusion Method to Analyze Antimicrobial Activity of Tested Compounds against Klebsiella pneumoniae
The reference strain K. pneumoniae ATCC 13883 and MDR clinical isolates from patients of Shumakov Federal Research Centre of Transplantology and Artificial Organs (Moscow, Russia) were used. Species identification of clinical isolates of K. pneumoniae and their sensitivity to antibiotics were determined on an automatic bacteriological analyzer for the identification of microorganisms (MicroScan WalkAway-96 plus System, Beckman Coulter, USA) following the manufacturer's instructions.
Bialaphos, L-Leu-L-PT and twelve commercial antibiotics were tested as follows. Different amounts of the substances under testing were applied to paper discs, the discs were air dried and placed on the surface of an agar plate containing M9 medium with the following composition: Na 2 HPO 4 •7H 2 O (12.8 g), anhydrous K 2 HPO 4 (3.0 g), NaCl (0.5 g), NH 4 Cl (1.0 g), MgSO4•7H 2 O (0.5 g), CaCl 2 (15 mg), glucose (4.0 g), thiamine (1 mg), agar (15 g) brought to 1 L with milliQ water and to a final pH 7.2. The plates were previously seeded with a lawn of K. pneumoniae ATCC 13883 or one of the clinical isolates; the seeding density was 10 6 bacteria per cm 2 of the agar surface. The plates were incubated for 20 h at 37 • C. Already prepared disks with the other antibiotics were from Becton, Dickinson & Co., Franklin Lakes, NJ, USA. The antibiotic activity of all the compounds was tested by the agar diffusion method [41] and determined based on the presence and size of non-growth zones around the disks.

Conclusions
Phosphinothricin (PT), an inhibitor of glutamine synthetase (GS), in the form of its actively penetrating di-and tripeptides, i.e., L-Leu-L-PT and Bialaphos, is effective in inhibiting multidrug-resistant (MDR) clinical isolates of K. pneumoniae, otherwise not sensitive to twenty four commonly used antibiotics of different classes. Our data suggest that nitrogen assimilation via glutamine biosynthesis is of crucial importance for K. pneumoniae and might be considered as a target to affect the growth of MDR strains of this pathogen. Furthermore, we describe the original synthesis and purification of L-Leu-L-PT.