Discovery of thiosemicarbazone derivatives as effective New Delhi metallo-β-lactamase-1 (NDM-1) inhibitors against NDM-1 producing clinical isolates.

New Delhi metallo-β-lactamase-1 (NDM-1) is capable of hydrolyzing nearly all β-lactam antibiotics, posing an emerging threat to public health. There are currently less effective treatment options for treating NDM-1 positive "superbug", and no promising NDM-1 inhibitors were used in clinical practice. In this study, structure-activity relationship based on thiosemicarbazone derivatives was systematically characterized and their potential activities combined with meropenem (MEM) were evaluated. Compounds 19bg and 19bh exhibited excellent activity against 10 NDM-positive isolate clinical isolates in reversing MEM resistance. Further studies demonstrated compounds 19bg and 19bh were uncompetitive NDM-1 inhibitors with Ki = 0.63 and 0.44 μmol/L, respectively. Molecular docking speculated that compounds 19bg and 19bh were most likely to bind in the allosteric pocket which would affect the catalytic effect of NDM-1 on the substrate meropenem. Toxicity evaluation experiment showed that no hemolysis activities even at concentrations of 1000 mg/mL against red blood cells. In vivo experimental results showed combination of MEM and compound 19bh was markedly effective in treating infections caused by NDM-1 positive strain and prolonging the survival time of sepsis mice. Our finding showed that compound 19bh might be a promising lead in developing new inhibitor to treat NDM-1 producing superbug.

Introduction

The discovery and development of antibiotics was considered one of the greatest breakthroughs in the 20th century. However, the emergence of multi-drug and even extensively drug-resistant pathogens due to the abuse or overuse of antibiotics in both clinical practice and farm animal production, posed an emerging threat to the treatment of bacterial infection. As a member of the main antibiotics, β-lactams agents were widely used in the clinic. Unfortunately, the clinical value of β-lactams has been challenged by the emergence and dissemination of β-lactamases which were able to destroy β-lactams by hydrolysis of the β-lactam ring. The β-lactamases can be grouped into four classes according to the Ambler classification scheme, the serine residue dependent class A, C, D, and the metallo-β-lactamases class B (MBLs)3, 4, 5, 6. MBLs are zinc-dependent and can be classified into three subunits, B1, B2 and B3, according to the number of zinc atoms in the active site,. New Delhi metallo-β-lactamase-1 (NDM-1) which belongs to the subclass B1, was first discovered in 2008 and was demonstrated with the capability of hydrolyzing nearly all β-lactam antibiotics except monobactams. NDM-1 has experienced the fastest and widest geographical spread among MBLs, and bacterial infections harbouring plasmid-encoded NDM-1 have rapidly emerged worldwide,, thus are the most troublesome. Therefore, there is a strong demand to develop effective inhibitors of NDM-1 to tackle this bacterial resistance.

The majority of reported inhibitors of NDM-1 were non-covalent inhibitors, including chelating agents and zinc coordinating agents, such as compounds 1–413, 14, 15, 16 (Fig. 1). Compounds 1 (AMA) and 2 could chelate zinc ions of protein to inhibit the activity of NDM-1. Compound 1 was an inhibitor of NDM-1 and other MBLs as well, whereas it was completely ineffective against the metallo-β-lactamases class A, C and D. Compounds 3 and 4 (methisazone) were zinc coordinating inhibitors, molecules able to coordinate the zinc ions within the NDM-1 binding site,. Although some effective molecules have been reported, concurrently, there are neither clinical drugs which can reverse the resistance mediated by NDM-1 nor approved NDM-1 inhibitors which are in clinical trials. Additionally, the variability of the loop arrangement at the active site as well as the rapidly evolved nature of NDM-1 variants made the development of efficient inhibitors more problematic17, 18, 19, 20.

Figure 1

The structure of reported inhibitors of NDM-1 and drugs containing thiosemicarbazone.

Thiosemicarbazones, known as an important metal ion chelating agents, have attracted much attention from researchers because of their extensively antibacterial and antitumor activities21, 22, 23, 24, e.g., 3-aminopyridine carboxaldehyde thiosemicarbazone (3-AP) and di-2-pyridylketone-4-cyclohexyl-4-methyl-3-thiosemicarbazone (DpC). 3-AP (Fig. 1), which is in clinical phase II studies, is a potent ribonucleotide reductase inhibitor for the treatment of non-small-cell lung cancer and renal carcinoma25, 26, 27, 28, 29. DpC has been in clinical phase I trials and performs good therapeutic effect for head and neck squamous cell carcinoma,.

Recently, studies have found that these analogs have specific antibacterial activity, such as compound 4 (methisazone), an old drug (Fig. 1), containing thiosemicarbazone and isatin structural skeleton, which had weak inhibitory activity against NDM-1. However, the structure–activity relationship (SAR) studies as well as the mechanism of action were rarely explored, especially in antibiotics. Previously, we have reported that some compounds with thiosemicarbazone skeletons performed a good antitumor activity,. In the present study, SAR based on thiosemicarbazones against NDM-1-producing Enterobacteriaceae was studied systematically (Fig. 2). Noteworthy, the thiosemicarbazone derivatives 19bg and 19bh were potent NDM-1 inhibitors that could restore the susceptibility of the meropenem (MEM) against clinical isolates. And the combination of compound 19bh and MEM was markedly effective in treating infections caused by NDM-1 positive strain and prolonging the survival time of sepsis mice. Herein, we reported the results of our studies that ultimately led to the discovery of thiosemicarbazone derivatives as possible leads.

Figure 2

The strategy of structure optimization based on methisazone and bacteriostasis activity evaluation.

Results and discussion

Chemistry

The general route for the synthesis of the target thiosemicarbazone derivatives referred to the Schiff base mechanism (Scheme 1). As shown, the hydrazine carbothioamide intermediates were prepared by two routes, with phenylamine and aliphatic amine as the starting reagents respectively. Firstly, commercially available substituted anilines 7a‒g, Et3N and 13 were mixed in tetrahydrofuran (THF) at room temperature (r.t.), following by adding di-tert-butyl dicarbonate ((Boc)2O) and 4-dimethylaminopyridine (DMAP) for a one-pot reaction to provide 8a‒g with 45%–70% yield. Then, the mixture of hydrazine hydrate and 8a‒g was refluxed in methanol, compounds 9a‒g were afforded in 65%–85% yield. On the other hand, compounds 10a‒d were mixed with sodium hydroxide in water, then added 13 and 14 in order, the hydrazine hydrate was added finally and the mixture refluxed and crystallized to get corresponding intermediates 15a‒d with 40%–55% yield. Likewise, compounds 16 and 17 were derived from reagents 11 and 12 respectively. Finally, appropriate aldehyde or ketone 18 was mixed with the prepared N-substituted hydrazine carbothioamide intermediates in methanol and refluxed after addition of a catalytic amount of acetic acid (AcOH), the precipitation was filtered, and purified by recrystallization or column chromatography to provide target compound 19 with yield ranging from 50% to 85%. Noteworthy, individual products needed to undergo de-protection process to remove the tert-butoxy carbonyl (Boc) group with concentrated hydrochloric acid (conc. HCl).

Scheme 1

Synthesis of the thiosemicarbazone derivatives. Reagents and conditions: (a) CS2, Et3N, (Boc)2O, DMAP, THF, r.t.; (b) hydrazine hydrate (80%, w/w), methanol, reflux; (c) sodium hydroxide, hydrazine hydrate (80%, w/w), water, reflux; (d) AcOH, methanol, reflux; (e) conc. HCl, dioxane, r.t.

The chemical structure characterization of all target compounds was confirmed by 1H NMR, 13C NMR and high-resolution mass spectra (HR-MS). Besides, these target thiosemicarbazone derivatives could possess either E or Z isomeric form because of the existence of azomethine (‒CR=N‒) moiety, which can be explained by 2D NOESY NMR. In 1H NMR spectra, the typical singlet signals were observed with chemical shifts ranging from δ 9.48 to 14.33 ppm (1H, hydrazine-NH) and δ 8.08 to 9.40 ppm (1H, –CHN‒); in the 13C NMR spectra, the chemical shift of CN moiety appeared to range from δ 138.45 to 178.02 ppm and CS moiety appeared at the region of δ 162.79–188.17 ppm. These data were all consistent with the previous reports involving thiosemicarbazone,. Additionally, in the 1H NMR spectrum analysis, the –CHN‒ signal was observed as a single peak for the singlet azomethine hydrogen, indicating that only one isomer was generated. To confirm the stereochemistry more accurately, compound 19ba was chosen to elucidate this isomerism by 2D NOESY NMR. In the NOESY spectra of compound 19ba, the spatial correlation between azomethine (‒CHN‒) at δ 8.41 ppm and the hydrogen of hydrazine at δ 12.06 ppm was observed, indicating that only E isomer existed because of the appropriate intramolecular H–H distance (Supporting Information Figs. S1 and S2). Collectively, these results could prove the successful synthesis of the target compounds and indicate that the target compounds were unambiguously confirmed as E isomer.

Antibacterial activity and SAR studies

Agar disk diffusion assay was conducted initially to evaluate the inhibitory activity of all synthesized compounds combined with MEM against the NDM-1 positive Escherichia coli strain, and ethylene diamine tetraacetic acid (EDTA) was used as the positive control,. The difference of bacteriostatic zone between MEM combined using with EDTA or candidate compounds and MEM alone was shown in Tables 1‒4.

Table 1

Structures of compounds and size of the bacteriostatic zone (SAR part 1).

Compd.	R1	R2	R3	ΔMEM+EDTA (mm)	ΔMEM+Compd.(mm)
19a	Image 4	H	Image 5	8.91	1.36
19b	Image 6	H	Image 7	8.92	0.16
19c	Image 8	H	Image 9	8.58	0.12
19d	Image 10	H	Image 11	8.58	0.12
19e	Image 12	H	Image 13	9.24	2.14
19f	Image 14	H	Image 15	9.24	0.04
19g	Image 16	H	Image 17	10.7	0.66
19h	Image 18	H	Image 19	9.88	‒0.02
19i	Image 20	H	Image 21	9.88	0.64
19j	Image 22	H	Image 23	8.24	0.78
19k	Image 24	H	Image 25	11.02	0.74
19l	Image 26	H	Image 27	10.70	0.92
19m	Image 28	H	Image 29	12.32	3.82
19n	Image 30	H	Image 31	11.78	0.68
19o	Image 32	H	Image 33	11.54	1.36
19p	Image 34	H	Image 35	8.32	‒0.30
19q	Image 36	‒CH3	Image 37	8.38	1.14
19r	Image 38	‒CH3	Image 39	8.38	3.93
19s	Image 40	‒CH3	Image 41	11.54	2.74
19t	Image 42	‒CH3	Image 43	8.24	0.22
19u	Image 44	‒C2H5	Image 45	10.52	2.2
19v	Image 46	‒C3H7	Image 47	11.16	0.79
19w	Image 48	‒CH(CH3)2	Image 49	11.23	0.13
19x	Image 50	Image 51	Image 52	10.96	0.26
19y	Image 53	Image 54	Image 55	9.60	0.31

ΔMEM+EDTA means the difference of bacteriostatic zone between MEM combined using with EDTA and MEM alone; ΔMEM+Compd. says the difference of bacteriostatic zone between MEM combined using with compound and MEM alone.

Firstly, we attempted to replace the isatin part to examine the contribution of thiosemicarbazone pharmacophore to the bacteriostatic activity, the SAR studies of thiosemicarbazone derivatives were performed on the R2 and R3 position, leading to the formation of compounds 19a‒y (Table 1). Compounds 19a‒i were designed to investigate the SAR of R3 position, where different aromatic heterocyclic rings were screened, such as pyridine, furan, thiophene and indole ring. Clearly, most compounds had no noticeable antimicrobial effect (Δzoom<4.0 mm). Only compound 19e containing 2-pyridyl exhibited very weak bacteriostatic activity, the difference of bacteriostatic zone between compound 19e combined MEM and MEM alone was 2.14 mm. Besides, it was observed that when the substituted site of pyridyl changed from 2- to 3- or 4-position, the antimicrobial effect decreased significantly (19e vs. 19f and 19g). In-depth structural optimization focusing on replacement of R2 groups affiliated the thiosemicarbazone core was performed. As shown in Table 1, the effect of methyl substitution at R2 position on the activity seemed to be uncertain (19m > 19t, but 19j < 19s). To further study this, different alkyl modifications were led in, and the results showed that the activity decreased with the length of alkyl chain increasing (19r vs. 19u and 19v). Also, the phenyl substitution resulted in a significant loss of potency (19x, Δzoom = 0.26 mm). Meanwhile, when R2 was pyridyl, the activity was about 10-fold weaker than corresponding no-substituted (19m vs. 19y). Thus, these results implied that appropriate sterically hindered effect of substitution at R2 location was critical for the antibacterial activity. And so far, changing substituents on R2 position appeared to be in vain, so we did not attempt other groups except methyl any more.

Based on the first-round of optimization results, the second-round structural modification primarily focused on the variations of R1. As shown in Table 1, we have tried to introduce different amendment at R1-phenyl to develop the activity, but unfortunately, it was little success. Inspired by wondering whether the change to the original scaffold of R1 would improve the activity, we initially tried to explore the effect of the carbon chain length between R1-phenyl and thiocarbamide N-terminal to the activity. As shown in Table 2, when the spacer extended to either one- or two-carbon, the activity all reduced obviously (19e vs. 19z and 19aa). Subsequently, we attempted to replace the phenyl by an aliphatic ring or an alkyl group. Among which, compound 19ag bearing morpholinyl exhibited good antimicrobial effect with an about 1.61-fold increase in the size of the bacteriostatic zone, comparable to compound 19r (Δzoom = 3.93 mm). Similarly, compound 19ai with piperidinyl substituted also displayed favorable activity (Δzoom = 5.75 mm), but was slightly weaker than compound 19ag. In this view, following structure optimization was based on R1 being morpholine group.

Table 2

Structures of compounds and size of the bacteriostatic zone (SAR part 2).

Compd.	R1	R2	R3	ΔMEM+EDTA (mm)	ΔMEM+Compd. (mm)
19z	Image 57	H	Image 58	9.28	0.02
19aa	Image 59	H	Image 60	9.52	0.06
19ab	Image 61	H	Image 62	9.62	0.24
19ac	Image 63	‒CH3	Image 64	10.75	1.56
19ad	Image 65	‒CH3	Image 66	8.22	0.88
19ae	Image 67	H	Image 68	9.23	1.38
19af	Image 69	‒CH3	Image 70	9.22	0.06
19ag	Image 71	‒CH3	Image 72	10.68	6.32
19ah	Image 73	H	Image 74	10.60	1.34
19ai	Image 75	H	Image 76	13.12	5.75
19aj	Image 77	‒CH3	Image 78	12.05	0.62
19ak	Image 79	‒CH3	Image 80	12.52	3.62

ΔMEM+EDTA means the difference of bacteriostatic zone between MEM combined using with EDTA and MEM alone; ΔMEM+Compd. says the difference of bacteriostatic zone between MEM combined using with compound and MEM alone.

As shown in Table 3, several compounds have shown excellent activity, suggesting that morpholinyl was indeed much tolerated than original phenyl. This might be due to the enhancement of the water solubility. It was noticed that 2-pyridyl at R3 position was still beneficial for the activity, such as 19ao and 19as, which showed high potency. Interestingly, replacement of the 2-pyridyl with 2-hydroxyphenyl resulted in a robust increase of antimicrobial activity (19an vs. 19ao) comparing with the control EDTA. Moreover, when the substituted site of phenyl changed from 2- to 3- or 4-position, namely 19ay and 19az, the activity decreased. This phenomenon was similar to 2-pyridine substitution which was discussed above. While, the amino group, bearing similar polarity to the hydroxyl group, did not contribute to the activity (19al, Δzoom = 1.96 mm). Besides, when the hydroxyl group was at the same substituted site, quinoline group, namely compound 19ax, displayed very weak efficacy.

Table 3

Structures of compounds and size of the bacteriostatic zone (SAR part 3).

Compd.	R1	R2	R3	ΔMEM+EDTA (mm)	ΔMEM+ Compd. (mm)
19al	Image 82	‒CH3	Image 83	11.22	1.96
19am	Image 84	‒CH3	Image 85	8.58	3.02
19an	Image 86	H	Image 87	11.22	11.34
19ao	Image 88	H	Image 89	8.96	5.62
19ap	Image 90	H	Image 91	9.56	0.62
19aq	Image 92	H	Image 93	9.56	1.58
19ar	Image 94	H	Image 95	11.87	8.42
19as	Image 96	H	Image 97	13.12	6.43
19at	Image 98	H	Image 99	10.32	0.72
19au	Image 100	H	Image 101	10.32	1.10
19av	Image 102	‒CH3	Image 103	8.58	0.10
19aw	Image 104	H	Image 105	9.76	4.34
19ax	Image 106	H	Image 107	12.02	0.74
19ay	Image 108	‒CH3	Image 109	8.72	0.64
19az	Image 110	‒CH3	Image 111	8.72	3.26

ΔMEM+EDTA means the difference of bacteriostatic zone between MEM combined using with EDTA and MEM alone; ΔMEM+Compd. says the difference of bacteriostatic zone between MEM combined using with compound and MEM alone.

As discussed above, the water solubility of these thiosemicarbazone derivatives was vital for the bacteriostatic activity. Inspired by this, further optimization focused on introducing hydrophilic groups to the scaffold, e.g., piperazinyl or N-methyl piperazinyl. Also, we tried to use alkyl group (methyl and ethyl) to modify the hydroxyl group of R3 position to improve the water solubility, the results were shown in Table 4. In comparison to morpholinyl, piperazinyl was likely a better choice, owing to the robust improvement of the activity. All the compounds showed more substantial effect than EDTA. In addition, we did observe that the modification of the hydroxyl group of R3 position led to a slight loss of activity (19ba vs. 19bb and 19bc, 19bd vs. 19be and 19bf).

Table 4

Structures of compounds and size of the bacteriostatic zone (SAR part 4).

Compd.	R1	R2	R3	ΔMEM+EDTA (mm)	ΔMEM+ Compd. (mm)
19ba	Image 113	H	Image 114	10.62	15.16
19bb	Image 115	H	Image 116	10.42	13.76
19bc	Image 117	H	Image 118	12.14	14.12
19bd	Image 119	H	Image 120	8.22	15.50
19be	Image 121	H	Image 122	8.54	12.62
19bf	Image 123	H	Image 124	8.54	10.42
19bg	Hydrochloride of compd. 19ba			11.14	15.20a
19bh	Hydrochloride of compd. 19bd			10.44	12.74a

The average value of three separate experiments; ΔMEM+EDTA means the difference of bacteriostatic zone between MEM combined using with EDTA and MEM alone; ΔMEM+Compd. says the difference of bacteriostatic zone between MEM combined using with compound and MEM alone.

In a word, a total of 7 compounds exhibited better activities than the positive control EDTA to increase the susceptibility of NDM-1 producing isolate to MEM. Moreover, compounds 19ba and 19bd showed the most potent effect. To further enhance the hydrophilicity, we made those two compounds into hydrochloride form to yield compounds 19bg and 19bh, which performed similar antibacterial effect (results of agar disk diffusion assay was shown in Supporting Information Fig. S3). Based on these observations, we supposed that these compounds containing thiosemicarbazone might serve as NDM-1 inhibitors by chelating zinc ion like EDTA, and compounds 19bg and 19bh were selected for further investigation.

Zinc homeostasis test

It has been documented that chelation of zinc ions, which played a vital role in maintaining the hydrolytic activity of NDM enzymes, by EDTA led to the inactivation of NDM-1,. Thus, zinc homeostasis testing was employed to verify whether compounds 19bg and 19bh could exert an inhibitory effect on NDM-1 enzyme by chelation of Zn2+. Bacterial growth rates of NDM-1 positive Klebsiella pneumoniae in different treating groups including monotherapy treatment groups (MEM, EDTA, compound 19bg and compound 19bh), MEM-inhibitors combination groups (MEM + EDTA, MEM+19bg and MEM+19bh) and untreated group (control) were shown in Fig. 3. Of note, combination use of MEM and one of the inhibitors EDTA, 19bg or 19bh led to a sharp decrease of the bacterial growth rates comparing with the control group, indicating compounds 19bg and 19bh exhibited good inhibition on NDM-1 enzyme. While, adding additional zinc ions to these combinations caused the bacterial growth rates raised again, comparable with the results of monotherapy treatment groups, suggesting that the presence of extra zinc ions could influence the activity of these inhibitors and maintain the hydrolytic activity of NDM-1 against MEM.

Figure 3

Metal suppression experiments. Bacterial growth rates of NDM-1-positive K. pneumoniae with monotherapy treatment groups (MEM = 2 μg/mL, EDTA = 32 μg/mL, compound 19bg = 32 μg/mL and compound 19bh = 32 μg/mL). The concentration of each component in the combination groups was the same as that in the monotherapy treatment groups, except Zn2 was provided by ZnCl2 (32 μg/mL); bacterial growth rate was denoted as the mean ± SD (n = 3).

Minimum inhibitory concentration (MIC) of MEM-inhibitors combination against clinical NDM-producing isolates

Antimicrobial susceptibility testing was conducted by using microbroth dilution method according to Clinical and Laboratory Standards Institute (CLSI) guidelines M07-A9 to determine the MIC of compounds 19bg and 19bh in combination with MEM against NDM-1 producing Enterobacteriaceae isolates. All of the 10 non-duplicated MEM resistant isolates (MIC≥16 μg/mL) which were recovered from a teaching hospital of Zhengzhou University (Zhengzhou, China) were confirmed carrying NDM-1-gene by PCR and sequencing methods.

As shown in Table 5, when the inhibitors were added at the concentration of 32 μg/mL, both of MEM + compound 19bg (MIC90 = 1 μg/mL) and MEM + compound 19bh (MIC90 = 0.5 μg/mL) combinations showed comparable activities as MEM + EDTA (MIC90 = 1 μg/mL) to restore activity of MEM against NDM-1 producing isolates. While, when the concentration of inhibitors increased to 64 μg/mL, either combination of MEM + compound 19bg or MEM + compound 19bh (MIC90<0.03 μg/mL) exhibited better effect than MEM + EDTA (MIC90 = 0.06 μg/mL) for rescuing MEM activity.

Table 5

MIC (μg/mL) of MEM-inhibitors combination against NDM-1 positive clinical isolates.

No. of CRE	Bacteria speciesa	MIC (μg/mL)
		EDTA	19bg	19bh	MEM	MEM/EDTAb		MEM+19bg		MEM+19bh
		64	64	64	64	32	64	32	64	32	64
1	Klebsiella pneumoniae	>64	>64	>64	64	2	0.06	1	<0.03	0.5	<0.03
2	Klebsiella pneumoniae	>64	>64	>64	32	0.06	<0.03	<0.03	<0.03	<0.03	<0.03
3	Klebsiella pneumoniae	>64	>64	>64	32	1	<0.03	0.125	<0.03	0.06	<0.03
4	Enterobacter cloacae	>64	>64	>64	32	1	0.25	0.5	0.06	0.125	<0.03
5	Enterobacter cloacae	>64	>64	>64	8	0.125	0.06	0.25	<0.03	0.06	<0.03
6	Escherichia coil	>64	>64	>64	32	0.25	<0.03	1	<0.03	0.5	<0.03
7	Klebsiella oxytoca	>64	>64	>64	32	0.125	0.06	0.25	<0.03	0.125	<0.03
8	Klebsiella oxytoca	>64	>64	>64	16	1	0.06	0.5	<0.03	0.25	<0.03
9	Citrobacter freundii	>64	>64	>64	64	0.06	<0.03	<0.03	<0.03	0.125	<0.03
10	Providencia ewing	>64	>64	>64	32	0.5	<0.03	<0.03	<0.03	<0.03	<0.03

All of the strains tested were isolated from clinical Enterobacteriaceae bacteria producing NDM-1 enzyme.

The concentration of EDTA in the combined susceptibility test was fixed also at 32 and 64 μg/mL.

Bactericidal kinetics

To further study the antibacterial timeliness of these two compounds, bactericidal killing kinetics assay was employed to evaluate activities of compound 19bg or 19bh together with MEM against NDM-1 producing strain. As shown in Fig. 4, combination use of 19bh and MEM performed excellent bactericidal activity, led to a 3 log populations reduction of NDM-1 producing strain within 24 h.

Figure 4

The time-dependent killing of bacterial by MEM combined with compound 19bg or 19bh. NDM-1 enzyme-producing K. pneumoniae was challenged with compounds 19bg and 19bh combined with MEM.

Toxicity analysis

Hemolytic activity

Since compounds 19bg and 19bh had an excellent antibacterial effect, we evaluated those two compounds for hemolysis activity. The results of hemolytic activities were shown in Fig. 5. The red blood cells (RBCs) hemolysis rates of compounds 19bg and 19bh were lower than 5% even at an extremely high concentration of 1000 mg/mL, indicating both of which exhibited no toxicity to RBCs.

Figure 5

Hemolysis rate of red blood cells treated with different concentrations of compounds 19bg and 19bh. Mocka: treated with phosphate-buffered saline (PBS); Triton Xb: treated with 0.5% Triton X-100 detergent. Hemolysis rate is the mean ± SD (n = 3).

Cytotoxicity

To further investigate the cytotoxicity of these compounds, HeLa cells were treated respectively with 19bg and 19bh at the concentrations with a range of 0–64 μg/mL for 48 h. Then, a CCK-8 assay kit was used to detect the anti-proliferation effect of the different concentration of the two compounds, and results were shown in Fig. 6A and B. The half-maximal inhibitory concentrations (IC50) of the two compounds were found to be more than 64 μg/mL.

Figure 6

Cell viability and fluorescence microscopy images of HeLa cells after treatment with compound 19bg or 19bh for 48 h and staining with EdU and DAPI. (A) and (B) Cell viability of HeLa cells treated with compounds 19bg and 19bh, respectively; (C)‒(E) non-treated HeLa cells (control group); (F)‒(H) cells treated with compound 19bg (64 μg/mL); (I)‒(K) cells treated with compound 19bh (64 μg/mL). The observation of the fixed and stained cells was performed with an Olympus laser scanning confocal microscope, magnification × 200. The photomicrographs shown are representative of at least two independent experiments performed.

Furthermore, in order to confirm the results shown in Fig. 6A and B, an immunofluorescent staining assay, by using the EdU cell proliferation kit, was conducted to determine whether the DNA synthesis in living cells could be affected by these two compounds. As shown in Fig. 6, the results clearly showed that, after treatment by compound 19bg (Fig. 6F‒H) or 19bh (Fig. 6I‒K) with the concentration of 64 μg/mL, there were more than 50% cells survived when compared with untreated control (Fig. 6C‒E). These results indicated that the NDM-1 inhibitors exerted low toxicity against HeLa cells. hERG inhibition of candidate compounds (result was shown in Supporting Information Table S1) indicated compounds 19bg and 19bh performed very low hERG inhibitory activity and they are not considered to increase the risk of drug-induced long QT syndrome (LQTS).

Inhibition on NDM-1

We have identified compounds 19bg and 19bh as the most potent antibacterial agents in combination with MEM against clinical blaNDM-1-producing isolates among our synthesized compounds. The NDM-1 protein was expressed and purified for evaluating the inhibit activities of compounds 19bg, 19bh and the positive control EDTA against NDM-1 enzyme. The results showed each one had NDM-1 enzyme inhibitory action, and IC50 of compound 19bg, 19bh and EDTA were 0.233, 0.237 and 0.0837 μg/mL, respectively.

To further study the mechanism of action of these compounds, the inhibition constants Ki against NDM-1 were also determined. Ki of compounds 19bg and 19bh against NDM-1 were calculated to be 0.63 and 0.44 μmol/L, smaller than that of EDTA with Ki of 1.25 μmol/L. These results indicated that compounds 19bg and 19bh inhibited NDM-1 enzyme more effectively than EDTA.

Lineweaver–Burk Plots, (Fig. 7) results revealed that compounds 19bg and 19bh acted as uncompetitive inhibitors, characterized by the formation of substrate‒enzyme‒inhibitor complexes with MEM and NDM-1 enzymes in bacteria, preventing NDM-1 from hydrolyzing MEM. While the EDTA was non-competitive inhibitor, with no competitive effect on the substrate, reduced the activity of the NDM-1 and binded equally well to the NDM-1 whether or not it had already bound the substrate.

Figure 7

Lineweaver–Burk plots of inhibition of the NDM-1 hydrolysis activity by compounds 19bg and 19bh. The concentrations (μg/mL) of inhibitors (A) compound 19bg, 0 (▲), 0.02 (●), and 0.05 (■); (B) Compound 19bh, 0 (▲), 0.02 (●), and 0.05 (■); (C) EDTA, 0.05 (▲), 0.1 (●), and 0.5 (■).

Molecular docking studies

To predict the binding mode of uncompetitive inhibitors with NDM-1, our docking study was based on the X-ray structure of NDM-1 with a hydrolyzed MEM ligand (PDB ID: 5N0H). Considering that compounds 19bg and 19bh were uncompetitive inhibitors, we suspected that they might bind at allosteric sites of NDM-1. First, we used Cavity and CorrSite program40, 41, 42 to identify potential allosteric ligand binding sites, the results outputted 4 allosteric binding sites (data not show). Because H122 and D124 were critical residues for NDM-1 catalyzed carbapenem hydrolysis, D124 was directly involved in proton transport in the catalytic process. We speculated that compounds 19bg and 19bh were most likely to bind in the allosteric pocket (grey pocket in Fig. 8C) which was adjacent to the substrate pocket (cyan pocket in Fig. 8C).

Figure 8

Overview of ligand-binding sites of compounds 19bg and 19bh in human NDM-1 and the detailed interactions. Small molecules were shown as sticks. Protein NDM-1 (PDB ID: 5N0H) was shown as cartoon (cyan), for clarity, NDM-1 was shown in 50% transparent cartoon. NDM-1 residues interacting with compounds were shown as sticks, hydrogen bonds were shown as dashed lines (yellow), Zn2+ ions shown as spheres (magenta). (A) Interaction model of compound 19bg in NDM-1. Three hydrogen bonds between 19bg (green) and NDM-1 residues were shown, bonds length were also marked. The interactions (grey) between hydrolyzed MEM (grey) and NDM-1 were displayed partly. (B) Interaction model of compound 19bh in NDM-1. Two hydrogen bonds between 19bh (salmon) and NDM-1 residues were shown, bonds length were also marked. (C) By comparison, compounds 19bg and 19bh occupied the same allosteric binding pocket (grey) in NDM-1, the hydrolyzed MEM occupied the substrate binding pocket (cyan) in NDM-1. These two binding pockets were closely adjacent.

Thus, compounds 19bg and 19bh were docked into the allosteric pocket (grey pocket in Fig. 8C) of NDM-1 respectively. As shown in Fig. 8A, 19bg (colored in green) formed two H-bond interactions with D95, additional H-bond interaction was formed with Q123. Differently, as shown in Fig. 8B, compound 19bh (colored in salmon) formed only one H-bond interaction with D95, another H-bond interaction was also formed with Q123. These H-bond lengths were labeled in Fig. 8A and B. As shown in Fig. 8C, compounds 19bg and 19bh were occupied the same region of the allosteric pocket, and they all formed H-bond interaction with Q123. Q123 was between H122 and D124, which were participated in the catalytic process of substrates. We speculated that the binding of compounds 19bg and 19bh would affect the catalytic effect of NDM-1 on the substrate MEM.

In vivo activities of compounds 19bg and 19bh

As stated above, compounds 19bg and 19bh were effective uncompetitive inhibitors of NDM-1 and had good bacteriostatic effect when used in combination with MEM. Hence, we evaluated the safety and bacteriostatic activity of the compounds 19bg and 19bh in vivo. Firstly, subcutaneous injection of BALB/c mice with compound 19bg or 19bh at the concentration of 64 mg/kg to evaluate the safety. As shown in Fig. 9A, 24 h after injection, there was no obvious pathological change in the liver, spleen and kidney of the medicated groups by H&E staining compared with the control group. Considering the flip side, compared with the combination group, H&E staining of sepsis mice results showed that parenchymatous organs appeared degeneration and infiltration with the inflammatory cells, just like liver and kidney of control group and MEM single group. Meanwhile, the number of megakaryocyte neutrophils increased significantly in the paracortex of spleen in control group and MEM single group compared to other combination groups. As shown in Fig. 9C‒E, combination therapy with candidate compounds (19bg and 19bh, respectively) and MEM significantly reduced the bacterial loaded in the liver by subcutaneous injection as well as spleen and kidney.

Figure 9

Safety evaluation of compounds 19bg and 19bh and their rescue MEM activity in vivo. (A) For safety evaluation experiments, BALB/c mice were given a single dose of compound 19bg (64 mg/kg), compound 19bh (64 mg/kg) or PBS by subcutaneous injection. Data was the means ± standard error (SE) from three separate experiments (n = 2 per group). (B) H&E staining results of BALB/c mice were given a median lethal dose (5 × 106 CFU) of NDM-1 positive K. pneumoniae by intraperitoneal injection and treated with a single dose of MEM (10 mg/kg) and combination of MEM (10 mg/kg) plus compound 19bg (10 mg/kg), compound 19bh (10 mg/kg) or PBS by subcutaneous injection. (C)‒(E) Bacterial load in the liver, spleen and kidney of NDM-1 positive K. pneumoniae infected BALB/c mice with different treatment approaches were determined by selective plating. Data was the means ± SE from four separate experiments (n = 6 per group). (F) For survival experiments, BALB/c mice were given a lethal dose (1 × 107 CFU) of K. pneumoniae by intraperitoneal injection and treated with a single dose of MEM (10 mg/kg), compound 19bg (30 mg/kg), compound 19bh (30 mg/kg) and combination of MEM (10 mg/kg) plus compound 19bg (30 mg/kg) or compound 19bh (30 mg/kg), or PBS by subcutaneous injection. Data was the means ± SE from six separate experiments (n = 6 per group).

Remarkably, the combined medicine group extended the life span of the sepsis mice and was significantly superior to the control group and MEM single group (Supporting Information Table S2). Especially, in the combination group of compounds 19bh and MEM, no mice were killed after infecting with NDM-1 positive strain within 96 h (Fig. 9F). According to experimental data, compound 19bh could restore the susceptibility of the MEM against NDM-1-positive K. pneumoniae in vivo and could significantly prolong the survival time of sepsis mice. The results of pharmacokinetic (PK) profiles indicated that compound 19bh had a favorable PK property (Supporting Information Fig. S4 and Table S3).

Conclusions

In this study, a total of 60 thiosemicarbazone derivatives were designed and synthetized and their potential antibacterial activities against NDM-1 positive clinical isolates combined with meropenem were evaluated for SAR studies. Among them, compounds 19bg and 19bh exhibited promising inhibitory activity during in vitro antimicrobial susceptibility tests and performed good inhibitory activity against NDM-1 (IC50 = 0.233 and 0.237 μg/mL, respectively) with uncompetitive inhibition toward MEM‒enzyme complex. Moreover, docking analysis illustrated that compounds 19bg and 19bh were most likely to bind in the allosteric pocket which was adjacent to the substrate pocket. In addition, the combination of compounds 19bh and MEM was markedly effective in treating infections caused by NDM-1 positive strain and prolonging the survival time of sepsis mice with low toxicity. In summary, for the first time, a series of thiosemicarbazone-based, specific NDM-1 inhibitors were explored and exhibited significant antibacterial activities in vivo, which might serve as leading compounds targeting NDM-1 producing drug-resistance clinical isolates.

Experimental

General information

All the reagents and solvents used in the chemical synthesis were obtained from commercial sources and were used without further purification. All the chemistry reactions were monitored by thin layer chromatography (TLC) on silica gel which was purchased from Qingdao Haiyang Chemical Co., Ltd., Qingdao, China. Column chromatography was carried out at medium pressure using silica gel (200–300 mesh) purchased from Qingdao Haiyang Chemical Co., Ltd. 1H NMR and 13C NMR spectra data were obtained on Bruker AVANCE Ⅲ 400 MHz spectrometer (Bruker Instruments, Inc., Ettlingen, Germany), chemical shifts (δ) were reported in parts per million (ppm) relative to tetramethylsilane (TMS) and J values were reported in Hertz (Hz). The NMR spectra data of compounds were provided in Supporting Information. High-resolution mass spectra (HR-MS) was recorded on a Waters Micromass Q-T of Micromass spectrometer by electrospray ionization (ESI; Waters Micro Q-TOF, Milford, MA, USA). The purity of the target compounds was determined by reverse-phase ultra-performance liquid chromatography (UPLC; Waters, ACQUITY H-Class, Milford, MA, USA) analysis. The signal was monitored at 235 nm with a UV detector. Water/acetonitrile served as a binary gradient mobile phase, with a flow rate of 0.3 mL/min.

General procedure for preparation of 8a‒g

Available substituted anilines 7a‒g (3 mmol) were dissolved in THF (8 mL) in a 25 mL round-bottom flask, then CS2 (30 mmol) and Et3N (3 mmol) were added dropwise in turn. While reacting for 3 h at r.t. and monitoring via TLC until the complete conversion of reagents 7a‒g. The system was cooled in an ice bath, then (Boc)2O (3 mmol, dissolved in 2 mL THF) and DMAP (0.03 mmol, dissolved in 0.5 mL THF) were added immediately, followed by stirring for another 0.3 h. Then the system was concentrated under reduced pressure and purified by column chromatography to provide phenyl isothiocyanates 8a‒g with 45%–70% yield.

General procedure for preparation of 9a‒g

In a 25 mL round-bottom flask, 8a‒g (1.5 mmol) was dissolved in methanol (5 mL), following by adding hydrazine hydrate (80%, w/w, 4.5 mmol) dropwise, while heating under reflux for 2–3 h and monitoring via TLC until the complete conversion of the reagents. Subsequently, the system was cooled to r.t., and filtered, washed with cold methanol to provide 9a‒g in 70%–75% yield.

General procedure for preparation of 15a‒d, 16 and 17

To a solution of 10a‒d (5 mmol) and NaOH (5 mmol) in water (15 mL) was injected compound 13 (5 mmol) dropwise, the mixture was stirred at r.t. for 3–4 h. Subsequently, compound 14 (5 mmol) was added in portions, after stirring for 2 h, hydrazine hydrate (80%, w/w, 15 mmol) was added, then the mixture was heated at 65 °C for 2–3 h until a large amount of solids was precipitated, kept stirring for another 0.5 h, then cooled to r.t., and allowed to stand overnight for crystallization to afford compounds 15a‒d with the yield ranging from 40% to 55%. These intermediates were used for the next step without purification. Likewise, compounds 16 and 17 were derived from reagents 13 and 14 respectively in the same way.

General procedure for preparation of 19a‒bf

To the solution of 9a‒g (1.2 mmol) in methanol (5 mL) containing catalytic amount of acetic acid (0.012 mmol), appropriate aldehyde or ketone 18 (1.3 mmol) was added dropwise. The mixture was heated under reflux at 70 °C for 2 h, then cooled to r.t., the precipitation was filtered and recrystallized with methanol, or purified by column chromatography (ethyl acetate/petroleum ether = 1:10) to get compounds 19a‒19ad as solid in 65%–85% yield. Likewise, the rest of target compounds were prepared by reagents 15a‒d, 16 and 17 in the same way. Noteworthy, compounds 19ba‒19bc were not directly derived from the intermediate 15c, the additional reaction occurred in the solvent of dioxane with conc. HCl as the reagent to remove the Boc protective group.

All the target compounds 19 were synthesized in >95% purity degree. The detailed information on characterization of some representative target compounds were shown below, the rest was summarized in Supporting Information. In addition, the UPLC chromatogram of compounds 19ba and 19bd were shown in Supporting Information Figs. S6 and S7.

(E)-N-Phenyl-2-(pyridin-2-ylmethylene)hydrazine-1-carbothioamide (19e)

White solid, Yield 78%, m.p. 200–201 °C. 1H NMR (400 MHz, DMSO-d6, ppm) δ 12.04 (s, 1H, ‒NH‒, D2O exchangeable), 10.26 (s, 1H, ‒NH‒, D2O exchangeable), 8.59 (d, J = 4.3 Hz, 1H, Ar–H), 8.45 (d, J = 8.0 Hz, 1H, Ar–H), 8.21 (s, 1H, –CHN‒), 7.85 (td, J = 7.8, 1.3 Hz, 1H, Ar–H), 7.56 (d, J = 7.6 Hz, 2H, Ar–H), 7.42–7.37 (m, 3H, Ar–H), 7.23 (t, J = 7.4 Hz, 1H, Ar–H). 13C NMR (100 MHz, DMSO-d6, ppm) δ 176.41, 153.15, 149.31, 143.05, 138.94, 136.47, 128.09, 126.07, 125.52, 124.22, 120.60. HR-MS (ESI), Calcd. C13H12N4S, [M+Na]+ m/z: 279.0680, Found: 279.0681.

(E)-N-Phenyl-2-(pyridin-3-ylmethylene)hydrazine-1-carbothioamide (19f)

White solid, Yield 85%, m.p. 234–235 °C. 1H NMR (400 MHz, DMSO-d6, ppm) δ 11.98 (s, 1H, ‒NH‒, D2O exchangeable), 10.22 (s, 1H, ‒NH‒, D2O exchangeable), 9.04 (d, J = 1.7 Hz, 1H, Ar–H), 8.59 (dd, J = 4.8, 1.5 Hz, 1H, Ar–H), 8.39 (dt, J = 8.0, 1.7 Hz, 1H, Ar–H), 8.19 (s, 1H, –CHN‒), 7.56 (d, J = 7.7 Hz, 2H, Ar–H), 7.45 (dd, J = 7.9, 4.8 Hz, 1H, Ar–H), 7.38 (t, J = 7.8 Hz, 2H, Ar–H), 7.22 (t, J = 7.4 Hz, 1H, Ar–H). 13C NMR (100 MHz, DMSO-d6, ppm) δ 176.26, 150.44, 149.05, 139.86, 138.98, 134.17, 130.01, 128.04, 126.04, 125.43, 123.69. HR-MS (ESI), Calcd. C13H12N4S, [M+Na]+ m/z: 279.0680, Found: 279.0681.

(E)-N-Phenyl-2-(pyridin-4-ylmethylene)hydrazine-1-carbothioamide (19g)

White solid, Yield 78%, m.p. 194–195 °C. 1H NMR (400 MHz, DMSO-d6, ppm) δ 12.08 (s, 1H, ‒NH‒, D2O exchangeable), 10.29 (s, 1H, ‒NH‒, D2O exchangeable), 8.63 (d, J = 6.0 Hz, 2H, Ar–H), 8.12 (s, 1H, –CHN‒), 7.89 (d, J = 6.0 Hz, 2H, Ar–H), 7.55 (d, J = 7.7 Hz, 2H, Ar–H), 7.39 (t, J = 7.8 Hz, 2H, Ar–H), 7.24 (t, J = 7.4 Hz, 1H, Ar–H). 13C NMR (100 MHz, DMSO-d6, ppm) δ 176.58, 150.00, 141.26, 140.00, 138.90, 128.11, 126.17, 125.61, 121.36. HR-MS (ESI), Calcd. C13H12N4S, [M+Na]+ m/z: 279.0680, Found: 279.0682.

(E)-N-(4-Bromophenyl)-2-(pyridin-2-ylmethylene)hydrazine-1-carbothioamide (19m)

White solid, Yield 65%, m.p. 225–226 °C. 1H NMR (400 MHz, DMSO-d6, ppm) δ 12.12 (s, 1H, ‒NH‒, D2O exchangeable), 10.28 (s, 1H, ‒NH‒, D2O exchangeable), 8.59 (d, J = 4.3 Hz, 1H, Ar–H), 8.43 (d, J = 8.0 Hz, 1H, Ar–H), 8.21 (s, 1H, –CHN‒), 7.86 (td, J = 7.7, 1.3 Hz, 1H, Ar–H), 7.60–7.54 (m, 4H, Ar–H), 7.43–7.38 (m, 1H, Ar–H). 13C NMR (100 MHz, DMSO-d6, ppm) δ 176.33, 153.04, 149.36, 143.41, 138.36, 136.48, 130.92, 127.97, 124.30, 120.63, 117.78. HR-MS (ESI), Calcd. C13H11BrN4S, [M+Na]+ m/z: 356.9785, Found: 356.9785.

(E)-N-(2-Bromophenyl)-2-(1-(pyridin-2-yl)ethylidene)hydrazine-1-carbothioamide (19r)

White solid, Yield 73%, m.p. 161–162 °C. 1H NMR (400 MHz, DMSO-d6, ppm) δ 10.93 (s, 1H, ‒NH‒, D2O exchangeable), 10.19 (s, 1H, ‒NH‒, D2O exchangeable), 8.64–8.58 (m, 1H, Ar–H), 8.52 (d, J = 8.1 Hz, 1H, Ar–H), 7.86–7.78 (m, 1H, Ar–H), 7.74 (ddd, J = 11.8, 8.0, 1.4 Hz, 2H, Ar–H), 7.47–7.38 (m, 2H, Ar–H), 7.25 (td, J = 7.8, 1.6 Hz, 1H, Ar–H), 2.48 (s, 3H, –CH3). 13C NMR (100 MHz, DMSO-d6, ppm) δ 177.59, 154.46, 149.16, 148.50, 137.94, 136.45, 132.38, 129.73, 128.04, 127.70, 124.13, 121.36, 121.03, 12.52. HR-MS (ESI), Calcd. C14H13BrN4S, [M+Na]+ m/z: 370.9942, Found: 370.9942.

(E)-N-Phenyl-2-(1-(pyridin-2-yl)ethylidene)hydrazine-1-carbothioamide (19ac)

Yellow solid, Yield 76%, m.p. 192–193 °C. 1H NMR (400 MHz, DMSO-d6, ppm) δ 10.67 (s, 1H, ‒NH‒, D2O exchangeable), 10.19 (s, 1H, ‒NH‒, D2O exchangeable), 8.61 (dd, J = 4.8, 0.7 Hz, 1H, Ar–H), 8.54 (d, J = 8.1 Hz, 1H, Ar–H), 7.82 (td, J = 8.0, 1.7 Hz, 1H, Ar–H), 7.56 (d, J = 7.6 Hz, 2H, Ar–H), 7.40 (dt, J = 14.2, 4.5 Hz, 3H, Ar–H), 7.23 (t, J = 7.4 Hz, 1H, Ar–H), 2.47 (s, 3H, –CH3). 13C NMR (100 MHz, DMSO-d6, ppm) δ 177.24, 154.50, 149.17, 148.45, 139.12, 136.35, 128.08, 126.11, 125.52, 124.08, 121.21, 12.46. HR-MS (ESI), Calcd. C14H14N4S, [M+H]+ m/z: 271.1017, Found: 271.1016.

(E)-N′-(2-Hydroxybenzylidene)benzohydrazide (19ae)

Yellow solid, Yield 77%, m.p. 172–173 °C. 1H NMR (400 MHz, DMSO-d6, ppm) δ 12.14 (s, 1H, –OH, D2O exchangeable), 11.32 (s, 1H, ‒NH‒, D2O exchangeable), 8.66 (s, 1H, –CHN‒), 7.96 (d, J = 7.4 Hz, 2H, Ar–H), 7.63 (t, J = 7.2 Hz, 1H, Ar–H), 7.56 (t, J = 7.1 Hz, 3H, Ar–H), 7.32 (t, J = 7.7 Hz, 1H, Ar–H), 7.00–6.89 (m, 2H, Ar–H). 13C NMR (100 MHz, DMSO-d6, ppm) δ 162.79, 157.43, 148.20, 132.76, 131.98, 131.38, 129.47, 128.54, 127.62, 119.33, 118.65, 116.39. HR-MS (ESI), Calcd. C14H12N2O2, [M+H]+ m/z: 241.0977, Found: 241.0970.

(E)-N′-(1-(5-Chloro-2-hydroxyphenyl)ethylidene)benzohydrazide (19af)

Yellow solid, Yield 79%, m.p. 206–207 °C. 1H NMR (400 MHz, DMSO-d6, ppm) δ 13.46 (s, 1H, –OH, D2O exchangeable), 11.45 (s, 1H, ‒NH‒, D2O exchangeable), 7.96 (s, 1H, Ar–H), 7.94 (s, 1H, Ar–H), 7.68–7.61 (m, 2H, Ar–H), 7.55 (t, J = 7.5 Hz, 2H, Ar–H), 7.34 (d, J = 8.7 Hz, 1H, Ar–H), 6.95 (d, J = 8.7 Hz, 1H, Ar–H), 2.50 (s, 3H, –CH3). 13C NMR (100 MHz, DMSO-d6, ppm) δ 164.51, 157.42, 156.62, 132.72, 132.08, 130.75, 128.41, 128.18, 127.74, 122.07, 120.81, 119.08, 14.16. HR-MS (ESI), Calcd. C15H13ClN2O2, [M+H]+ m/z: 289.0744, Found: 289.0738.

(E)-N′-(1-(Pyridin-2-yl)ethylidene)morpholine-4-carbothiohydrazide (19ag)

Yellow solid, Yield 75%, m.p. 213–215 °C. 1H NMR (400 MHz, DMSO-d6, ppm) δ 9.96 (s, 1H, ‒NH‒, D2O exchangeable), 8.67 (d, J = 52.9 Hz, 1H, Ar–H), 8.11–7.80 (m, 2H, Ar–H), 7.47 (d, J = 47.2 Hz, 1H, Ar–H), 3.95 (s, 4H, –CH2CH2‒), 3.65 (s, 4H, –CH2CH2‒), 2.40 (s, 3H, –CH3). 13C NMR (101 MHz, CDCl3, ppm) δ 188.17, 186.70, 136.24, 133.63, 123.76, 122.36, 71.03, 70.91, 54.00, 52.98. HR-MS (ESI), Calcd. C12H17N4OS, [M+H]+ m/z: 265.1123, Found: 265.1122.

(E)-N′-(Pyridin-2-ylmethylene)piperidine-1-carbothiohydrazide (19ai)

Brown solid, Yield 75%, m.p. 213–215 °C. 1H NMR (400 MHz, DMSO-d6, ppm) δ 11.25 (s, 1H, ‒NH‒, D2O exchangeable), 8.58 (d, J = 4.9 Hz, 1H, Ar–H), 8.17 (s, 1H, –CHN‒), 7.84 (d, J = 4.2 Hz, 2H, Ar–H), 7.37 (q, J = 4.5 Hz, 1H, Ar–H), 3.88 (t, J = 4.9 Hz, 4H, –CH2), 1.74–1.46 (m, 6H, –CH2CH2CH2‒). 13C NMR (101 MHz, DMSO-d6, ppm) δ 180.15, 153.47, 149.41, 143.35, 136.71, 123.85, 119.27, 51.23, 25.75, 23.85. HR-MS (ESI), Calcd. C12H16N4S, [M+H]+ m/z: 249.1174, Found: 249.1167.

(E)-N-Cyclohexyl-2-(1-(pyridin-2-yl)ethylidene)hydrazine-1-carbothioamide (19aj)

Yellow solid, Yield 75%, m.p. 171–172 °C. 1H NMR (400 MHz, DMSO-d6, ppm) δ 10.26 (s, 1H, ‒NH‒, D2O exchangeable), 8.59 (d, J = 4.7 Hz, 1H, ‒NH‒, D2O exchangeable), 8.29 (d, J = 8.1 Hz, 1H, Ar–H), 8.14 (d, J = 8.5 Hz, 1H, Ar–H), 7.84 (td, J = 7.8, 1.7 Hz, 1H, Ar–H), 7.44–7.33 (m, 1H, Ar–H), 4.29–4.15 (m, 1H, –CH), 2.40 (s, 3H, –CH3), 1.94–1.86 (m, 2H, –CH2‒), 1.74 (d, J = 12.9 Hz, 2H, –CH2‒), 1.62 (d, J = 12.5 Hz, 1H, –CH2‒), 1.47 (qd, J = 12.2, 3.1 Hz, 2H, –CH2‒), 1.37–1.24 (m, 2H, –CH2‒), 1.22–1.10 (m, 1H, –CH2‒). 13C NMR (100 MHz, DMSO-d6, ppm) δ 176.72, 154.62, 148.50, 148.25, 136.44, 123.92, 120.69, 52.81, 31.59, 25.10, 24.84, 12.26. HR-MS (ESI), Calcd. C14H20N4S, [M+Na]+ m/z: 299.1306, Found: 299.1308.

(E)-N,N-Dimethyl-2-(1-(pyridin-2-yl)ethylidene)hydrazine-1-carbothioamide (19ak)

Yellow solid, Yield 75%, m.p. 195–196 °C. 1H NMR (400 MHz, DMSO-d6, ppm) δ 9.60 (s, 1H, ‒NH‒, D2O exchangeable), 8.59 (s, 1H, Ar–H), 7.89–7.77 (m, 2H, Ar–H), 7.39 (s, 1H, Ar–H), 3.32 (s, 6H, –CH3), 2.40 (s, 3H, –CH3). 13C NMR (101 MHz, DMSO-d6, ppm) δ 179.92, 152.20, 147.43, 138.67, 124.34, 124.19, 120.02, 42.18, 21.87. HR-MS (ESI), Calcd. C10H14N4S, [M+H]+ m/z: 223.1017, Found: 223.1015.

(E)-N′-(Pyridin-2-ylmethylene)morpholine-4-carbothiohydrazide (19ao)

White solid, Yield 77%, m.p. 201–203 °C. 1H NMR (400 MHz, DMSO-d6, ppm) δ 11.44 (s, 1H, ‒NH‒, D2O exchangeable), 8.58 (dd, J = 4.8, 1.5 Hz, 1H, Ar–H), 8.18 (s, 1H, –CHN‒), 7.85 (dt, J = 3.4, 1.5 Hz, 2H, Ar–H), 7.38 (td, J = 5.2, 3.0 Hz, 1H, Ar–H), 3.94 (t, J = 4.8 Hz, 4H, –CH2CH2‒), 3.76–3.62 (m, 4H, –CH2CH2‒). 13C NMR (101 MHz, DMSO-d6, ppm) δ 180.83, 153.26, 149.43, 144.04, 136.77, 124.02, 119.47, 65.97, 50.58. HR-MS (ESI), Calcd. C11H14N4OS, [M+H]+ m/z: 251.0967, Found: 251.0967.

(E)-N′-(4-Bromobenzylidene)morpholine-4-carbothiohydrazide (19ap)

Yellow solid, Yield 80%, m.p. 169–170 °C. 1H NMR (400 MHz, DMSO-d6, ppm) δ 11.27 (s, 1H, ‒NH‒, D2O exchangeable), 8.12 (s, 1H, –CHN‒), 7.66–7.53 (m, 4H, Ar–H), 3.93–3.89 (m, 4H, –CH2CH2‒), 3.70–3.65 (m, 4H, –CH2CH2‒). 13C NMR (100 MHz, DMSO-d6, ppm) δ 180.75, 142.66, 133.58, 131.82, 128.61, 122.87, 65.98, 50.47. HR-MS (ESI), Calcd. C12H14BrN3OS, [M+H]+ m/z: 330.0099, Found: 330.0090.

(E)-N′-((1H-Indol-3-yl)methylene)morpholine-4-carbothiohydrazide (19ar)

Yellow solid, Yield 77%, m.p. 165–166 °C. 1H NMR (400 MHz, DMSO-d6, ppm) δ 11.57 (s, 1H, ‒NH‒, D2O exchangeable), 10.98 (s, 1H, ‒NH‒, D2O exchangeable), 8.36 (s, 1H, –CHN‒), 8.17 (d, J = 7.6 Hz, 1H, Ar–H), 7.78 (d, J = 2.7 Hz, 1H, Ar–H), 7.43 (d, J = 7.9 Hz, 1H, Ar–H), 7.22–7.11 (m, 2H, Ar–H), 3.96–3.93 (m, 4H, –CH2CH2‒), 3.72–3.68 (m, 4H, –CH2CH2‒). 13C NMR (100 MHz, DMSO-d6, ppm) δ 180.34, 141.91, 137.02, 130.38, 124.03, 122.57, 121.76, 120.48, 111.80, 111.45, 66.03, 50.17. HR-MS (ESI), Calcd. C14H16N4OS, [M−H]− m/z: 287.0967, Found: 287.0966.

(E)-N′-((5-Bromopyridin-2-yl)methylene)morpholine-4-carbothiohydrazide (19as)

Yellow solid, Yield 80%, m.p. 205–206 °C. 1H NMR (400 MHz, DMSO-d6, ppm) δ 11.49 (s, 1H, ‒NH‒, D2O exchangeable), 8.71 (d, J = 2.3 Hz, 1H, Ar–H), 8.15 (s, 1H, –CHN‒), 8.09 (dd, J = 8.6, 2.3 Hz, 1H, Ar–H), 7.80 (d, J = 8.6 Hz, 1H, Ar–H), 3.93 (t, J = 4.7 Hz, 4H, –CH2CH2‒), 3.68 (t, J = 4.7 Hz, 4H, –CH2CH2‒). 13C NMR (101 MHz, DMSO-d6, ppm) δ 180.76, 152.13, 150.13, 142.83, 139.46, 121.03, 120.29, 65.94, 50.51. HR-MS (ESI), Calcd. C11H13BrN4OS, [M+H]+ m/z: 329.0072, Found: 329.0064.

(E)-N′-(2-Hydroxybenzylidene)piperazine-1-carbothiohydrazide (19ba)

Yellow solid, Yield 75%, m.p. 215–216 °C. 1H NMR (400 MHz, DMSO-d6, ppm) δ 12.12 (s, 1H, ‒NH‒, D2O exchangeable), 8.41 (s, 1H, –CHN‒), 7.35 (dd, J = 7.9, 1.8 Hz, 1H, Ar–H), 7.23 (td, J = 7.7, 1.7 Hz, 1H, Ar–H), 6.86 (dd, J = 7.8, 5.8 Hz, 2H, Ar–H), 3.89 (dd, J = 6.1, 3.9 Hz, 4H, –CH2CH2‒), 2.88–2.71 (m, 4H, –CH2CH2‒). 13C NMR (101 MHz, DMSO-d6, ppm) δ 179.84, 157.28, 145.79, 130.22, 129.67, 118.86, 118.70, 116.37, 48.78, 44.97. HR-MS (ESI), Calcd. C12H16N4OS, [M+H]+ m/z: 265.1123, Found: 265.1116.

(E)-N′-(2-Methoxybenzylidene)piperazine-1-carbothiohydrazide (19bb)

Yellow solid, Yield 81%, m.p. 204–205 °C. 1H NMR (400 MHz, DMSO-d6, ppm) δ 8.54 (s, 1H, –CHN‒), 7.75 (dd, J = 7.8, 1.7 Hz, 1H, Ar–H), 7.45–7.36 (m, 1H, Ar–H), 7.10 (d, J = 8.4 Hz, 1H, Ar–H), 7.01 (t, J = 7.5 Hz, 1H, Ar–H), 4.09 (t, J = 5.0 Hz, 4H, –CH2CH2‒), 3.86 (s, 3H, –CH3), 3.33–3.02 (m, 4H, –CH2CH2‒). 13C NMR (101 MHz, DMSO-d6, ppm) δ 181.08, 157.66, 140.33, 131.45, 120.73, 111.84, 55.72, 47.05, 42.93. HR-MS (ESI), Calcd. C13H18N4OS, [M+H]+ m/z: 279.1280, Found: 279.1269.

(E)-N′-(2-Ethoxybenzylidene)piperazine-1-carbothiohydrazide (19bc)

Yellow solid, Yield 83%, m.p. 209–210 °C. 1H NMR (400 MHz, DMSO-d6, ppm) δ 11.13 (s, 1H, ‒NH‒, D2O exchangeable), 8.52 (s, 1H, –CHN‒), 7.74 (d, J = 7.7 Hz, 1H, Ar–H), 7.36 (t, J = 8.0 Hz, 1H, Ar–H), 7.07 (d, J = 8.4 Hz, 1H, Ar–H), 6.98 (t, J = 7.5 Hz, 1H, Ar–H), 4.11 (q, J = 6.9 Hz, 2H, –CH2‒), 3.86 (t, J = 4.8 Hz, 4H, –CH2CH2‒), 2.82 (t, J = 4.9 Hz, 4H, –CH2CH2‒), 1.38 (t, J = 6.9 Hz, 3H, –CH3). 13C NMR (101 MHz, DMSO-d6, ppm) δ 157.06, 140.62, 120.69, 112.84, 43.03, 42.67, 21.06, 14.63. HR-MS (ESI), Calcd. C14H2ON4OS, [M+H]+ m/z: 293.1436, Found: 293.1426.

(E)-N′-(2-Hydroxybenzylidene)-4-methylpiperazine-1-carbothiohydrazide (19bd)

Yellow solid, Yield 82%, m.p. 176–177 °C. 1H NMR (400 MHz, DMSO-d6, ppm) δ 11.59 (s, 1H, ‒NH‒, D2O exchangeable), 8.46 (s, 1H, –CHN‒), 7.41 (dd, J = 7.9, 1.5 Hz, 1H, Ar–H), 7.30–7.23 (m, 1H, Ar–H), 6.93–6.87 (m, 2H, Ar–H), 4.01–3.74 (m, 4H, –CH2CH2‒), 2.43–2.38 (m, 4H, –CH2CH2‒), 2.23 (s, 3H, –CH3). 13C NMR (100 MHz, DMSO-d6, ppm) δ 171.98, 157.05, 146.13, 130.82, 129.92, 119.04, 118.47, 116.49, 54.14, 48.25, 45.26, 21.03. HR-MS (ESI), Calcd. C13H18N4OS, [M+H]+ m/z: 279.1280, Found: 279.1278.

(E)-N′-(2-Methoxybenzylidene)-4-methylpiperazine-1-carbothiohydrazide (19be)

Yellow solid, Yield 85%, m.p. 178–179 °C. 1H NMR (400 MHz, DMSO-d6, ppm) δ 11.16 (s, 1H,‒NH‒, D2O exchangeable), 8.49 (s, 1H, –CHN‒), 7.74 (d, J = 7.7 Hz, 1H, Ar–H), 7.39 (t, J = 8.0 Hz, 1H, Ar–H), 7.09 (d, J = 8.4 Hz, 1H, Ar–H), 7.00 (t, J = 7.6 Hz, 1H, Ar–H), 3.89 (t, J = 4.9 Hz, 4H, –CH2CH2‒), 3.84 (s, 3H, –CH3), 2.39 (t, J = 4.9 Hz, 4H, –CH2CH2‒), 2.21 (s, 3H). 13C NMR (101 MHz, DMSO-d6, ppm) δ 180.43, 157.53, 139.49, 131.11, 125.21, 122.39, 120.71, 111.79, 55.65, 54.42, 49.65, 45.38. HR-MS (ESI), Calcd. C14H2ON4OS, [M+H]+ m/z: 293.1436, Found: 293.1427.

(E)-N′-(2-Ethoxybenzylidene)-4-methylpiperazine-1-carbothiohydrazide (19bf)

Yellow solid, Yield 78%, m.p. 182–183 °C. 1H NMR (400 MHz, DMSO-d6, ppm) δ 11.18 (s, 1H, ‒NH‒, D2O exchangeable), 8.52 (s, 1H, –CHN‒), 7.74 (d, J = 7.7 Hz, 1H, Ar–H), 7.36 (t, J = 8.0 Hz, 1H, Ar–H), 7.06 (d, J = 8.4 Hz, 1H, Ar–H), 6.98 (t, J = 7.6 Hz, 1H, Ar–H), 4.10 (q, J = 6.9 Hz, 2H, –CH2‒), 3.89 (t, J = 4.8 Hz, 4H, –CH2CH2‒), 2.40 (t, J = 4.9 Hz, 4H, –CH2CH2‒), 2.21 (s, 3H, –CH3), 1.37 (t, J = 7.0 Hz, 3H, –CH3). 13C NMR (101 MHz, DMSO-d6, ppm) δ 180.49, 156.94, 139.73, 131.06, 125.20, 122.64, 120.66, 112.83, 63.82, 54.43, 49.67, 45.38, 14.64. HR-MS (ESI), Calcd. C15H22N4OS, [M+H]+ m/z: 307.1593, Found: 307.1588.

Biological

Standard agar disk diffusion assay

Standard operation was performed according to CLSI M07-A9 guidelines. E. coli ATCC 25922(American Type Culture Collection, Maryland, USA) was used as control. DMSO (Tianjing Hengxing Chemical Reagent Co., Ltd., Tianjin, China) was used as a solution. After 6 h of culture, 109 CFU/mL strains were taken and diluted to 105 CFU/mL with Mueller-Hinton Broth (MHB, Beijing Aoboxing Biotechnology Co., Ltd., Beijing, China). The bacterial solution was evenly coated on Mueller-Hinton Agar (MHA, Beijing Aoboxing Biotechnology Co., Ltd.) plate. MEM/MEM-EDTA or test paper was analyzed by agar disc diffusion method with or without 0.5 mol/L EDTA (Beijing Solarbio Science and Technology Co., Ltd., Beijing, China)/test material. After 16–18 h, the bacteriostatic circle was cultured at 37 °C. The results were in accordance with the CLSI standard.

Minimum inhibitory concentration (MIC) assay

According to the regulations of CLSI, broth microdilution method was used to evaluate the antimicrobial activity. Bacteria were cultured in MHB to 109 CFU/mL. After diluting it 1000 times, bacteria with a concentration of about 106 CFU/mL was obtained. The compounds were dissolved in water and diluted with MHB. MEM was gradient diluted to 0.3125–16 mg/mL. The diluted MHB drug (256 or 128 μg/mL) and the diluted bacterial culture medium (106 CFU/mL) was added to 96-well plate. The final volume of each hole in 96-well plate was 200 μL. The concentration of other compounds in 96-well plate was 4 times as high as that in bacterial culture medium, except that the dilution of bacterial culture medium was 2 times as 5 × 105 CFU/mL. The plate was incubated at 37 °C for 16–18 h. In addition to the blank control group (medium only), the positive control group (containing only bacterial solution) was set up in all experiments. The MIC of the tested drug showed no bacterial growth in the tube at the lowest concentration. The reported experimental results were the average of three independent experiments.

Metal suppression experiments

Zinc homeostasis testing experiment was carried out according to the MIC of the combination drug. Zn2+ plays a major role in MBL. Therefore, we designed a Zn2+ metal ion inhibition experiment to test the sensitivity of the compound to MEM in the presence of Zn2+ ions. The concentration of monotherapy treatment groups of MEM = 2 μg/mL, EDTA = 32 μg/mL, compound 19bg = 32 μg/mL and compound 19bh = 32 μg/mL. The operation was the same except that an equimolar amount of ZnCl2 (32 μg/mL, Heowns Biochem Technologies LLC., Tianjin, China) was added. Incubated at 37 °C for 1 h and measured the absorbance at OD = 600 nm. Results were plotted as mean ± SD of three samples.

Haemolytic activity

Preparation of 5% red blood cell suspension: fresh sterile sheep blood (Bianzhen Biotechnology Co., Ltd., Nanjing, China) was suspended in sterile 1 × PBS buffer (Beijing Solarbio Science and Technology Co., Ltd.) and diluted by centrifugation (1500×g, 10 min). In the experiment, a hemolysis assay was carried out using water as a solvent (final concentration not exceeding 0.5%) to prevent false positive results, and 0.1% Triton-X detergent (Beijing Solarbio Science and Technology Co., Ltd.) was used as a positive control. A 150 μL/well 5% red blood cell suspension was pipetted into a 96-well plate. The drug was diluted in 1 × PBS buffer, 50 μL per well, and three parallel wells were made at the same concentration. After incubating for 1 h in a 37 °C incubator, the plates were placed in a centrifuge (3500 rpm, 5 min, TDZ5-WS, Xiangyi Laboratory Instrument Development Co., Ltd., Changsha, China). Transfer 100 μL/well of supernatant to a new 96-well plate. Finally, the results were measured at 540 nm on a microplate reader (318MC, Boteng Instrument Equipment Co., Ltd., Shenyang, China). The percentage of hemolysis was determined as Eq. (1):where A was the test well, A0 was the negative control, and ATotal was positive control.

CCK-8 and EdU cell proliferation assay

Target HeLa cells (Thermo Fisher scientific, Waltham, MA, USA) were diluted into 5 × 104 cells/mL with the medium, 100 μL of the cell suspension was seeded in each well of 96-well cell culture plates and incubated in the presence of different concentrations of compounds 19bg and 19bh at 37 °C for 48 h. Cells proliferation was detected with CCK-8 cell counting kit (vazyme, A311-01, Nanjing, China), and the protocol was followed according to the manufacturer's instructions. All measurements were performed in triplicate. DNA synthesis in living cells was detected by using the EdU cell proliferation assay Kit (Beyotime, C0075S, Shanghai, China), and the protocol was followed according to the manufacturer's instructions. The EdU fluorescence was observed under an Olympus laser scanning confocal microscope (Leica microsystems, Shanghai, China).

Enzyme inhibition

The final concentrations of NDM-1 inhibitor compounds 19bg and 19bh were set to 20, 10, 5, 2.5, 1.25, 0.625, 0.3125 and 0 μg/mL in a 200 μL enzyme reaction system; three parallel control groups were set up on a 96-well UV plate according to the corresponding concentration dilution gradient. Dilute the NDM-1 enzyme solution in 10 mmol/L HEPES buffer, add 98 μL of the diluted enzyme solution and two μL of the screening sample to each well, incubate at 30 °C for 15 min; then 100 μL of the diluted substrate solution was added to each well to initiate the reaction with end concentration of 250 μmol/L; the 96-well UV plate was placed in the Multimode Plate Reader (EnSpire, PerkinElmer, Waltham, MA, USA). It was measured at intervals of 1 min at 30 °C for a total of 60 min. The absorbance at OD300 was measured. The inhibition rate of the enzyme was calculated. The IC50 of each inhibitor was analyzed using GraphPad Prism version 5.0 (GraphPad Software, San Diego, CA, USA).

Expression and purification of NDM-1 enzyme

The blaNDM-1 sequence that encoding N-terminus truncated protein (residue G29‒R270) was amplified by PCR (Life technologies, QuantStudio 6 Flex, Thermo Fisher scientific, MA, USA) and purified blaNDM-1 PCR product was cloned into the pET28a vector between NdeI and XhoI restriction site. The recombinant plasmid was sequenced and introduced into E. coli BL21 (DE3) for expressing. E. coli BL21 (DE3)/pET-28a-NDM-1 was induced by 0.5 mmol/L IPTG, 20 °C, to cultivate 16 h. After induction, bacterial cells were collected and sonicated. The system was purified by using His Trap Q 5-mL column (GE healthcare, Boston, USA) as a purification medium, and the buffer was 250 mmol/L imidazole-containing PBS solution (pH 7.5) to obtain a higher purity recombinant protein solution. At last purified NDM-1 was stored in the buffer (150 mmol/L NaCl, 2 mmol/L ZnCl2, 20 mmol/L Tris, pH 7.5) and frozen at −80 °C until further use.

Ki measurement method

A 200 μL NDM-1 enzyme reaction system (2 μL inhibitor, 98 μL NDM-1 enzyme and 100 μL MEM) was selected with final concentrations of inhibitors of 0, 2.5, and 5.0 mg/mL. The NDM-1 enzyme was diluted with 10 mmol/L HEPES, the inhibitor was diluted with water to the set concentration gradient, and the sample was incubated at 30 °C for 15 min. For different concentrations of compound, the substrate concentration was varied and the final concentrations of substrate MEM were 8.25, 6.25, 12.5, 25, 50, 100, 200 and 400 mmol/L. After measuring the absorbance in the OD300 using a microplate reader (318MC, Boteng Instrument Equipment Co., Ltd., Shenyang, China), the measurement was continued for 60 min and measured every 1 min. Finally, the reaction rates of each compound concentration were calculated separately, and the Lineweaver-Burke curve at different compound concentrations, the double reciprocal curve of the reaction rate and the substrate concentration were calculated. The Ki value was calculated using GraphPad prism software to analyze the type of inhibition of the NDM-1 enzyme by the compound.

Time-dependent killing assay

The strains (A total of 10 NDM-1 positive non-duplicate K. pneumoniae isolates, which were obtained from a teaching hospital of Zhengzhou University, Zhengzhou, China) were grown overnight (incubated at 37 °C with aeration at 225 rpm) with Shaker (ZWY-1102C, Zhicheng Analytical Instrument Manufacturing Co., Ltd., Shanghai, China). On the next day, the bacterial solution was diluted 1:10,000 with MHB medium, and placed on a constant temperature shaker at 37 °C for 225 rpm to continue growth for 2 h, after which different concentrations of MIC compounds and meropenem were added. At this time, it was 0 h, 100 μL of the sample solution was taken out from the EP tube, centrifuged at 10,000×g in a centrifuge, the supernatant was discarded, and a sterile 96-well plate was taken and diluted with 1 × PBS buffer. Diluted 10 times in sequence, and took 10 μL of the sample droplet from the diluted well and added it to the MHA solid medium to mark it. This was the colony count at 0 h. At this time, according to the above sampling method, the colony counts were performed for 1, 2, 3, to 24 h, and the time sterilization curve was obtained with Origin version 8.0 (OriginLab, Northampton, MA, USA). The bacterial colonies were counted and results represented in log10 (CFU/mL) using the origin 8.0. And photos the bacteria control for 24 h.

In vivo sepsis animal experiment

Animals experiment was carried out according to the approved guidelines of the Institutional Animal Care and Use Committee. Female BALB/c nude mice weighing 15–18 g and aged 7–9 weeks were purchased from Henan Hua Xing of Laboratory Animal Co., Ltd. (HXDW20010004, Zhengzhou, China). Drug safety evaluation experiment was to examine the safety of the compound by tissue staining. Experimental mice were randomized to cages of two per group for this experiment. The mice were divided into three groups: NS (normal saline, subcutaneous injection), 19bg (64 mg/kg, subcutaneous injection) and compound 19bh (64 mg/kg, subcutaneous injection). Twenty-four hours after injection, all of the mice were euthanized, and these mice were dissected, and liver, spleen and kidney tissues were sectioned and H&E stained. We established murine sepsis models to verify that compounds 19bg and 19bh could reverse the resistance of clinical strains of NDM-1 enzyme to MEM. Mice were randomized to cages of 6 per group for this experiment. For all organ bacterial load experiments, BALB/c mice were given a median lethal dose (5 × 106 CFU) of K. pneumoniae by i.p. injection and after 30 min post-infection treated with a single dose of MEM (10 mg/kg) and combination of MEM (10 mg/kg) plus compound 19bg (10 mg/kg), compound 19bh (10 mg/kg) or PBS by subcutaneous injection. Mice were euthanized 48 h post-infection and treatment, and spleen, liver and kidney were collected. Organs were placed into one mL sterile PBS on ice, and then homogenized. Organ homogenates were then serially diluted in PBS, and selected plating for CFU enumeration. At the same time, H&E staining was used for observing whether liver, spleen and kidney tissue lesions. For survival experiments, BALB/c mice were given a lethal dose of K. pneumoniae (1 × 107 CFU) by i.p. injection and after 30 min post-infection treated with a single dose of MEM (10 mg/kg), compound 19bg alone (30 mg/kg), compound 19bh alone (30 mg/kg) and combination of MEM (10 mg/kg) plus compound 19bg (30 mg/kg) or compound 19bh (30 mg/kg), or PBS by subcutaneous injection. Mice were monitored for endpoint until 96 h post-infection and treatment.

Molecular docking assay

To predict the binding mode of uncompetitive inhibitors with NDM-1, we used CavityPlus (http://www.pkumdl.cn/cavityplus) for molecular docking studies. The X-ray structure of NDM-1 with a hydrolyzed meropenem ligand (PDB ID: 5N0H) was selected for docking. Eleven binding sites were identified by using the Cavity Module with default parameters. The location, max pKd and average pKd value of every binding sites were shown in Supporting Information Fig. S5 and Table S4). Then, cavity one was selected as ortho-site (hydrolyzed meropenem ligand binds at cavity 1), CorrSite Module with default parameters of CavityPlus was used to identify potential allosteric binding site. The results outputted 4 allosteric binding sites with a Z-score larger than 0.5 (Supporting Information Table S5).

Compound 19bh was docked into allosteric binding sites (cavities 4, 2, 7 and 10) respectively using Glide Docking Module of Schrödinger (Version 6.7, Schrodinger, New York, NY, USA) and without hydrolyzed meropenem ligand in the substrate binding site (cavity 1). The docking scores were shown in Supporting Information Table S6. When 19bh was docked to cavity 7, it formed two H-bond interactions with D95 and Q123. Because H122 and D124 were critical residues for NDM-1-catalyzed carbapenem hydrolysis, we speculated that compound 19bh was most likely to bind in cavity 7 which was adjacent to the substrate binding site. When we docked 19bh into cavity 7 with hydrolyzed meropenem ligand in the substrate binding site (cavity 1), there was little change in values. The docking results showed that 19bh formed two H-bond interactions with D95 and Q123, just like the docking results when hydrolyzed meropenem ligand in the substrate binding site. In addition, compound 19bg showed a similar binding pattern to compound 19bh.