Deciphering the Role of V88L Substitution in NDM-24 Metallo- β -Lactamase

: The New Delhi metallo- β -lactamase-1 (NDM-1) is a typical carbapenemase and plays a crucial role in antibiotic-resistance bacterial infection. Phylogenetic analysis, performed on known NDM-variants, classiﬁed NDM enzymes in seven clusters. Three of them include a major number of NDM-variants. In this study, we evaluated the role of the V88L substitution in NDM-24 by kinetical and structural analysis. Functional results showed that V88L did not signiﬁcantly increase the resistance level in the NDM-24 transformant toward penicillins, cephalosporins, meropenem, and imipenem. Concerning ertapenem, E. coli DH5 α / NDM-24 showed a MIC value 4-fold higher than that of E. coli DH5 α / NDM-1. The determination of the k cat , K m , and k cat / K m values for NDM-24, compared with NDM-1 and NDM-5, demonstrated an increase of the substrate hydrolysis compared to all the β -lactams tested, except penicillins. The thermostability testing revealed that V88L generated a destabilized e ﬀ ect on NDM-24. The V88L substitution occurred in the β -strand and low β -sheet content in the secondary structure, as evidenced by the CD analysis data. In conclusion, the V88L substitution increases the enzyme activity and decreases the protein stability. This study characterizes the role of the V88L substitution in NDM-24 and provides insight about the NDM variants evolution.


Introduction
Metallo-β-lactamases (MBLs) are a group of enzymes that confer high resistance to most β-lactams. The active site of these enzymes contains one or two zinc ions, that are crucial for catalytic mechanism [1]. Based on their amino acid sequences, MBLs have been divided into subclasses B1, B2, and B3 [2]. Among subclass B1, the New Delhi metallo-β-lactamase (NDM-1) is one of the most widespread carbapenemase. NDM-1 was first identified in 2008 in a clinical strain of Klebsiella pneumoniae [3]. NDM-1 producing bacteria can hydrolyse all β-lactams (except monobactams), including carbapenems, the "last resort" antibiotics used in clinical therapy. NDM-1 genes are located on plasmids that mediate their dissemination across different bacterial strains [4,5]. However, the clinical success of NDM is also due to the fact that it is a lipoprotein anchored to the outer membrane, resulting in an unusual stability of NDM-1 and enabling secretion, in Gram-negative bacteria [6][7][8].

Phylogenetic analysis
A phylogenetic analysis of NDM-1 variants was performed in order to classify these enzymes based on their amino acid similarities. Overall, the NDM variants were classified into three major clusters (NDM-1, NDM-4, and NDM-24), two minor clusters (NDM-3 and NDM-6), and two divergent proteins (NDM-14 and NDM-10). As shown in Figure 1, the NDM variants are well categorized. The NDM-1 cluster includes eight variants that showed only one amino acid replacement, except for NDM-18 where an insertion of five amino acids have been found (position 48-52). In the NDM-4 group, all variants possess the replacement at position 154. In particular, except for NDM-11 containing the M154V substitution, all variants shared M154L. In the NDM-24 group, Valine at position 88 has been replaced by a Leucine (V88L). Concerning the two minor groups, similar characteristics were observed with the D95N and A233V substitution for the NDM-3 and NDM-6 clusters, respectively.

Functional Study
The NDM-24 variant was obtained by a site-directed mutagenesis by using the NDM-1 as template. All genes (bla NDM-1 , bla NDM-5 , and bla NDM-24 ) were cloned into pHSG398, which were controlled by the same promoter and thus the same expression. The E. coli DH5α recombinant strains were used to test susceptibility to a wide panel of β-lactams. As shown in Table 1, the results of the susceptibility test revealed that NDM-1, NDM-5, and NDM-24 confer resistance to most β-lactams with similar MIC values, suggesting that the NDM enzymes were successfully expressed in the host cells. A different behaviour was observed for carbapenems, for which the MIC values were markedly lower than those of penicillins and cephalosporins with the exception of cefepime, as previously reported [15][16][17][18][19][20]. Concerning ertapenem, NDM-24 and NDM-5 showed an increase of the MIC values of the 4-and 8-fold with respect to NDM-1. Based on the data obtained, the V88L substitution enhances resistance toward ertapenem.

Characteristics of Enzyme Activity
In order to obtain soluble and active enzymes, the recombinant plasmids were successfully expressed in the E. coli BL21 (DE3) cells as described in the methods section. After purification, the enzymes were checked on SDS-PAGE to confirm the solubility and purity (>90%) ( Figure S1). The MALDI-TOF mass spectrometry was used to confirm the molecular mass of the three enzymes, which corresponds to 24884,024 Da ( Figure S2). To investigate the enzyme activity, the kinetic parameters for NDM-1, NDM-5, and NDM-24 were determined ( Table 2).
All the NDM variants of this study were able to hydrolyse all the β-lactams tested. Compared with NDM-1, NDM-24 showed lower K m values for penicillins and ceftazidime whereas for carbapenems they are quite similar. Comparing the k cat values, NDM-24 hydrolyses all β-lactams, except penicillins, better than NDM-1 and NDM-5. In particular, the k cat values of NDM-24 are 2.26-, 1.61-, 2.73-, 2.02-, 2.17-, and 1.75-fold higher than NDM-1 towards penicillin G, ceftazidime, cefepime, imipenem, meropenem, and ertapenem, respectively. This was also confirmed by a slight increase of catalytic efficiency. This result stated the important role of V88L in the substrate hydrolysis. The contribution of V88L is likely that of M154L as demonstrated by the calculation of the k cat /K m rates ( Table 2). This was possibly due to differences in the intrinsic properties, such as the enzyme stability, protein expression, and adaptability [21][22][23][24], and nutritional conditions of bacteria in vivo/vitro. Comparing the k cat /K m values of carbapenems, the carbapenemase activity of NDM-5 was similar to NDM-24, but higher than NDM-1. A recent study showed that an increase of the catalytic efficiency (k cat /K m ) for meropenem has been ascertained in NDM-5 (V88L and M154L). In NDM-4, which contains only M154L, no significant change has been observed, suggesting that V88L might play a role in enhancing the NDM enzymes activity rather than M154L. Moreover, an increase of the carbapenemase activity was also observed in the evolutionary NDM variants, such as NDM-17 (V88L, M154L, and E170K) and NDM-20 (V88L, M154L, and R270H) [10,15,16].

Thermal Stability
As previously reported, mutations in the NDM variants can affect the enzymes stability, resulting in changing the persistence lifetime in the bacterial host, and consequent antibiotics resistance [25]. For determining whether the V88L substitution influences the NDM-24 stability property, circular dichroism CD was used to assay the protein stability by recording signal changes. NDM-5 was used as reference to analyze the effect of M154L. Compared with NDM-1 and NDM-5, NDM-24 (59.41 ± 0.06 • C) possessed the lowest melting temperature ( Figure 2). Notably, the V88L destabilized effect was compensated by M154L in NDM-5 with a remarkable higher thermal temperature than NDM-24 (69.13 ± 3.6 • C compared to 59.41 ± 0.06 • C). Moreover, NDM-5 showed a higher stability than NDM-1 suggesting the destabilized role of M154L. This was in agreement with a previous document that the M154L mutation would be a turning point for the NDM variants, in which combing M154L with additional substitutions benefit for the NDM enzymes exhibiting increased thermostability [10]. In the NDM-24 group there are four variants (NDM-5, -17, -20, and -21) in which the combination of the V88L and M154L substitutions takes favorable results in terms of the stability and environmental selection.  Table 2). This was possibly due to differences in the intrinsic properties, such as the enzyme stability, protein expression, and adaptability [21][22][23][24], and nutritional conditions of bacteria in vivo/vitro. Comparing the kcat/Km values of carbapenems, the carbapenemase activity of NDM-5 was similar to NDM-24, but higher than NDM-1. A recent study showed that an increase of the catalytic efficiency (kcat/Km) for meropenem has been ascertained in NDM-5 (V88L and M154L). In NDM-4, which contains only M154L, no significant change has been observed, suggesting that V88L might play a role in enhancing the NDM enzymes activity rather than M154L. Moreover, an increase of the carbapenemase activity was also observed in the evolutionary NDM variants, such as NDM-17 (V88L, M154L, and E170K) and NDM-20 (V88L, M154L, and R270H) [10,14,15].

Thermal stability
As previously reported, mutations in the NDM variants can affect the enzymes stability, resulting in changing the persistence lifetime in the bacterial host, and consequent antibiotics resistance [25]. For determining whether the V88L substitution influences the NDM-24 stability property, circular dichroism CD was used to assay the protein stability by recording signal changes. NDM-5 was used as reference to analyze the effect of M154L. Compared with NDM-1 and NDM-5, NDM-24 (59.41 ± 0.06 °C) possessed the lowest melting temperature (Figure 2). Notably, the V88L destabilized effect was compensated by M154L in NDM-5 with a remarkable higher thermal temperature than NDM-24 (69.13 ± 3.6 °C compared to 59.41 ± 0.06 °C). Moreover, NDM-5 showed a higher stability than NDM-1 suggesting the destabilized role of M154L. This was in agreement with a previous document that the M154L mutation would be a turning point for the NDM variants, in which combing M154L with additional substitutions benefit for the NDM enzymes exhibiting increased thermostability [10]. In the NDM-24 group there are four variants  in which the combination of the V88L and M154L substitutions takes favorable results in terms of the stability and environmental selection.

Structure Analysis
Previous reports indicated that mutations in the NDM influence the α-helical, β-sheets content, and loop flexibility [26]. For example, the Q123A substitution in NDM-1 leads to a decrease of the α-helical content [27]. To know if the V88L substitution could modify the NDM-24 structure, a secondary structure was determined by the Far-UV CD spectrum. All NDM variants CD spectrum data were fitted and shown in Figure 3. The spectrum signals were superimposable at most wavelengths, and showed characteristics of αβ/βα fold, a typical and conservative protein structure in MBLs [28]. The presence of positive bands at 192 nm and two negative peaks at 208 nm, a minimum peak, and 220 nm, suggesting the dominance of the β-sheets and α-helical content. The major differences were observed in the nearby 192 nm, symbolizing α-helical peak, and 208-220 nm, a α-helical and β-sheets bonds. Overall, the α-helical content was found ranging between 13%-20% in NDM-1, NDM-5, and NDM-24 (Table 3), in agreement with previous reports and the content of the β-sheet was high around 30% [27]. Compared to NDM-1, NDM-24 possesses a higher α-helical content and lower β-sheet content, suggesting that V88L was responsible for the secondary structure content changes of NDM-24. Furthermore, the secondary predicted result ( Figure S3) confirmed that the V88L substitution occurred in the β-strand terminal, which may be prematurely terminated, leading to a decrease in the β-sheet content. Kumar et al. claimed that 152A, located in the β-strand, drastically influenced the NDM-5 Catalysts 2019, 9, 744 6 of 11 activity and protein thermolability, by reducing the β-sheet content [26]. Our analysis demonstrates that the emergence of M154L (in NDM-5) caused the α-helical to decrease and the β-sheet to increase relative to NDM-24, while the α-helical and β-sheet content of NDM-5 were between NDM-1 and NDM-24. In addition, the 3D model of NDM-24 ( Figure 4) was generated by using NDM-1 (PDB accession: 5N0H, 4RBS, 4HKY, and 4EYL) and NDM-5 (PDB accession: 6MGY, and 4TZE) as a template. Although the residue 88L is away from the active site groove and far to the active loops (Loop 3 and Loop 10), differences in the structure content, stability, and enzyme activity were ascertained. Several studies confirmed that non-activity sites substitution can influence the NDM catalytic efficiency [29], and our results about the V88L substitution support this theory.     To acquire a credible model, the 6 NDM crystal structure (NDM-1(5N0H, 4RBS, 4HKY, and 4EYL) and NDM-5(6MGY, and 4TZE)) were adopted. The residue 88L and active pocket were labelled.

Site-directed mutagenesis, cloning and expression of NDM variants
The blaNDM-1 and blaNDM-5 encoding genes were obtained from clinical Escherichia coli strains as previously described [14,15]. Site-directed mutagenesis was performed to generate blaNDM-24 using the pHSG398/NDM-1 plasmid as template and primers listed in Table S1, as previously described [30].
First, the blaNDM-genes were cloned into a pHSG398 vector (TaKaRa Bio, Dalian, China) using the BamHI and XhoI restriction sites. In a second PCR experiment, the blaNDM variants were amplified without a signal peptide introducing the Tobacco Etch Virus (TEV) at the N-terminal sequence. In order to obtain enzymes overexpression, the amplicons were inserted into a pET-28a(+) plasmid. The E. coli DH5α competent cells were used as a non-expression host. E. coli BL21(DE3) was used for enzymes overexpression. The authenticity of recombinant plasmids was verified by PCR and sequencing was with Sanger. To acquire a credible model, the 6 NDM crystal structure (NDM-1(5N0H, 4RBS, 4HKY, and 4EYL) and NDM-5(6MGY, and 4TZE)) were adopted. The residue 88L and active pocket were labelled.

Site-Directed Mutagenesis, Cloning and Expression of NDM Variants
The bla NDM-1 and bla NDM-5 encoding genes were obtained from clinical Escherichia coli strains as previously described [15,16]. Site-directed mutagenesis was performed to generate bla NDM-24 using the pHSG398/NDM-1 plasmid as template and primers listed in Table S1, as previously described [30].
First, the bla NDM -genes were cloned into a pHSG398 vector (TaKaRa Bio, Dalian, China) using the BamHI and XhoI restriction sites. In a second PCR experiment, the bla NDM variants were amplified without a signal peptide introducing the Tobacco Etch Virus (TEV) at the N-terminal sequence. In order to obtain enzymes overexpression, the amplicons were inserted into a pET-28a(+) plasmid. The E. coli DH5α competent cells were used as a non-expression host. E. coli BL21(DE3) was used for enzymes overexpression. The authenticity of recombinant plasmids was verified by PCR and sequencing was with Sanger.

Antimicrobial Susceptibility Tests
The phenotypic profile has been characterized by a microdilution method using a bacterial inoculum of 5 x 10 5 CFU/ml according to the Clinical Laboratory and Standards Institute [31,32]. E. coli ATCC25922 was used as a negative control.

Production and Purification of NDM-1, NDM-5, and NDM-24
NDM-1, NDM-5, and NDM-24 were extracted from 0.5 L of culture of E. coli BL21(DE3)/pETNDM-1, E. coli BL21 (DE3) /pETNDM-5, and E. coli BL21 (DE3)/pETNDM-24, respectively. The cultures were grown at 37 • C to achieve an A 600 of 0.5 L, and 0.4 mM of IPTG was added. After addition of the IPTG, the cultures were incubated at 20 • C for 16 h, under aerobic conditions. Thereafter, a cell supernatant containing the soluble NDM protein was obtained from the lytic bacteria by centrifuging at 10 4 rpm. Proteins purification followed the manufacturer's instructions (Qiagen, Hilden, Germany) by using the Ni-nitrilotriacetic acid (NTA) agarose. The turbo tobacco etch virus (TEV) protease (Accelagen, San Diego, CA, USA) was used to gain the untagged protein without the His tags. The SDS-PAGE was performed to estimate the NDM purity enzymes. The protein concentration was determined using a BCA Protein Quantification Kit (Vazyme Biotech Co., Ltd, Nanjing, China). The β-lactamase activity was monitored at each purification step using the colour change of nitrocefin 1 mg/mL, a chromogenic cephalosporin, according to the previous report [20].

Determination of Kinetic Parameters
Steady-state kinetic experiments were performed following the hydrolysis of each substrate at 25 • C in a 50 mM phosphate buffer, pH 7.0 in the presence of 20 µM Zn SO 4 . The data were collected with a SpectraMax M5 multi-detection microplate reader (Molecular Devices, Sunnyvale, CA, USA) as previously described [33]. Kinetic parameters were determined under initial-rate conditions using the GraphPad Prism ® version 5.01 (San Diego, CA, USA) to generate the Michaelis-Menten curves, or by analysing the complete hydrolysis time courses [34,35]. Each kinetic value is the mean of the results of three different measurements. The error was below 5%. NDM-5 was used as a reference to normalize.

Circular Dichroism and Structure Analysis
The circular dichroism (CD) spectra (185 to 260 nm) were determined with a Chirascan Plus CD spectrophotometer (Applied Photophysics Ltd, Leatherhead, UK) equipped with a Peltier temperature-controlled cell holder, at 25 • C using a 1-mm pathlength cuvette and the Savitzky-Golay filter was explored to the baseline-correct spectra data. Protein concentrations were diluted to 0.05-0.2 mg/mL with a 5 mM sodium phosphate buffer pH 7.0 [36]. 207 nm spectrum data was used as the baseline to normalize and calibrate data for eliminating minor errors due to different concentrations [37]. The analysis was performed using the CONTINLL and SELCON3 algorithms with reference data sets three and nine, respectively [38]. The super-secondary (tertiary) structures of the proteins were analysed by the CDPro software package, which is available at the CDPro website: https://sites.bmb.colostate.edu/sreeram/CDPro/ [38,39]. To assay the location of the V88L substitution and analyse its effect on the structure, the pharmacophore modeling and screening software program Discovery Studio (version 2018) was employed to generate a three-dimensional (3D) interconnected model of NDM-24 using NDM-1 and NDM-5 as a template, in which reliable data of the crystal structure were collected from the PDB database.

Thermal Stability Testing
The melting temperature (T m ) was used to show the protein stability. The determination of T m was performed by recording the CD signal change at 222 nm. Data were collected at a ramp of 1 • C /min with a temperature range from 20-90 • C. The two-state model using nonlinear regression (Boltzmann) in the OriginPro 9.1.64 (OriginLab, Northampton, MA, USA) was applied to analyse the data. When 50% of the protein melts, the temperature is defined as the Tm, representing thermal stability.

Conclusions
Our study explored the NDM-24 biological function and probed the V88L substitution role on the structure, enzyme activity, and stability. In brief, this non-active site change enhances the enzyme activity by increasing the turnover rate, related with an indirect effect on conformation. However, the loss cost caused by V88L significantly decreased the protein stability, and would shorten the persistence lifetime in the cell, so that the resistance to antibiotics hardly exhibits an outstanding elevation for the NDM-24-producing transformants. Meanwhile, alterations in the secondary content, such as lowering the β-sheet, have an interesting role in the NDM instability, being relevant to the V88L substitution occurring in the β-strand. According to previous data, the V88L/M154L combination appears to be favorable in the NDM evolution under an environmental pressure selection [14]. Further analysis about the significance of non-active-site residues will help in the comprehension of the resistance mechanism and broaden insight in the development of inhibitors, such as potential antibiotics candidate by reducing the protein stability lifetime.
Author Contributions: J.S. designed the study. Z.L., D.L., and W.L. collected the data. Z.L., Y.W., and D.L. analyzed and interpreted the data. Z.L., A.P., D.L., Y.W., and J.S. wrote the report. All authors revised, reviewed and approved the final report.

Conflicts of Interest:
The author declares no conflict of interest.