1. Origin of indologenes #3362
C. indologenes #3362 was isolated in 2019 from a sink swab sample from a residential aged care facility as part of an antimicrobial surveillance project. The genome size of C. indologenes was 5105493 bp with 37.28% GC content (Accession number: JAPSGE000000000). Preliminary antimicrobial susceptibility assays indicated high-level intrinsic resistance to ceftazidime, imipenem, meropenem and colistin (Fig. 1a). C. indologenes is an opportunistic pathogen that is intrinsically resistant against the last resort antibiotics carbapenems and colistin, hence the presence of this pathogen in close proximity of a vulnerable population warranted a closer investigation into its resistance profile and genomic content.
- Bioinformatic analysis of indologenes isolate #3362 revealed a putative subclass B1 β-lactamase
Whole genome sequencing was performed on a resistant C. indologenes isolate #3362. Sequencing analysis revealed three β-lactamases including an Ambler Class A Extended-Spectrum β-lactamase, blaCIA, and two metallo-β-lactamases, blaIND-2 and a putative metallo β lactamase (MBL) gene. The blaCIA and blaIND-2 genes code for two predominant β-lactamase genes in C. indologenes and co-harbouring of both has been reported worldwide (Matsumoto et al., 2012; Yeh, Li, Huang, & Liu, 2022; Zhang et al., 2021). The putative MBL is composed of 251 amino acids and has not been characterized before hence was named CIM-1 for C. indologenes metallo-β-lactamase 1 in this study.
CIM-1 was further studied by an amino acid sequence alignment with the most widely disseminated MBL, the New-Dehli metallo β-lactamase (NDM-1). CIM-1 and NDM-1 revealed low sequence identity (33.5%), however, some level of similarity was observed in the region of loops 3 and 10, which have been identified as important for substrate binding (Andersson, Jarvoll, Yang, Yang, & Erdelyi, 2020; Palica et al., 2022) (Fig. 1b). Assessment of known conserved regions found within MBLs revealed that CIM-1 has all the highly conserved residues in the zinc binding site of class B1 MBLs (Fig. 1b).
The primary amino acid sequence of CIM-1 was used to generate a 3D- CIM-1 model using the SWISS-MODEL server (Fig. 1c), with results identifying a MBL carbapenemase, MYO-1 (PDB: 6T5L) as the top template. MYO-1 is a plasmid-encoded subclass B1 MBL isolated from Myroides odoratimimus. MYO-1 and CIM-1 were predicted to share 49 % amino acid sequence identity (Supplementary Fig. 1). The model generated using this template was shown to be of high quality and high reliability, displaying a high GMQE (global model quality estimate)(Waterhouse et al., 2018) score of 0.81 and high QMEANDisCo (distance constraints applied on model quality estimation)(Studer et al., 2020) global score of 0.84. The active site of CIM-1 hosts two Zn (II) ions in a shallow grove in between loop 3 and 10 (Fig. 1c). The confidence level of the model at loop L3 region is relatively low despite the high amino acid sequence similarity (Fig. 1c). This may be due to the flexibility of this loop as it is involved in substrate binding. The highly conserved Zn1 coordinators (His116, His118, and His196) and Zn2 site coordinators (Asp120, Cys221, and His263) among subclass B1 MBLs are presented in the CIM-1 amino acid sequence as denoted in Fig. 1b and presented in Fig. 1d.
- CIM-1 is a lipidated MBL with non-canonical lipobox
As MBLs are typically translocated into the periplasmic space the CIM-1 signal peptide and translocation pathway were assessed by SignalP 6.0 (Teufel et al., 2022). CIM-1 was predicted to be a lipoprotein transported by the Sec-translocation pathway and cleaved by signal peptidase II at a high likelihood (0.9909). The cleavage site of the signal peptide was predicted between asparagine 18 and the cysteine 19. Interestingly, assessment of the lipobox revealed that CIM-1 has a non-canonical lipobox in comparison to the canonical lipobox exemplified by NDM-1 (Fig. 2a). The only conserved amino acid within the CIM-1 sequence is cysteine (C19), which is essential in the attachment of lipid moiety during lipoprotein biosynthesis (Kovacs-Simon, Titball, & Michell, 2011).
To verify membrane tethering and the classification of CIM-1 as a lipoprotein, cellular fractionation was carried out with CIM-1 cloned in a pET41a(+) plasmid and transformed into E. coli C41(DE3) cells. Results revealed that CIM-1 was mainly observed in the membrane fraction indicating that it is a membrane associated protein (Fig. 2b). To investigate the role of the highly conserved cysteine (C19) in lipidation, a CIM-1 C19A mutant was generated, and its cellular localisation investigated. A comparable amount of CIM-1 C19A was observed in both soluble and membrane fractions (Fig. 2c).
Further solubilization of the membrane fraction revealed that CIM-1 was not removed by the addition of 1 M KCl which are designed to remove loosely associated peripheral proteins nor 0.1 M Na2CO3, which remove peripheral proteins associated with membrane through strong electrostatic interactions. However, the majority of CIM-1 could only be solubilized by the addition of Triton X-100. This result suggests that CIM-1 is hydrophobically tethered to the membrane. The β-lactamase activity in each cellular fraction was assessed by its ability to hydrolyse the chromogenic cephalosporin antibiotic, nitrocefin. A comparison between the end-point wavelength changes correlates with the immunoblotting result (Fig. 2b-c). Notably, a similar amount of CIM-1 protein was observed in both the Triton X-100 solubilized fraction and the remaining membranes, while the β-lactamase activity of the former is significantly higher than the later. This indicates that the protein in the last fraction was not fully functional or may have been misfolded. On the other hand, the majority of CIM-1 C19A was removed by 0.1 M Na2CO3 indicating that this mutant is associated with the membrane through electrostatic interactions.
- N-terminal sequences alters protein localisation, expression level, and resistance profiles of MBL in coli
The activity of CIM-1 as a subclass B1 MBL was validated by investigating its potential to confer carbapenem and cephalosporin resistance on antibiotic-sensitive E. coli C41(DE3) (Table 1). For a comparison, the periplasmic MBL, IND-2 from C. indologenes #3362 and NDM-4 from Klebsiella pneumoniae D53 were also cloned and expressed in E. coli C41(DE3). Analysis of resistance conferred by CIM-1 in E. coli C41(DE3) cells revealed that CIM-1 was able to confer resistance upon E. coli to all the β-lactam antibiotics tested albeit at lower levels (lower MIC values) compared to IND-2 and NDM-4.
To further investigate the localisation of CIM-1 and the effect of the N-terminal amino acid sequence on the protein localisation, signal peptide constructs were made. I-CIM and C-IND were constructed to assess if the exchange of signal peptides and the first three amino acids of the mature protein will be able to alter protein localisation (Fig. 3a). As NDM is the only characterized lipidated MBL to date, N-CIM and C-NDM were constructed to assess whether exchange of signal peptides between two lipidated proteins will have any effect on protein localisation and their resistance profiles. Finally, the mature protein of CIM-1 with the signal peptide removed (DSig) was constructed to generate a soluble version of CIM-1. As shown in Table 1, all the constructs are viable and able to confer resistance to E. coli.
Table 1. Minimum inhibitory concentrations (MIC) for E. coli C41(DE3) expressing CIM-1, IND-2, NDM-4, the cysteine mutant CIM-1 C19A, the chimeric constructs I-CIM, C-IND, C-NDM, N-CIM, and the construct lacking the signal peptide CIM-1ΔSig sub-cloned into the pET41a(+) vector. E. coli C41(DE3) propagating the pET41a(+) vector was used as a control. C. indologenes #3362 carrying both IND-2 and CIM-1 was used as a comparator.
β-lactams
|
|
MIC (μg/mL) for E. coli C41(DE3)
|
|
|
C. indologenes #3362
|
pET41a(+)
|
CIM-1
|
IND-2
|
NDM
|
CIM-1 C19A
|
I-CIM
|
C-IND
|
N-CIM
|
C-NDM
|
CIM-1ΔSig
|
Ampicillin
|
512
|
4
|
8
|
>128
|
>128
|
8
|
>128
|
64
|
64
|
256
|
>128
|
Cefazolin (1st)
|
256
|
2
|
4
|
64
|
>128
|
4
|
128
|
2
|
64
|
64
|
128
|
Cefoxitin (2nd)
|
32
|
0.5
|
1
|
32
|
>128
|
1
|
32
|
1
|
16
|
16
|
>128
|
Ceftazidime (3rd)
|
16
|
0.125
|
0.25
|
0.25
|
>128
|
0.25
|
64
|
0.125
|
64
|
32
|
>128
|
Cefepime (4th)
|
2
|
0.031
|
0.031
|
0.031
|
64
|
0.031
|
16
|
0.031
|
8
|
0.5
|
>128
|
Doripenem
|
128
|
0.063
|
0.25
|
>128
|
>128
|
0.5
|
128
|
0.25
|
16
|
8
|
>128
|
Ertapenem
|
64
|
0.016
|
0.063
|
128
|
64
|
0.063
|
16
|
0.016
|
4
|
2
|
>128
|
Imipenem
|
128
|
0.5
|
2
|
>128
|
>128
|
2
|
128
|
2
|
16
|
8
|
>128
|
Meropenem
|
128
|
0.031
|
0.125
|
>128
|
>128
|
0.25
|
128
|
0.063
|
16
|
8
|
>128
|
Aztreonam
|
>512
|
<0.02
|
<0.02
|
<0.02
|
<0.02
|
<0.02
|
<0.02
|
<0.02
|
<0.02
|
<0.02
|
<0.02
|
*The generation of cephalosporins is indicated in brackets for each cephalosporin
z fractions while IND-2 was only identified in soluble fractions (Fig. 3d). As mentioned above, a comparable amount of CIM-1 C19A was identified in both soluble and membrane fractions due to an electrostatic interaction with membranes. Results also revealed that replacing the periplasmic signal peptide with a lipoprotein signal peptide leads to protein relocation as C-IND was found to be localised to the membrane fraction. Interestingly, replacing the lipoprotein signal peptide with a periplasmic signal peptide did not fully alter the protein localisation as I-CIM was identified in both membrane and soluble fractions.
Assessment of the MIC results revealed that resistance profiles among constructs differs significantly (Table 1), however, it is unclear if expressions levels may be a contributing factor in results obtained (Fig. 3d). The replacement of the highly conserved cysteine to alanine did not fully alter CIM-1 localisation. This protein construct was still lowly expressed (Fig. 3c). We postulated that the non-canonical signal peptide influences the expression level of protein as both CIM-1 and C-IND as low expression levels were observed for both proteins. In addition, E. coli expressing C-IND had a drastic decrease in the MIC values when compared to E. coli expressing IND-2. Removal or exchange of the CIM-1 non-canonical signal peptide (CIMDSig, I-CIM and N-CIM) resulted in increased expression levels and vastly improved antimicrobial activity (Fig. 3b and Table 1). The general cephalosporin resistance trend, as indicated by the MIC assay, revealed that all constructs had reduced β-lactamase activity against the later generation cephalosporins (Table 1). As expected, none of the constructs conferred resistance to the monocyclic β-lactam (aztreonam). The unusually high resistance profile observed for CIM-1ΔSig expressing E coli cells was unexpected. This construct should be in the cytoplasm as its signal peptide, essential for translocation to the periplasm, was removed and the site of action of β-lactam antibiotics is in the periplasm. We postulate that this high resistance was due to the cytoplasmic CIM-1ΔSig not being able to reach the periplasmic space where the β-lactam attacks, leading to cell lysis and the release of cell content. High expression levels of CIM-1ΔSig (Supplementary Fig. 2) and its presence within the medium may thus be deactivating antibiotics in the broth allowing the remaining cells to grow in this antibiotic free environment. This result suggests that CIM-1 is a very active enzyme against all tested bicyclic β-lactams.
- CIM-1 has a high affinity for cephalosporins and carbapenems
Kinetic assays were performed on purified proteins to compare the biochemical properties of the three enzymes, including their abilities to hydrolyse various classes of b-lactam antibiotics. Hydrolysis in the presence of the MBL inhibitor 1,10-phenanthroline (zinc chelating agent) was included as a control for all the antibiotics tested. 1,10-phenanthroline was able to inhibit 70% of the hydrolysis activity (Supplementary Fig. 3).
All three enzymes (IND-2ΔSig, CIM-1ΔSig, and NDM-4ΔSig) showed catalytic activity against the substrates tested in this study with the exception of IND-2, which was unable to hydrolyse cefepime (Table 2). The two MBLs carried by C. indologenes have very different kinetic properties. CIM-1ΔSig was broadly active against all seven β-lactam substrates in the study, showing the highest affinity for nitrocefin, ertapenem and imipenem with Km values less than 100 μM, and relatively high turnover rates against ceftazidime. IND-2ΔSig had high Km values (low affinity) against all tested substrates apart from nitrocefin for which it had a Km at 115 μM. These in vitro results suggest CIM-1 is likely to be an effective and clinically important broad-spectrum MBL. Comparing CIM-1 with NDM-4, which is the most disseminated and clinically relevant MBL, similar low Km values were observed, especially against carbapenems. However, compared to IND-2, a lower turnover rate was observed for CIM-1 that was comparable with the turnover rates of NDM-4. In addition, CIM-1 was observed to have a lower affinity for cephalosporin than NDM-4.
Table 2. Kinetic parameters of recombinantly expressed and purified CIM-1ΔSig, IND-2ΔSig, and NDM-4ΔSig. Data shown are the results generated using Prism GraphPad 8.0 from three replicates. Errors are reported as standard errors.
β-lactams Substrates
|
Km (μM)
|
kcat (s−1)
|
kcat/Km (s−1 M-1)
|
IND-2ΔSig
|
CIM-1ΔSig
|
NDM-4ΔSig
|
IND-2ΔSig
|
CIM-1ΔSig
|
NDM-4ΔSig
|
IND-2ΔSig
|
CIM-1ΔSig
|
NDM-4ΔSig
|
Nitrocefin
|
115±9.6
|
33±8
|
11±1.8
|
78
|
2.8
|
3.4
|
6.8 x 105
|
8.5 x 104
|
3.1 x 105
|
Ceftazidime
|
>1000
|
324±42
|
93±14
|
3.4
|
30
|
6.3
|
ND
|
9.3 x 104
|
6.8 x 104
|
Cefepime
|
NH
|
538±56
|
136±16
|
-
|
15.3
|
5.4
|
-
|
2.8 x 104
|
4.0 x 104
|
Doripenem
|
>1000
|
175±20
|
253±19
|
ND
|
21
|
51
|
ND
|
1.2 x 105
|
2.0 x 105
|
Ertapenem
|
1069±105
|
91±11
|
159±22
|
127
|
9.2
|
17
|
1.2 x 105
|
1.0 x 105
|
1.1 x 105
|
Imipenem
|
728±102
|
94±7
|
151±22
|
208
|
8.5
|
31
|
2.9 x 105
|
9.0 x 104
|
2.1 x 105
|
Meropenem
|
>1000
|
143±19
|
163±22
|
ND
|
15
|
25
|
ND
|
1.0 x 105
|
1.5 x 105
|
Aztreonam
|
NH
|
NH
|
NH
|
-
|
-
|
-
|
-
|
-
|
-
|
*NH, no hydrolysis was detected; ND, not determined
- The non-canonical lipobox is highly conserved in environmental bacteria
To investigate whether CIM-1 is carried by bacteria other than C. indologenes, a BlastP search was performed with the amino acid sequence of CIM-1. Homologs of CIM-1 were found distributed widely cross various species of mainly environmental bacteria (Fig. 4). The sequence identity of between top 100 homologs and CIM-1 is between 100% (Accession number: WP_123865086.1) and 54% (Accession number: WP_035589998.1). A phylogenetic tree comprised of the top 100 CIM-1 homologs revealed that there are three major clusters (Fig. 4). Of these, clade 1 contained the smallest group of four proteins from Hymenobacter spp., Gram-negative bacteria found in soil and water (Lee et al., 2017; Srinivasan, Joo, Lee, & Kim, 2015). Clade 2 contained a larger number of proteins from a diverse group of bacterial species. Assessment of the lipoboxes revealed that the two homologs from Flavobacterium jejuense and one from Avrilella dinanensis were predicated to be lipoproteins with a high likelihood that fulfills the conserved lipobox sequence ([LVI][ASTVI][GAS]C), while the rest of the predicted lipoproteins did not follow the conserved sequence. The second clade contains a broad variety of environmental bacteria, mainly found in water and soil. Sequence alignments show that two protein homologs from Pedobacter (Accession number: RZK54418.1 and RZL36466.1) and two from Elizabethkingia anopheles (Accession number: WP_035589998.1 and MBG0505245.1) carry a highly conserved mature protein sequence, but a low similarity in signal peptide sequence (Supplementary Figure 4). The third clade contains mostly the Chryseobacterium spp. homologs with most predicted to be lipoproteins with a variety of lipobox sequences. We also observed that lipobox sequences are well-conserved within bacterial species.