Mechanistic Insights into Crosstalk of Tet(X) and MCR-1, Two Resistance Enzymes Co-produced by A Single Plasmid

Abstract

Tigecycline and colistin are few of last-resort defenses used in anti-infection therapies against carbapenem-resistant bacterial pathogens. The successive emergence of plasmid-borne tet(X) tigecycline resistance mechanism and mobile colistin resistance (mcr) determinant, renders them clinically ineffective, posing a risky challenge to global public health. Here, we report that co-carriage of tet(X6) and mcr-1 gives co-resistance to both classes of antibiotics by a single plasmid in E. coli. Genomic analysis suggested that transposal transfer of mcr-1 proceeds into the plasmid pMS8345A, in which a new variant tet(X6) is neighbored with Class I integron. The structure-guided mutagenesis finely revealed the genetic determinants of Tet(X6) in the context of phenotypic tigecycline resistance. The combined evidence in vitro and in vivo demonstrated its enzymatic action of Tet(X6) in the destruction of tigecycline. The presence of Tet(X6) (and/or MCR-1) robustly prevents the accumulation of reactive oxygen species (ROS) induced by tigecycline (and/or colistin). Unlike that mcr-1 exerts fitness cost in E. coli, tet(X6) does not. In the tet(X6)-positive strain that co-harbors mcr-1, tigecycline resistance is independently of colistin resistance caused by MCR-1-mediated lipid A remodeling, and vice versa. Co-production of Tet(X6) and MCR-1 gives no synergistic delayed growth of the recipient E. coli. Similar to that MCR-1 behaves in the infection model of G. mellonella, Tet(X6) renders the treatment of tigecycline ineffective. Therefore, co-transfer of such two AMR genes is of great concern in the context of “one health” comprising environmental/animal/human sectors, and heightened efforts are required to monitor its dissemination.

Author summary We report that tet(X6), a new tigecycline resistance gene, is co-carried with the other resistance gene mcr-1 by a single plasmid. Not only have we finely mapped genetic determinants of tet(X6), but also revealed its biochemical action of tigecycline destruction. Crosstalk of Tet(X6) with MCR-1 is addressed. Tet(X6) tigecycline resistance is independently of MCR-1 colistin resistance, and vice versa. Similar to MCR-1 that renders colistin clinically ineffective, Tet(X6) leads to the failure of tigecycline treatment in the infection model of G. mellonella. This study extends mechanistic understanding mechanism and interplay of Tet(X6) and MCR-1, coproduced by a single plasmid. It also heightens the need to prevent rapid and large-scaled spread of AMR.

Introduction

Antimicrobial resistance is an increasingly-devastating challenge in the context of “one health” that covers the environmental, animals and human sectors. Colistin is one of cationic antimicrobial polypeptides (CAMP) with an initial target of the surface-anchored lipid A moieties on the Gram-negative bacterium ¹. In contrast, tigecycline is the third-generation of tetracycline-type antibiotic, which interferes the machinery of protein synthesis of both Gram-negative, and Gram-positive bacteria ². In general, both colistin and tigecycline are an ultimate line of antibiotics to combat against lethal infections with carbapenem-resistant pathogens ^3,4. Unfortunately, the emergence and global distribution of MCR family of mobile colistin resistance (mcr-1 ^5,6 to mcr-10 ⁷) has potentially threatened the renewed interest of colistin in clinical therapies ⁸. The majority of transferable colistin resistance depends on the surface lipid A remodeling by the MCR enzymes via the “ping-pong” trade-off ⁹.

In addition to the two well-known actions, efflux and ribosome protection ^10,11, antibiotic degradation also constitutes in part the mechanism of tigecycline resistance ^12-14. The Tet(X) enzyme is a class of flavin-requiring monooxygenase ^15,16, which possesses the ability of modifying tetracycline and its derivatives (like the glycylcycline, tigecycline) ¹³. In general, Tet(X) inactivates tigecycline to give 11a-Hydroxytigecycline, rendering the carrier host bacterium insusceptible to tigecycline ¹³. In particular, functional meta-genomics of soils performed by Forsburg et al. ¹⁷ revealed a number of new tetracycline destructases (10 in total, namely Tet47 to Tet56). Among them, tet56 is only one exclusively from the human pathogen Legionella longbeachae, indicating its potential spread from environments to clinical sector ¹⁷. Subsequent structural studies suggested that these Tet tetracycline destructases accommodate antibiotics in diverse orientation, which highlights their architectural plasticity ¹². Indeed, the discovery of an inhibitor blocking the entry of flavin adenine dinucleotide cofactor into Tet(50) enzyme also paved a new way to reversing tigecycline resistance ¹².

Very recently, two new variants [tet(X4) and tet(X5)] of tet(X)-type determinants that encode a tigecycline-inactivating enzyme, was found to spread by distinct plasmids of Escherichia coli in China ^18,19. Although the limited distribution of tet(X4) thus far ²⁰, it constitutes an expanding family of Tet(X) resistance enzyme [Tet(X) ^21,22 to Tet(X5) ¹⁹], and raises the possibility of rendering tigecycline (and even the newly-FDA approved eravacycline, the fourth-generation of tetracycline derivatives ²³) clinically ineffective. Worrisomely, the co-transfer of mcr-1 and tet(X) probably promotes the emergence of a deadly superbug with the co-resistance to polymyxin and tigecycline. However, this requires further epidemiological evidence.

Here, we report that this is the case. It underscores an urgent need to monitor and evaluate a potential risk for the convergence of Tet(X) tigecycline resistance to MCR colistin resistance by a single highly-transmissible plasmid in an epidemic ST95 lineage of virulent E. coli ²⁴.

Results and Discussion

Discovery of tet(X6), a new variant of Tet(X) resistance enzyme

To address this hypothesis, we systematically screened the whole NCBI nucleotide database, in which each of the six known tet(X) variants [tet(X) to tet(X5)] acts as a query. Among them, it returned six hits with the significant score (97%-99% identity) when we used the tet(X3) of Pseudomonas aeruginosa (1137bp, Acc. no.: AB097942) as a searching probe. The resultant hits corresponded to four contigs of uncultured bacterium and two plasmids. These contigs include TE_6F_Contig_7 (3328bp, Acc. no.: KU547125), TG_6F_Contig_3 (3323bp, Acc. no.: KU547130), TG_7F_Contig_3 (2575bp, Acc. no.: KU547185); and TE_7F_Contig_3 (2588bp, Acc. no.: KU547176). The matched plasmids refer to pMS8345A (241,162bp; Acc. no.: CP025402) ²⁴ and p15C38-2 (150,745bp; Acc.no.: LC501585), in which the gene of MS8345_A00031 exhibits 97.1% identity and 100% coverage when compared to tet(X3) (Fig. S1). This tet(X3)-like gene, MS8345_A00031, is thereafter renamed tet(X6) (Acc. no.: BK011183) in this study (Fig. 1A). Strikingly, we found that this plasmid co-harbors the mcr-1 gene encoding a phosphoethanolamine (PEA)-lipid A transferase ²⁵. Because that the plasmid is detected in an epidemic clone of ST95 Extraintestinal Pathogenic E. coli (ExPEC) ²⁴, this single plasmid pMS8345A possesses the potential to confer its recipient host E. coli the co-resistance to both tigecycline and colistin. It is unusual, but not without any precedent. In fact, mcr-1 has ever coexist with bla_NDM in a single isolate ²⁶ and even a single plasmid ²⁷.

Figure 1 Genetic context-based analyses for co-transfer of tet(X6) and mcr-1 by a single plasmid

A. Co-linear genome alignment of the tet(X6), and mcr-1-coharboring plasmid pMS8345A with other three IncHI2-type mega plasmids (pDJB-3, pYps.F1, and pHNSHP45-2) The backbones of the four resistance plasmids are of high similarity. Except that pYps.F1 (Acc. no.: LT221036) of Yersinia pseudotuberculosis lacks mcr-1, all the other three resistance plasmids here carry this gene. pDJB-3 (Acc. no.: MK574666) is a new variant of the mcr-1-containing plasmid we reported here. Unlike that pDJB-3 lacking multidrug resistance region (MDR), the remaining three plasmids possess different MDRs. ICEs are colored as follows: oriT in brown, the relaxase-encoding gene in orange, T4SS in green, and T4CP in purple.

B. Fine dissection of the tet(X6)-positive MDR regions in pMS8345A and p15C38-2 As for this unique MDR region in pMS8345A, three Class 1 integrons are circled by dashed lines, 14 kinds of antibiotic resistance determinants like addA1 (with red arrows), are integrated, and no less than 10 different mobile elements such as ISEcp1 and TnAs1 (indicated with yellow arrows) are determined. In particular, a new tet(X) variant of tigecycline resistance enzymes, named tet(X6) (in bold red letters), is adjacent to sul1 that conferring resistance to sulfonamide. Additionally, five attL sites (highlighted in purple) are detected. The gene environment of tet(X6) on pMS8345A is highly matched to the counterpart on plasmid p15C38-2, with an exception of a nucleotide deletion of “A” on the position of 287. This A287 deletion of tet(X6) highlighted with the symbol of “*” results in a frameshift and premature termination of its corresponding ORF (1137bp vs 282bp; 378aa vs 93aa). In contrast, it gives limited overlap with the three tet(X6)-containing contigs, P. aeruginosa Pa-3 contig, TE_7F_Contig_3, and TG_7F_Contig_3.

C. Genetic context of the mcr-1-centering transposal region in pMS8345A, pHNSHP45-2, and pDJB-3 Colored arrows indicate ORFs and the shaded region depicts sequence similarity (70%-100%). The resistance genes are highlighted in red and mobile elements are highlighted in yellow.

Designations: pDJB-3 (156.5kb, IncHI2/E. coli, China/2017, Acc. no.: MK574666); pYps.F1 (200.66kb, IncH1/Y. pseudotuberculosis, France/2017, Acc. no.: LT221036); pHNSHP45-2 (251.5kb, IncHI2/E. coli, China/2013, Acc. no.: KU341381); p15C38-2 (150.745kb, IncA/C2, E. coli/Japan/2003, Acc. no.: LC501585); Pseudomonas aeruginosa strain Pa-3 contig (3.796kb, P.aeruginosa, Pakistan/2016, Acc. no.: GCA_011947185.1); Uncultured bacterium clone TE_7F_Contig_3 (Acc. no.: KU547176); Uncultured bacterium clone TG_7F_Contig_3 (Acc. no.: KU547185) and pMS8345A (241.162kb, IncHI2/E. coli, Qutar/2017, Acc. no.: CP025402).

Characterization of a tet(X6)-harboring plasmid

The tet(X6)-positive pMS8345A (Acc. no.: CP025402) that attracts us much attention during the period of in silico search, is an IncHI2-type large plasmid (∼241kb). In fact, it was initially discovered by Beatson and coworkers ²⁴ from a virulent lineage of E. coli with multiple drug resistance (MDR) through the routine screen of colistin resistance (Fig. 1A). Generally, this plasmid has an average GC% value of 46.29%, and is predicted to contain 548 putative open reading frames (ORFs). Importantly, pMS8345A is found to possess a series of Integrative and Conjugative Elements (ICEs) using the web-based tool of oriTfinder (Fig. 1A) ²⁸. Although that pMS8345A is not accessible in China right now, we applied its 2 surrogate plasmids, pHNSHP45 and pDJB-3 (Fig. 1A), in the conjugation assays. In general consistency with an observation of Zhi et al. ²⁹, the efficiency of pHNSHP45-2 transfer is calculated to be 3.7×10⁻⁵ in our experiment of conjugations. In contrast, pDJB-3 can’t survive in the conjugation trials. Unlike pDJB-3 that carries T4SS alone (Fig. 1A), pHNSHP45-2 is fulfilled with all the four essential modules for self-transmission, namely oriT region, relaxase gene, type IV coupling protein (T4CP) gene and type IV secretion system (T4SS) (Fig. 1B). Notably, the aforementioned modules are shared by the plasmid pM8345A and pHNSHP45-2 (Fig. 1B). Taken together, we believed that pM8345A is self-transmissible.

The hallmark of pMS8345A lies in two disconnected/unique resistance regions, one of which refers to the tet(X6)-bearing MDR region of appropriate ∼40kb long (Fig. 1B), and the other denotes the mcr-1-containing region (Fig. 1C). In total, 11 kinds of antimicrobial resistance (AMR) genes have been recruited and integrated into this unusual MDR region (Fig. 1B), giving multiple drug resistance. Apart from tigecycline resistance caused by tet(X6), colistin resistance arises from the “mcr-1-ISApl1” transposon alone (Fig. 1C). It is reasonable to believe that the occupation of MDR (including, but not limited to the co-resistance to tigecycline and polymyxin, two last-resort anti-infection options) by pMS8345A can be a serious risk in the clinic sector, once it successfully enters and further disseminates across pathogenic species.

Genomic analyses of tet(X6)-containing MDR region

The linear genome alignment of MDR plasmids revealed that pMS8345A displays high level of similarity to at least three other resistance plasmids (Fig. 1A). Namely, they include i) the mcr-1-harboring plasmid pHNSHP45-2 with 99.75% identity and 87% query coverage (Acc. no.: KU341381), ii) and a tetracycline resistance plasmid of Yersinia pseudotuberculosis, pYps.F1 with 100% identity and 74% query coverage (Acc. no.: LT221036), and iii) a typical IncHI2-group, mcr-1-carrying plasmid pDJB-3 with 99.74% identity and 63% query coverage (Acc. no.: MK574666). Unlike that the mcr-1-lacking plasmid pYps.F1 exists in Y. pseudotuberculosis, all the other three mcr-1-harboring plasmids disseminate in different clones of E. coli with varied sequence types (Fig. 1A). In brief, i) pMS8345A is detected in ST95, a globally-distributed clone having the relevance to bacterial bloodstream infections and neonatal meningitis ²⁴; ii) pHNSHP45-2 is recovered from an intensive pig farm ²⁹; and iii) pDJB-3 is recently determined by our group to occur in ST165 in an pig farm, a rare sequence type (Fig. 1A).

Among them, the organization of MDR differs greatly (Fig. 1). Unlike the pMS8345A having both tet(X6)-positive MDR (Fig. 1B), and the mcr-1-containing cassette (Fig. 1C), the plasmid pDJB-3 has mcr-1, but not MDR region (Fig. 1A). In contrast, the plasmid pYps.F1 carries the MDR region, but not mcr-1 (Fig. 1A). Genetic analysis elucidated that three class I integrons are located in the MDR region. Given that a pool of gene cassettes can be integrated, the majority of which encode resistance to antibiotics ³⁰, Class I integron facilitates the global spread of AMRs ³¹. In particular, the MDR region of pMS8345A seems structurally unusual (Fig. 1B). First, it comprises a cluster of transposons and insert sequences (IS) with a boundary of two integrases (one is an integrase-encoding gene, int1, on the forward strand, and the other denotes a truncated version of int1 on the reverse strand, Fig. 1B); Second, int1 is adjacent to an integron-associated recombination site attI, and then followed by several attC sites (Fig. 1B); Third, the occurrence of Tn2 and TnAs1 (two copies of Tn3 family transposons) in the pMS8345A MDR region implies an association with its mobility (Fig. 1B); Fourth, the multiple IS elements located within the MDR region (namely bla_CTX-M-1 carried on an ISEcp1, the operon of strA-strB adjacent to an IS26-IS1133 structure ³², the aac(3)-IIa/tmrB plus bla_TEM-1B genes neighbored by IS26), facilitate the formation of transposons via the recombination events (Fig. 1B); Fifth, circular gene cassettes [such as arr-2/ere(A)/aadA1] are presumably integrated by site-specific recombination between attI and attC, a process mediated by the integron integrase (Fig. 1B). Along with class I integron (Fig. 1B), the fact that the GC content (37.8%) of tet(X6) is far less than the average GC% (46.29%) of the pMS8345A, allowed us to speculate that it is probably acquired via gene horizontal transfer. Sequence alignment reveals that the tet(X6)-positive MDR region in pMS8345A is highly similar to the MDR region in p15C38-2 (Fig. 1B). In brief, tet(X6) and sul1 are downstream of a class 1 integron carrying the aadA family of resistance genes (aadA22 in pMS8345A and aadA12 in p15C38-2). p15C38-2 harbors a Tn3-like transposon, TnAs2 with 85% identity.

Functional insights into Tet(X6) tigecycline resistance

In addition to the pMS8345A plasmid, the tet(X6)-based in silico search uncovers two more tet(X6)-containing contigs (Fig. S2), namely TG_7F_Contig_3 (2575bp, Acc. no.: KU547185) and TE_7F_Contig_3 (2588bp, Acc. no.: KU547176). More intriguingly, the two contigs derive from uncultivated bacterium from latrine, in EI Salvador, 2012 (Fig. 2). It seems likely that tet(X6) appears earlier than that of tet(X4) initially detected in a contig of K. pneumoniae (4069bp, Acc. no.: NQBP01000050) from Thailand, 2015 ²⁰. Further database mining suggests a number of tet(X) homologs [designated Tet(X7) to Tet(X13)] that are similar to tet(X6) at the identity ranging 91.27% to 97.62%. To relieve the confused nomenclature, we renamed the two redundant genes tet(X5) and tet(X6) appropriately (Fig. 2), which are chromosomally encoded in certain species like Myroides ³³ and Proteus ^33-35. Different from the pMS8345A plasmid-borne tet(X6), the designation of tet(X6) from four different species [Proteus genomospecies T60 ³⁴, P. cibarius strain ZF2 ³⁵, Acinetobacter (contig) ³³ and A. johnsonii (contig) ³³] is identical to that of tet(X12) we proposed here. Accordingly, the pAB17H194-1 plasmid-encoding tet(X5) in Acinetobacter pittii strain AB17H194 (Acc. no.: CP040912) is relabeled with tet(X14). The three inconsistent tet(X6) genes that arise separately from P. cibarius [contig, Acc. no.: WURM01000016] P. mirabilis [contig, Acc. no.: WURR01000048], and M. phaeus [genome, Acc. no.: CP047050] ³³, were re-assigned with three distinct variants, namely tet(X15), tet(X16), and tet(X17). Phylogeny of these Tet(X) enzymes illustrates an ongoing Tet(X) family of resistance determinants (Fig. 2), raising a possible ancestor shared amongst these tet(X) variants.

Figure 2 Phylogeny of Tet(X6)

The protein sequences of Tet(X) variants were selected from NCBI database, and the accession numbers were labeled accordingly on the graph. Clustal Omega (https://www.ebi.ac.uk/Tools/msa/clustalo/) was applied to generate the phylogenetic tree, whose final output is built with TreeView (https://www.treeview.co.uk/). In addition to Tet(X3) (colored pink), all the twelve newly-renamed TetX variants (X6 to X17) are labeled in red. Besides the pMS8345A plasmid producing TetX6, two uncultured bacterium contigs (TE_7F_Contig_3 and TG_7F_Contig_3) from Latrine, EI Salvador, were found to carry an intact version of tet(X6). Of note, a number of tet(X) variants in duplicated and/or redundant nomenclature were corrected.

Given that i) the statement by He et al. ¹⁸ that TetX3 of Pseudomonas (Acc. no.: AB097942) is not active, is argued by our recent study ²⁰; and ii) as a new variant, Tet(X6) displays 96.03% identity to the Pseudomonas Tet(X3) (Fig. S3), integrative evidences are highly demanded for the functional assignment of Tet(X6) in the context of tigecycline resistance. Therefore, we cloned tet(X6) into an arabinose-inducible pBAD24 expression vector and test its function in the strain MG1655 of E. coli. As predicted, the presence of tet(X6) can restore the growth of its recipient strain on LB agar plates with tigecycline (16 to 32μg/ml, Fig. S4A). This level of resistance is almost as same as tet(X3) does, but slightly lower than that of Tet(X4) (Fig. S4A). As predicted, structural modeling of Tet(X6) presents a substrate-loading channel (Figs 3A-B). Similar to the scenario with Tet(X4) ²⁰, it consists of a tigecycline substrate-binding motif (Figs 3C and E) and a FAD cofactor-occupied cavity (Figs 3D and F). As for Tet(X6), the substrate-binding requires the cooperation of five residues E182, R203, H224, G226 and M365 (Table 1 and Fig. 3E). Similarly, the FAD cofactor is surrounded with the following six residues E36, R37, R107, D301, P308 and Q312 in Tet(X6) (Table 1 and Fig. 3F). Except that the substitution of H224T occurs in Tet(X1) [and/or H234Y in Tet(X5), in Fig. S3], all the aforementioned residues are relatively-conserved across the newly-proposed family of Tet(X) tigecycline-inactivating enzymes (Table 1). Among them, a number of residues have been functionally verified. In the case of Tet(X4), two of 5 substrate-binding cavity (H231 and M372) and three of 6 FAD-interactive residues (E43, R114, and D308) somewhat play roles in the phenotypic tigecycline resistance ²⁰. Structure-guided alanine substitution of Tet(X6) revealed that i) two of 5 tigecycline-binding residues (H224A and M365A) partially determine its phenotypic resistance to tigecycline (Fig. 3G); and ii) three of 6 FAD-surrounding residues (E36A, R107A, and D301A) give differential level of impact on its resultant tigecycline resistance (Fig. 3H). Thus, this result represents a functional proof for Tet(X6) as a new member of the expanding family of Tet(X) enzymes that have a role in the action of tigecycline degradation.

Table 1

Substrate-loading cavities across Tet(X)-type enzymes

The substituted residues are indicated in yellow background.

*Tet(X14) is renamed from Tet(X5) of Acinetobacter pittii p13C018-1 [acc. no.: CP040912]; Tet(X15) is relabeled from Tet(X6) of Proteus cibarius [acc. no.: WURM01000016]; Tet(X16) replace the redundant Tet(X6) from Proteus mirabilis [acc. no.: WURR01000048]; and Tet(X17) is corrected from Tet(X6) of Myroides phaeus [acc. no.: CP047050].

Figure 3 Structural and functional analysis of the substrate tigecycline-loading tunnel in Tet(X6)

Surface structure of Tet(X6) with tigecycline-loading tunnel in the front view (A) and the rotated view (B).

The substrate-loading tunnel in open state is indicated with an arrow.

C. Sectional view of tigecycline-occupied cavity in Tet(X6) protein

D. Sectional view of FAD-binding cavity in Tet(X6) enzyme

E. An enlarged view of tigecycline-binding cavity in Tet(X6)

F. Structural snapshot of the FAD cofactor-binding motif in Tet(X6) The critical residues are labeled. The molecule of tigecycline is colored in cherry red, and the FAD molecule is labeled in purple.

G. Use of site-directed mutagenesis to assay the five tigecycline-binding residues in Tet(X6)

H. Structure-guided alanine substitution analyses of FAD-interacting residues in Tet(X6)

Negative control is the E. coli MG1655 alone, and positive control refers to the MG1655 expressing the wild-type of tet(X6) (Table S1). The pBAD24-driven expression of Tet(X6) and its derivatives is triggered by the addition of 0.2% arabinose. Three independent assays were conducted, each of which is indicated with a dot.

Action of inactivation of tigecycline by Tet(X6)

To further elucidate biochemical mechanism of Tet(X6)-catalyzed tigecycline destruction (Fig. 4), we integrated an in vivo approach of microbial bioassay (Fig. 4A) with the in vitro system of enzymatic reaction (Figs 4B-D). This tigecycline bioassay we developed is dependent on the indicator strain DH5α of E. coli in that it was verified to be tigecycline susceptibility (Fig. 4A). As predicted, a zone of bacterial inhibition was clearly seen to surround a paper disk of the blank control, on which 2.5µg/ml tigecycline is spotted (Fig. 4A). A similar scenario was also seen with the negative control, i.e., the supernatant from E. coli MG1655 having the empty vector pBAD24 alone (Fig. 4A and Table S1). In contrast, the zone of tigecycline inhibition disappears around the paper disk containing supernatants of E. coli MG1655 expressing either tet(X6) or its homologous gene tet(X3) (Fig. 4A). This highlighted an in vivo role of tet(X6) [and/or tet(X3)] in the degradation of tigecycline.

Figure 4 Action and mechanism for Tet(X6) tigecycline resistance

A. The bioassay for Tet(X6) [and Tet(X3)]-mediated destruction of tigecycline The presence of tigecycline on paper disks leads to the appearance of a transparent zone of microbial inhibition. In contrast, the inhibition zone is absent upon the destruction of tigecycline by Tet(X6) [and/or Tet(X3)].

B. Gel filtration analysis of the purified Tet(X6) resistance enzyme To address the solution structure of Tet(X6) and Tet(X3), the method of size exclusion chromatography was applied. The protein samples were loaded into Superdex 200/300GL size exclusion column (GE Healthcare, USA). The eluted protein of interest (indicated with an arrow) is visualized in the right-hand inside gel (15% SDS-PAGE). The almost-identical elution volume (∼16ml) of Tet(X6) and Tet(X3) is generally consistent with the apparent size of molecular weight ∼36kDa in the inside PAGE gel, validating the monomeric form. Of note, the yellow color of Tet(X6) [Tet(X3)] solution is given in inside gel on the left hand, indicating the presence of a FAD cofactor-bound protein form.

LC/MS identification of tigecycline (C) and its oxygenated product by Tet(X6) enzyme (D)

Compared with the peak of tigecycline at 586.2 (m/z), the oxygenated derivative of tigecycline is reflected by a unique peak of 602.2 (m/z) in that it is added with an oxygen atom.

E. A chemical reaction model that Tet(X6) destructs/inactivates tigecycline It was adapted appropriately from Xu et al. ²⁰ with permission. FAD-Tet(X6) denotes Tet(X6) enzyme in the form of binding FAD cofactor. Chemical structures were given with ChemDraw.

Designations: blank, the LBA media with only tigecycline; vec, the vector of pET21a; LC/MS, liquid chromatography mass spectrometry; FAD, flavin adenine dinucleotide; NADP⁺, the oxidized form of nicotinamide adenine dinucleotide phosphate; NADPH, the reduced form of nicotinamide adenine dinucleotide phosphate.

Subsequently, we produced the recombinant forms of Tet(X6) and its homolog Tet(X3) and examined their enzymatic activities in vitro. Notably, Tet(X6) and Tet(X3) protein consistently gives yellow in solution (Figs 4B and 5B), hinting a possibility of being occupied with a FAD cofactor (Fig. 5A). Indeed, optical absorbance spectroscopy demonstrated the presence of Tet(X6)-bound FAD cofactor (Figs 5A-B). Gel filtration analysis indicated that both Tet(X6) and Tet(X3) display the solution structure of being a monomer (Fig. 4B). This generally agrees with the apparent molecular mass (∼36kDa) seen in the SDS-PAGE (Fig. 4B). The identification of polypeptide fingerprint with liquid chromatography (LC)/mass spectrometry allowed us to further study the catalytic action of Tet(X6) [Tet(X3)] using the in vitro reconstituted system of tigecycline oxygenation (Figs S5A-B). As expected, LC/MS-based detection of the substrate tigecycline showed a unique peak at the position of 586.2 m/z (Fig. 4C). In particular, the reaction mixture of Tet(X6) [Tet(X3)] consistently gave two distinct peaks in the spectrum of LC/MS (Figs 4D and S6). Namely, they correspond to a peak of substrate tigecycline (586.2 m/z), and an additional peak assigned to its oxygenated product of tigecycline at the position of 602.2 m/z (Figs 4D and S6). Notably, the method of double-reciprocal plot (Figs 5C-D) was exploited to measure the kinetic parameters (Fig. 5E) of Tet(X6) enzyme for the reactant tigecycline. As a result, Km of Tet(X6) was calculated to be 42.6±4.3 (Figs 5E-F), which is comparable to those of Tet(X2), Tet(X4) and Tet(X5) (Fig. 5F). This finding is consistent with an observation with the newly-identified Tet(X4) by He and coauthors ¹⁸.

Figure 5 Kinetic characteristics of Tet(X6) enzyme

A. Use of optical absorbance spectroscopy to detect the Tet(X6)-bound FAD cofactor The FAD solution (positive control) with yellow color (inside gel), features with two unique peaks at the wave-lengths of 370nm and 450nm. A similar scenario was also seen with the sample of Tet(X6) protein.

B. Purity of the recombinant Tet(X6) protein judged with gel filtration The purified protein of Tet(X6) with the yellow color (inside gel on left hand), was separated with 15% SDS-PAGE (inside gel on right hand). The gel filtration was developed with the Superdex 200 column run on AKTA Pure.

C. The standard curve of absorption of tigecycline at the wavelength of 400nm. The slope of the line is described the symbol “ε”. As for tigecycline, ε₄₀₀=4300 M^-1 cm^-1.

Use of double-reciprocal plot (D) to measure the kinetic parameters (E) of Tet(X6) enzyme Four independent experiments were conducted to generate the above double-reciprocal plots, giving kinetic parameters.

F. Kinetic constant of Tet(X6) is comparable with the counterpart of the other three known Tet(X) enzymes, namely Tet(X2), Tet(X4) and Tet(X5) The value of Tet(X6) arising from the data (panel D) are expressed in averages±SD. All the values of other three enzymes were reported by He and coworkers ¹⁸.

Designations: K_m, Michaelis-Menten constant; V_max, the maximum velocity of enzymatic reaction; K_cat, catalytic constant; V₀, the initial velocity; SD, standard deviation.

As Forsberg et al. ¹⁷ stated, similar scenarios were also seen with both Tet(X3) and Tet(X6) (Fig. S7), which is evidenced by the fact that liquid culture of tet(X3) [and/or tet(X6)]-bearing E. coli gives dark (Fig. S7). Unlike that the negative-control strain MG1655 with empty vector pBAD24 alone displays a big inhibition circle, E-test of tigecycline showed that expression of tet(X3) [or tet(X6)] renders the recipient strains significantly antagonistic to the tigecycline challenge (Fig. S8). Thereafter, we formulated a working model that Tet(X6) exploits a FAD cofactor to oxygenate/destruct the last-line antibiotic tigecycline (Fig. 4E). It seems likely that this chemical reaction proceeds via a ‘ping-pong’ action. However, this hypothesis requires further experimental evidence.

Variation in the mcr-1-containing cassettes

Sequence analysis of mcr-1-bearing elements from the three IncHI2-type plasmids (pMS8345A, pSA186_MCR1, and pDJB-3) reveals the core structure of “ISApl1-mcr-1” (Fig. 1C). Unlike the plasmid of pDJB-3 containing a cassette of “ISApl1-mcr-1-pap2-ISApl1”, the pSA186_MCR1 plasmid possesses an inactivated pap2 inserted with an inverted copy of ISApl1 (Fig. 1C). As a member of the IS30 family, the ISApl1 can transpose into its target by formatting a synaptic complex between an inverted repeat (IR) in the transposon circle and an IR-like sequence in the target ³⁶. After the initial formation of this composite transposon, one or both copies of ISApl1 might be lost. As such, this loss might improve the stability of mcr-1 in a diverse range of plasmids and then intensify its spread of mcr-1 ³⁷. Therefore, we favor to believe this model that pMS8345A having “ISApl1-mcr-1” alone (Fig. 1C) proceeds the loss of its downstream ISApl1 following the transposition. Not surprisingly, the E. coli strain carrying mcr-1 gives the minimum inhibitory concentration (MIC) at 4.0μg/ml. In addition, functional expression of a single mcr-1 allows the polymyxin-susceptible recipient strain of E. coli MG1655 to appear on the LB agar plate with colistin of up to 16μg/ml (Fig. S4B). Evidently, these data demonstrated that both tet(X6) and mcr-1 are actively co-carried by a single plasmid in an epidemic ST95 clone of pathogenic E. coli.

Physiological alteration by Tet(X6) and MCR-1

To address physiological consequence of Tet(X6) and MCR-1, we separately examined the pool of intracellular reactive oxygen species (ROS) and various growth curve-based metabolic fitness in an array of different E. coli strains (Figs 6 and 7A). As illustrated in the assay of fluorescence activated cell sorting (FACS), the cytosolic ROS level in the M1655 with empty vector alone was relatively low (Fig. 6A). Similar scenarios were also seen in derivatives of the MG1655 strain, regardless of the presence of mcr-1 (Fig. 6D), tet(X6) (Fig. 6E), and even both (Fig. 6F). As a consequence, the level of intracellular ROS was increased greatly upon its exposure to either colistin (Figs 6B and J) or tigecycline (Figs 6C and J). The expression of mcr-1 effectively prevented the colistin-stimulated ROS formation (Figs 6G and J). Similarly, the presence of tet(X6) robustly interfered the ROS production triggered by tigecycline (Figs 6H and J). In fact, the addition of both colistin and tigecycline only gave slight increment of ROS accumulation in the MG1655 strain co-harboring mcr-1 and tet(X6) (Figs 6I-K). Therefore, Tet(X6) attenuates the tigecycline-induced ROS generation as does MCR-1 in response to colistin (Fig. 6).

Figure 6 Use of flow cytometry to measure intracellular ROS level induced by colistin and/or tigecycline

A. FACS analysis of basal level of ROS in the negative strain E. coli MG1655 with empty vector alone Intracellular ROS level is boosted upon the addition of either colistin (B) or tigecycline (C) into the MG1655 strain

Cytosolic ROS level of the MG1655 strains expressing mcr-1 (D) or tet(X6) (E)

F. Determination of ROS level in the MG1655 carrying both mcr-1 and tet(X6)

G. Colistin cannot stimulate the ROS production in the MG1655 strain expressing mcr-1

H. Tigecycline cannot activates the ROS production in the MG1655 strain expressing tet(X6)

I. The mixture of tigecycline and colistin cannot significantly alter the cytosolic ROS accumulation in the MG1655 strain co-harboring mcr-1 and tet(X6)

J. Flow cytometry-based determination of relative level of intracellular ROS in different E. coli MG1655 strains in response to colistin and/or tigecycline Along with Tukey-Kramer multiple comparisons post-test, the data is given using one-way analysis of variance (ANOVA). *p-value is less than 0.001.

K. A scheme for colistin/tigecycline-induced accumulation of ROS, and its interfered formation by MCR-1/Tet(X6) Blue asterisk denotes colistin, and green square refers to tigecycline. The oxygenated form of tigecycline is indicated with green square attached with a red dot. The reactive oxygen species (ROS) are illustrated with the symbols of lightning.

As for the induction of ROS, the two antibiotics used here denoted 0.2mg/ml colistin and 0.2mg/ml tigecycline.

Abbreviations: FACS, fluorescence activated cell sorting; ROS, reactivated oxygen species.

As expected, the presence of plasmid-borne mcr-1 can cause the delayed growth of its recipient host E. coli MG1655, whereas the empty vector not (Figs 7B&D). In agreement with earlier observations ^38-42, this underscored that MCR-1 exerts significantly fitness cost in E. coli. In contrast, the expression of tet(X6) fails to trigger any detectable retardation of bacterial growth (Fig. 7C), indicating that Tet(X6)-causing metabolic burden/disorder is minimal. To further probe whether or not the crosstalk between Tet(X6) and MCR-1 occurs in E. coli, we engineered an E. coli strain that coharbors derivatives of two compatible plasmids [one arises from a low copy number, lactose promoter-driven pWSK129 ⁴³, and the other is constructed from an arabinose-inducible pBAD24 with ampicillin resistance ⁴⁴]. In fact, the two resistance enzymes MCR-1 and Tet(X6) are produced by lactose-inducible pWSK129::mcr-1, and arabinose-activating pBAD24::tet(X6), respectively (Table S1). Evidently, the coexistence of tet(X6) and mcr-1 cannot exert any synergism on bacterial retarded growth (Fig. 7E).

Figure 7 Use of growth curves to monitor metabolic fitness caused by mcr-1 and tet(X6) in E. coli

A. Schematic representative of various E. coli strains carrying plasmid-borne

tet(X6), mcr-1 or both

B. The empty vector exerts no effect on bacterial growth of the E. coli MG1655

C. Expression of tet(X6) causes slightly the delayed growth of the E. coli MG1655

D. Expression of mcr-1 gives significantly metabolic fitness in its recipient host E. coli MG1655 slightly the delayed growth of the E. coli MG1655

E. Co-expression of mcr-1 and tet(X6) cannot lead to synergistic fitness cost of the recipient strain

The data was given in means ± SD from three independent plotting of growth curves. vec, pBAD24.

As recently performed with MCR-3/4, we also adopted an approach of LIVE/DEAD cell staining to analyze an array of engineered strains (Fig. 8A). Unlike that the negative control, MG1655 strains with vector alone are almost fulfilled with alive cells (Figs 8B-C), the mcr-1-producing strains contained around 30% dead cells (Figs 8D-E, and J). Consistent with that no retarded growth is associated with Tet(X6) (Fig. 6C), confocal microscopy assays illustrated that relatively-low level of DAED/LIVE ratio is present in the tet(X6)-carrying MG1655 (Figs 8F-G, and J). Not surprisingly, the co-expression of mcr-1 and tet(X6) cannot promotes significant increment in the ratio of DEAD/LIVE cells (Figs 8H-J), when compared with the mcr-1-positive strains (Figs 8D-E). The remaining question to ask is whether or not Tet(X6) tigecycline resistance can crosstalk with MCR-1 colistin resistance in a given strain (Fig. 9). Thus, we designed such an E. coli strain FYJ4022 (Table S1), which co-harbors pWSK129::mcr-1 and pBAD24::tet(X6). In this engineered strain, the expression of mcr-1 is turned on by the addition of lactose, and tet(X6) is finely tuned by the supplementation of arabinose (Fig. 9A). The colistin resistance by MCR-1 was found to be independently of the presence of Tet(X6) (Fig. 9B), and vice versa (Fig. 9C). Further, MALDI-TOF mass spectrometry confirmed that the insusceptibility to polymyxin, arises from the PEA addition to lipid A by MCR-1, regardless of the presence of Tet(X6) in E. coli (Figs 9D-E).

Figure 8 Confocal microscopy-based visualization for bacterial fitness cost caused by functional expression of tet(X6) (and/or mcr-1) in E. coli

A. Scheme for engineered strains of E. coli with inducible expression of tet(X6), mcr-1, or both

B&C Bacterial viability of the negative control strains, E. coli MG1655 with the empty vector alone is not affected by the presence of the inducer arabinose and/or lactose

D&E The ratio of DEAD/LIVE cells suggests that arabinose-induced expression of mcr-1 gives appreciable level of fitness cost in E. coli

F&G Regardless of the inducer lactose added, the MG1655 strains harboring tetX6 is indistinguishable, when compared with those of negative control strains

H. The mcr-1 and tet(X6)-coharboring MG1655 does not display the phenotypic fitness cost, without the addition of neither arabinose nor lactose

I. The co-expression of MCR-1 and Tet(X6) cannot exert synergistic effect of metabolic fitness

J. Bacterial viability-based measurement of bacterial fitness costs in various MG1655 strains bearing either mcr-1 or tet(X6) (or both) Bacterial viability was determined through cell staining with LIVE/DEAD kit, which is followed by imaging with confocal laser scanning microscopy. The color of green and red separately denotes alive and dead cells. Data was collected from four independent trials and evaluated using one-way analysis of variance (ANOVA) followed by Tukey–Kramer multiple comparisons post hoc test. **p-value is less than 0.01. ns, no significance.

Figure 9 No crosstalk between MCR-1 and Tet(X6) in a given E. coli strain

A. Schematic representative of an engineered E. coli strain co-carrying mcr-1 and tet(X6) in four different modes of expression A symbol of triangle denotes intact LPS-lipid A, whereas the symbol “” refers to the PPEA-4’-lipid A, a derivative of lipid A with the addition of PEA moiety.

B. MCR-1 colistin resistance occurs independently of Tet(X6) tigecycline resistance

C. Tet(X6) tigecycline resistance proceeds independently of MCR-1 colistin resistance

D. Scheme for remodeling of bacterial lipid A by MCR-1

E. Structural identification of lipid A species from different strains of E. coli expressing a single tet(X6)/mcr-1 or both The strain tested here denotes FYJ4022 (Table S1), which is MG1655 co-harboring pWSK129::mcr-1 and pBAD24::tet(X6). The assays of susceptibility to colistin and tigecycline were performed with LB Agar plates with varied levels of antibiotics. The symbol “-” refers to no addition of arabinose or lactose, whereas “+” denotes addition of arabinose and/or lactose.

Abbreviations: PE, Phosphatidylethanolamine; DAG, Diacylglycerol; PPEA, Phosphoethanolamine; MW, molecular weight.

Together, these data suggested that no synergism is detected in fitness costs caused by the lipid A modifier MCR-1 and the tigecycline-inactivating enzyme Tet(X6). Unlike that MCR-1 modifies bacterial lipid A, the initial target of the cationic antimicrobial peptide colistin receptor ⁵, Tet(X6) hydrolyzes the family of tetracycline and its derivatives, like tigecycline (Fig. 4) ^15,20. The former action results in bacterial surface remodeling by MCR-1 through an addition of PEA moiety to 1(4’)-phosphate position of lipid A ^8,9. Consequently, this might in part shape metabolic flux of the recipient microbe to balance mcr-1 expression and bacterial survival stressed with colistin, producing the phenotypic fitness cost ³⁸. In contrast, it seems likely that the destruction of tigecycline by the flavin-dependent Tet(X) enzyme exerts minor effects on metabolic process (or claims few metabolic requirement). While such explanation for the limited fitness cost by Tet(X6) needs more experimental explorations.

Inability of tigecycline to treat Tet(X)-producing E. coli

Since that Tet(X6) possesses the activity of oxygenating tetracycline (Fig. S7) and its derivative tigecycline (Figs 3-4), it is reasonable to anticipate it might interfere effectiveness of tigecycline in clinical sector. As very recently Song and coworkers ⁴⁵ established in the case of mcr-1, we also adopted the infection model of Galleria mellonella (G. mellonella) to address this prediction (Fig. 10A). Given the constitutive expression of resistance enzymes in the recipient host, both mcr-1 and tet(X6) were fused with the native promoters and then cloned into a low-copy vector of pWSK129 to give pWSK129::Pmcr-1 and pWSK129::P2tet(X6), respectively (Table S1). Subsequently, these two recombinant plasmids were separately engineered into the well-known virulent strain EDL933 of E. coli O157:H7, which generated strain FYJ4039 carrying pWSK129::Pmcr-1 and strain FYJ4040 containing pWSK129::P2tet(X6) (Table S1). Unlike the negative control group that are consistently killed within 36 hrs after the treatment of PBS alone (Fig. 10B), 5 out of 8 larvae survived in the treatment of colistin (7.5mg/kg), 1h post-infection of virulent EDL933 strains (Fig. 10B). Notably, all the eight larvae were killed by the MCR-1-producing pathogenic strains of EDL933, regardless of colistin treatment (Fig. 10B). This revealed that mcr-1 renders colistin inefficient in the infection model of G. mellonella. In fact, similar scenarios were observed with mcr-1 in the infection models of G. mellonella ⁴⁵ and mouse thighs ^27,45. As expected, the tigecycline-based therapy (4mg/kg) seemed effective in part (if not all), because that 6 of 8 larvae (75%) are alive within the whole monitoring period of 72hrs post-infection of virulent E. coli O157:H7 (Fig. 10C). Whereas in the negative control of PBS, none of larvae is exempt from the killing by the virulent strain EDL933 (Fig. 10C). Not surprisingly, nearly all the 8 infected G. mellonella still were dead, despite that they were treated with tigecycline (Fig. 10C). Consistent with that of tet(X4) reported by Sun et al. ¹⁹, the observation also enabled us to believe that Tet(X6) abolishes clinical effectiveness of tigecycline. In summary, MCR-1 and Tet(X6) are posing challenges to the renewed interests of colistin and tigecycline, as two last-resort antibiotics used in clinical therapies against severe infections by pathogenic bacteria with multiple resistance.

Fig. 10 Survival curves of G. mellonella suggested that the two treatments of both colistin and tigecycline are ineffective for the E. coli infections producing either MCR-1 or Tet(X6)

A. Schematic representative for G. mellonella infection model

B. The failure in colistin treatment as for G. mellonella infected with mcr-1-harboring virulent E. coli

C. Tet(X6) renders tigecycline useless in the infection model of G. mellonella

In addition to the virulent strain EDL933 of EHEC O157:H7, the two derivative strains were included here (Table S1). Namely, they referred to FYJ4039 with pWSK129::Pmcr-1(panel B) and FYJ4040 harboring pWSK129::P2tet(X6) (panel C).

A representative result is given from three independent assays.

Conclusions

The pMS8345A, a large IncHI2-type MDR-plasmid is firstly identified by Beatson and coworkers ²⁴ to coexist with a big ColV-like virulence plasmid in the ST95 virulent lineage of E. coli. This alarms us that the spread of such pathogen might herald an era of post-antibiotic where we stand. The data we report here furthers our understanding tigecycline resistance mechanism of TetX family enzymes (Fig. 4E). To the best of knowledge, it is a first report addressing a case of the co-transfer of tet(X6) and mcr-1 by a single plasmid. Since that no known mobile elements are adjacent to tet(X6), we hypothesize that the transposon of “ISApl1-mcr-1-pap2-ISApl1” mediates the transfer of mcr-1 into this plasmid. The discovery of Tet(X6), a new member of Tet(X) family, allows us to engineer an array of Tet(X)-expressing bacteria, which paves a way to the development of bioremediation strategy for the environmental tetracycline contamination in the agricultural/industrial productions.

Also, the co-carriage of tet(X6) and mcr-1 on a single IncHI2-type plasmid is far different from the observation by Sun et al. ¹⁹ that the co-existence of tet(X4) and mcr-1 is mediated by two distinct plasmids in an E. coli clone. Unlike that the mobility of tet(X4) relies on ISCR2-mediated transposition ¹⁸, the gain of tet(X6) transferability is not clear (Fig. 1B). Not surprisingly, the IncHI2-type plasmid carries mcr-1 along with tet(X6) here, because that it has ever been found to act as a vehicle of global mcr-1 dissemination ^8,46. As expected, two types of different antibiotics (colistin and tigecycline) consistently stimulate the formation of hydroxyl radicals in E. coli (esp. ROS, in Fig. 6), which might constitute an additional example and/or evidence for an improved postuate of “efficient antibiotic killing associated with bacterial metabolic state [ATP ^47-49 and ROS ^50-53]. As for a given version of MCR resistance determinants ^39,41,42, the recipient bacterial host has been demonstrated to give fitness cost exemplified with the delayed growth prior to the entry into log-phase. It is reasonable that the presence of either mcr-1 or tet(X6) also cause metabolic fitness to some extent. Consistent with scenarios with MCR-like members by Yang et al. ³⁸ and Zhang et al. ^39,40,42, we verified the fitness cost caused by MCR-1 (Figs 7-8). This is in part (if not all) explained by the fact that bacterial membrane integrity is altered by MCR-1-mediated lipid A remodeling ⁹. In contrast, the expression of tet(X6) does not lead to metabolic burden detected (Figs 7-8), which is probably because that Tet(X6) destructs the antibiotic of glycyl-cycline tigecycline, rather than the ribosome target ¹⁰. It is unusual, but not without any precedent. A similar scenario was seen with the other resistance enzyme β-lactamase-encoding gene bla_TEM1b (i.e., no fitness cost is correlated with it) ³⁸. Therefore, we are not surprised with that no synergistic fitness arises from the co-carriage of tet(X6) and mcr-1 in E. coli. Although that a growing body of new tet(X) variants [tet(X6) to tet(X17)] have been proposed in this study (Fig. 2), most members of this family, apart from tet(X6), await experimental demonstration in the near future. This is because that rare case of cryptic version might occur naturally, such as the prototypical tet(X) [we called tet(X0)] preexisting in an obligate anaerobe Bacteroides fragilis ²¹. Given that i) tigecycline and colistin both are one of few alternative options to combat against carbapenem-resistant Enterobacteriaceae and Acinetobacter species, ii) both MCR-1 ⁶ and Tet(X4) ¹⁹ have accordingly rendered colistin and tigecycline ineffective in the therapy of mice with MDR infection, the co-occurrence and co-transfer of tet(X6) and mcr-1 by a single plasmid amongst epidemic pathogens is a risky challenge to public health and clinical therapies.

Taken together, it is plausible and urgent to introduce mcr variants along with, but not only limited to, tet(X) variants in the routine national (and/or international) investigation in the context of “one health” (environmental/animal/human sectors). Along with major findings of other research group ¹², functional definition of Tet(X3) and its homologue Tet(X6) here extends mechanistic insights into Tet(X) tigecycline resistance, and even benefit the development of anti-Tet(X) resistance enzyme inhibitors.

Materials and Methods

Sequencing, assembly and annotation of plasmids

The plasmid pDJB-3 was isolated from the colistin-resistant E. coli DJB-3 of swine origin, verified with mcr-1-specific PCR, and then subjected to genome sequencing with the Hiseq X ten PE150 sequencer platform (Illumina, USA). As a result, the DNA library of pDJB-3 plasmid prepared by KAPA Hyper Prep Kit (Roche, Basel, Switzerland) gave a pool of 150 bp paired-end reads that are destined to be assembled into a contig by the SPAdes Genome Assembler (version 3.11.0). A BLASTN search was conducted to probe whether or not the resultant mcr-1-containing contig has a best-hit plasmid candidate. Together with Sanger sequencing, PCR was applied to close all the suspected gaps.

The resultant plasmid genome was annotated through the prediction of open reading frames (ORFs) with RAST (rapid annotation using subsystem technology, http://rast.nmpdr.org). PlasmidFinder 1.3 (https://cge.cbs.dtu.dk/services/PlasmidFinder-1.3/) was used to type the plasmid incompatibility, and ResFinder 3.1 (https://cge.cbs.dtu.dk/services/ResFinder/) was applied to screen possible antimicrobial resistance genes. The plasmid map was given with GenomeVx (http://wolfe.ucd.ie/GenomeVx/), and its linear alignment was proceeded with Easyfig ⁵⁴.

Plasmid conjugation experiments

As recently described by Sun et al. ²⁷, the experiments of plasmid conjugation were performed, in which the rifampin-resistant E. coli recipient strain EC600 (and/or strain DJB-3) acted as a donor. In brief, overnight cultures were re-grew in LB broth, donor and recipient strains were mixed at the logarithmic phase and spotted on a filter membrane, and then incubated at 37°C overnight. Subsequently, bacteria were washed from filter membrane and spotted on LB agar plate containing 400µg/ml rifampin and 4 µg/ml colistin for selection of transconjugants. The suspected transformants were validated with PCR assays.

Molecular and microbial manipulations

With all the known tet(X) variants [tet(X0)-tet(X5)] as queries, BLASTN was carried out. In particular, a tet(X3)-based search returned a plasmid pMS8345A with significant hit, leading to the discovery of new variant of tet(X6). Then, tet(X6) was synthesized in vitro, and cloned into pABD24, giving pABD24::tet(X6) (Table S1). Following the verification of its identity with direct DNA sequencing, this recombinant plasmid was introduced into the MG1655 strain of E. coli to assess its role in vivo. The generation of all the point-mutants of tet(X6) were based on pBAD24::tet(X6) (Table S1) using site-directed mutagenesis kit (Vazyme Biotech), along with an array of specific primers (Table S2). To test relationship of MCR-1 with Tet(X6), mcr-1 was cloned into pWSK129, giving pWSK::mcr-1, compatible with pBAD::tet(X6) within a single E. coli colony (Table S1). After experimental validations of phenotypic clolistin resistance (and/or tigecycline resistance), all the bacterial were subjected to routine isolation of crude lipo-polysaccharides-lipid A as earlier recommended by Caroff et al. ⁵⁵. The identity of purified lipid A species were verified with MALDI-TOF/TOF mass spectrometry (Bruker UltrafleXtreme, Germany) ²⁵.

As recently described with tet(X4) ²⁰, the ability of tet(X6) and tet(X3) in phenotypic tigecycline resistance was evaluated with LB agar plates containing tigecycline in a series of dilution. The strain expressing tet(X4) is used as a positive control. In addition, the MCR-1 colistin resistance was also judged as we earlier conducted with mcr-1 ⁵⁶ with little change. All the examined E. coli strains were cultivated at 37°C overnight. Overnight cultures were standardized to OD₆₀₀ 0.05, inoculated (1:10; v/v) into 96-well glass-bottomed plates in fresh LB broth ± arabinose or lactose (0, 0.02%, and 0.2%, w/v), and shaken (180 r.p.m) at 37°C. Of note, arabinose acted as an inducer of pBAD24, and lactose was used to trigger expression of pWSK129-based MCR-1. As the establishment with NMCR-1 ³⁹ and MCR-3/5 ^41,42, bacterial growth curves were automatically plotted with spectrophotometer (Spectrum lab S32A) to evaluate the fitness cost caused by tet(X6) and mcr-1. During the total period of 20 hours, the value of optical absorbance (i.e., OD600) was consistently recorded at an interval of 1 hour.

Bioassays for tigecycline destruction

The hydrolytic activity of Tet(X6) [and/or Tet(X3)] enzyme was determined as Balouiri et al. ⁵⁷ described with little change. In brief, the strain of MG1655 harboring pBAD24::tet(X6) [or pBAD24::tet(X3)] was cultivated overnight on LB agar plates supplemented with 0.1% arabinose. As a result, bacterial colonies stripped, were suspended with 0.5ml of LB broth containing 0.1% arabinose and 2.5mg/ml tigecycline, whose optical density at 600nm (OD600) was adjusted to about 2.0. Then, the suspension cultures were proceeded to 8h stationary growth at 37°C. Following centrifugation (13,600rpm, 20min) and filtration (at 0.22μm cut-off), bacterial supernatants were prepared. The E. coli DH5α here referred to an indicator strain of tigecycline susceptibility. Of note, the overnight culture of E. coli DH5α (∼100μl) was spread on a LB agar plate, which is centered with a paper disk of 6mm diameter. To visualize the inhibition zones, the supernatant of interest (∼20µl) was spotted the paper disk on the aforementioned bioassay plates, and incubated at 37°C for 16h. The negative-control denotes the supernatant from E. coli MG1655 bearing the empty vector pBAD24 (Table S1), and the blank control referred to the LB broth containing 2.5mg/ml tigecycline.

Expression, purification and identification of Tet(X) enzymes

To produce the Tet(X6) protein and its homologue Tet(X3), the strains of E. coli BL21 carrying pET21::tet(X6) [and pET21::tet(X3)] were engineered (Table S1) for the inducible expression via the addition of 0.5 mM isopropyl β-D-1-thiogalactopyranoside (IPTG). Bacterial lysates obtained by a French Press (JN-Mini, China), were subjected to 1h of centrifugation at 16,800 rpm at 4°C, and the resultant supernatants were incubated with pre-equilibrated Ni-NTA agarose beads on ice for 3 hours. Following the removal of protein contaminants, the Tet(X6) [and/or Tet(X3)] protein was eluted from the Ni-NTA agarose beads using the elution buffer [20mM Tris-HCl (pH 8.0), 150mM NaCl, 20mM imidazole, and 5%glycerol], and concentrated with a 30kDa cut-off ultra-filter (Millipore, USA). Subsequently, gel filtration was performed to probe solution structure of Tet(X6) [Tet(X3)], using a Superdex 200/300GL size exclusion column (GE Healthcare). The purity of protein pooled from the target peak was judged with SDS-PAGE (15%), and its identity was validated with MALDI-TOF/TOF mass spectrometry (LTQ orbitrap Elite, Thermo Fisher).

Enzymatic activity for Tet(X6) in vitro

To confirm the enzymatic activity of Tet(X6) enzyme, the in vitro reaction system was established as recently described by Sun et al. ¹⁹ with little change. The components of this assay (50μl in total) consisted of 20mM Tris (pH7.5), 150mM NaCl, 1mM NADPH, 4mg/ml tigecycline, and the purified enzyme [2mg/ml for either Tet(X6) or Tet(X3)]. Following the maintenance (∼12h) of enzymatic reaction at room temperature, the resultant reaction mixture was subjected to further analysis of liquid chromatography mass spectrometry (LC/MS) using an Agilent 6460 triple quadrupole mass spectrometer (Agilent Technologies, USA) ¹⁸. As for LC/MS here, it was carried out as follows: i) Nitrogen acted as the sheath gas and drying gas, the nebulizer pressure was set to 45 psi, and the flow rate of drying gas was 5 liter/min. The flow rate and temperature of the sheath gas were 11 liter/min and 350°C, respectively; ii) Chromatographic separation proceeded on a Zorbax SB C8 column (150 × 2.1mm, 3.5µm); iii) Mass spectrometric detection was completed using an electrospray ionization (ESI) source in positive mode. Scan range was 100∼1000amu; and iv) The resultant data was processed with Agilent Mass Hunter Workstation.

The steady state kinetic assay of Tet(X6)

The decrease in absorbance corresponding to tigecycline hydroxylation by Tet(X6) were monitored at 400nm (ε₄₀₀ =4300 M⁻¹cm⁻¹) over 6 min. To determine the steady-state kinetics parameters for Tet(X6), we measured initial velocities (V₀) of tigecycline inactivation at varied concentration of tigecycline (60μM, 80μM, 100μM, and 120μM), 1mM NADPH, 5mM MgCl₂, 0.5μM Tet(X6) protein, 20mM Tris-HCl (pH8.5) concentrations at 37°C ¹⁵. Each 200μl reaction in 96-well micro-titre was monitored using SPECTROstar ^Nano. All assays were performed in triplicate. Steady-state kinetic parameters were determined by fitting initial reaction rates (V₀). The data was analyzed according to the standard Michaelis-Menten equation. The double-reciprocal plot featuring with the formula “1/V₀=(Km/V_max)/[S]+1/V_max” was used to calculate the K_m. Accordingly, V₀=V_max [S]/(K_m+[S]). The catalytic constant k_cat was determined according to the V_max = k_cat [E₀], and E₀ is total enzyme concentration ⁵⁸.

Flow cytometry

Mid-log phase cultures (OD600, ∼1.0) were prepared for the detection of intra-cellular reactive oxygen species (ROS). The oxidant sensor dye, DCFH2-DA (sigma) was mixed with bacterial strains and kept for 0.5h. Accordingly, the 2.0mg/ml of antibiotics (colistin and/or tigecycline) were supplemented. Then, bacterial samples (10⁵∼10⁶) diluted with 0.85% saline were subjected to the analysis of flow cytometry ^40,42. The resultant FACS data was recorded with a BD FACSVerse flow cytometer through counting 10,000 cells at a flow rate of 35ml/min (and/or 14ml/min). In particular, DCFH florescence was excited with a 488nm argon laser and emission was detected with the FL1 emission filter at 525nm using FL1 photomultiplier tub.

Confocal microscopy

As recently described ³⁸, confocal microscopy was conducted to examine the potential effects on bacterial viability exerted by resistance enzymes [MCR-1 and/or Tet(X6)]. Prior to assays of confocal microscopy, mid-log phase cultures were processed with the LIVE/DEAD BacLight™ Bacterial Viability Kit (Cat. No. L7012) ³⁸. Namely, the three strains tested here included i) E. coli MG1655 (mcr-1/pWSK), ii) MG1655 [tet(X6)/pBAD], and iii) MG1655 [tet(X6)/pBAD and mcr-1/pWSK). Of note, 0.2% lactose is an inducer of mcr-1 expression, and 0.2% arabinose acts as an activator for Tet(X6) enzyme production. After the removal of supernatants, bacterial biofilms were stained with 3% LIVE/DEAD kit solution, and maintained at room temperature in the dark for 15 minutes. Photographs were captured by the confocal laser scanning microscopy (Zeiss LSM 800) with a 63× oil immersion lens and analyzed using COMSTAT image analysis software. The Tukey–Kramer multiple comparison post hoc test was applied to judge the COMSTAT data. Statistical significance was set at p< 0.01 with T-test.

Infection model of G. mellonella

To probe possible interferences of mcr-1 and/or tet(X6) in the anti-bacterial treatment with colistin (and/or tigecycline), the infection model of Galleria mellonella (G. mellonella) was applied here. Prior to bacterial infections, the larvae of G. mellonella (Tianjin Huiyude Biotech Company, Tianjin, China) was assessed as for the weight (0.3-0.4g each) and its active status, and then grouped appropriately (8 per group). The mid-log phase cultures of the virulent E. coli (EHEC O157:H7) with or without plasmid-borne mcr-1 [and/or tet(X6)] were prepared (Table S1), and then suspended with 1xPBS buffer, in which the final OD600 is 0.1. As recently Song et al. performed ⁴⁵ with minor change, each larvae was injected with 10ul of bacterial solution (1.0x 10⁵ cfu) at the left posterior gastropoda. After 1h post-challenge, the infected larvae separately received the different treatments on the right posterior gastropoda ⁴⁵. Namely, they referred to PBS, colistin (7.5mg/kg), and tigecycline (4mg/kg) ⁴⁵. Survival rate of G. mellonella was monitored over 72hrs, of which an interval is 12hrs. Three biological replicates were performed.

Bioinformatics

Multiple sequence alignments of Tet(X) variants at the levels of both amino acids and nucleic acids proceeded with ClustalOmega (https://www.ebi.ac.uk/Tools/msa/clustalo). Consequently, the phylogeny of Tet(X) was generated with TreeView (https://www.treeview.co.uk/). Tet(X6) was structurally modeled using Swiss-Model (https://swissmodel.expasy.org/interactive) ⁵⁹, in which the structural template detected refers to Tet(X2) (PDB: 2Y6Q) ^16,58. Both GMQE (global model quality estimation) and QMEAN (a global and local absolute quality estimate on the modeled structure) was applied to judge the quality of the modeled structure. Finally, structural presentation and cavity illustration of Tet(X6) was given with PyMol (https://pymol.org/2).

Accession numbers

Nucleotide sequence data of tet(X6) reported here is available in the Third-Party Annotation Section of the DDBJ/ENA/GenBank databases under the accession number TPA: BK011183. The full genome sequence of the mcr-1-harboring plasmid pDJB-3 of swine origin is accessed under the accession number: MK574666.

Author contributions

YF designed and supervised this study; YF, YX, LL and HZ conducted experiments and analyzed data; YF and HZ contributed regents and interpreted data; YF and HZ drafted and reviewed this manuscript.

Competing Interests

We declare that no conflict of interest is present.

Supporting information

Supporting tables

Table S1

Bacterial strains and plasmids used in this study

Table S2

Primers used in this study

Supporting figures

Fig. S1 Nucleotide acid sequence analysis of tet(X3) and tet(X6)

To determine the similarity between tet(X3) and tet(X6), their nucleotide acid sequences were subjected to Clustal Omega (https://www.ebi.ac.uk/Tools/msa/clustalo/), and resultant form of sequence is given with ESPript 3.0 (http://espript.ibcp.fr/ESPript/cgi-bin/ESPript.cgi). Identical residues are in white letters with black background, and different residues are black letters with white background. The identity between tet(X3) and tet(X6) is 97.1%, with 31 substitution of 1137 residues in total.

Fig. S2 Evidence that an intact tet(X6) is only harbored on one plasmid and two contigs

Sequence alignment was conducted as described in Fig. S1.

An initial codon “ATG” and the stop codon “TAA” are underlined.

Fig. S3 Sequence alignment of the Tet(X6) enzyme with other five homologs (X0 to X5)

Multiple sequence alignment was conducted with Clustal Omega (https://www.ebi.ac.uk/Tools/msa/clustalo/), generating its output with ESPript 3.0 (http://espript.ibcp.fr/ESPript/cgi-bin/ESPript.cgi). The putative substrate-loading cavity is composed of FAD-interactive residues (with red triangles) and Tigecycline-binding residues (with blue triangles).

Fig. S4 Co-occurrence of tet(X6) and mcr-1 gives co-resistance to colistin and tigecycline

A. Contrasting the level of tigecycline resistance by Tet(X) variants The E. coli MG1655 carrying different variants of tet(X) [X3, X4, and X6] were maintained at 37°C on LB agar plates containing tigecycline in series of dilution.

B. The MCR-1 confers phenotypic colistin resistance in the E. coli MG1655 The derivatives of E. coli MG1655 bear mcr-1 or tet(X6) alone (or both) were spotted on LB agar plates containing colistin in series of dilution, and maintained at 37°C for overnight.

Fig. S5 MS identity of the purified two proteins Tet(X3) and Tet(X6)

A. MS-based identification of polypeptide fragments from the purified Tet(X3) protein

B. MS-based determination of the purified Tet(X6) protein The underlined letters denote the polypeptides that match Tet(X3) [and/or Tet(X6)] protein.

Fig. S6 Use of LC/MS to identify the oxygenated product of tigecycline by Tet(X3) enzyme

In the spectrum, the peak of 586.2 (m/z) refers to tigecycline, whereas the other peak of 602.2 (m/z) denotes an oxygenated product of tigecycline. Chemical structures were given with ChemDraw.

Fig. S7 Visualization for destruction of tetracycline by Tet(X3) [and Tet(X6)] enzyme

Unlike the fact that the blank and the negative control (liquid culture of E. coli with empty vector alone) display yellow, bacterial culture of Tet(X3) [and/or Tet(X6)]-expressing E. coli gives dark. This indicates that the expression of Tet(X3) [and Tet(X6)] leads to the oxygenation of tetracycline, which is fully consistent with the observation of other tetracycline-inactivating enzymes by Forsberg and coworkers ¹⁷. 0.1% arabinose is used to induce the expression of pBAD24-borne tet(X3) [and tet(X6)].

Designation: Blank, the LB liquid medium containing tetracycline; Vec, pBAD24

Fig. S8 Use of E-test to evaluate phenotypic growth of MG1655 strains expressing either tet(X3) or tet(X6)

Semi-solid medium was poured, which was supplemented with the MG1655 strains with the empty vector alone, or the plasmid-borne tet(X3) [and/or tet(X6)], accordingly. E-test strip is featuring with a gradient concentration of tigecycline. The cut-off value of tigecycline resistance is indicated with an arrow. A representative result of three independent experiments is given.

Acknowledgements

We would like to thank Dr. Yuanyuan Zhang for the technical assistance in flow cytometer assays with BD FACSVerse (Shared Management Platform for Large Instrument, College of Animal Sciences, Zhejiang University). We are grateful to Prof. Jian-hua Liu (South China Agricultural University, Guangzhou, China) for providing us the mcr-1-harboring plasmid, pHNSHP45-2. This work was supported by National Key R&D Program of China (2017YFD0500202, YF) and National Natural Science Foundation of China (31830001, 31570027 & 81772142, YF). Dr. Feng is a recipient of the national “Young 1000 Talents” Award of China.

Footnotes

• It is an improved version having multiple lines of text correction, image replacement, and new data!