Emergence of Morganellaceae Harboring blaIMP-27 Metalloenzyme in Canada

ABSTRACT In 2018 to 2019, PCR for carbapenemases in routine Gram-negative isolates submitted to the National Microbiology Laboratory revealed an increase in IMP-type metalloenzyme-positive isolates, mostly among Morganellaceae. Whole-genome sequencing revealed that 23 Morganellaceae harbored blaIMP-27 within a chromosomal Tn7 element. Phylogenomics indicated diversity of isolates but also the presence of a few clonal isolates dispersed geographically. These isolates may be difficult to detect due to carbapenem susceptibility and false-negative results in phenotypic testing. IMPORTANCE Over the last decade or so, the frequency of isolation of clinical carbapenemase-producing organisms (CPOs) has increased among health care-associated infections. This may seriously compromise antimicrobial therapy, as carbapenems are considered the last line of defense against these organisms. The ability of carbapenemases to hydrolyze most β-lactams in addition to the co-occurrence of mechanisms of resistance to other classes of antimicrobials in CPOs can leave few options for treating infections. The class B metalloenzymes are globally distributed carbapenemases, and the most commonly found include the NDM, VIM, and IMP types. Our study describes a sudden emergence of IMP-27-harboring Morganellaceae during 2018 to 2019 in Canada. There is a paucity of literature on IMP-27 isolates, and our data bolster the information on the genetic context, antimicrobial profiles, and phylogenomics of this group of CPOs.

variants of IMP have been assigned (https://www.ncbi.nlm.nih.gov/pathogens/beta -lactamase-data-resources/). In Canada, the first IMP carbapenemase identified was bla IMP-7 from an outbreak of nosocomial P. aeruginosa isolated from 1995 to 1997 in a single region (3). The Canadian Nosocomial Infections Surveillance Program identified only two IMP producers among 615 carbapenemase-producing Enterobacterales collected from 2010 to 2016, an Enterobacter cloacae isolate harboring bla IMP-13 and an Acinetobacter pittii isolate harboring bla IMP-26 (4,5). IMP-27 was first reported in 2011 from Proteus mirabilis PM185, isolated in 2009, with further studies determining that bla IMP-27 was on the chromosome in PM185, on an IncX8 plasmid and the chromosome in P. mirabilis PM187, and on a plasmid of unknown Inc type isolated from Providencia rettgeri PR1 (6)(7)(8). P. mirabilis GN855 harboring bla IMP-27 was reported from a patient in Ontario, Canada, in 2012 (9). Another study reported bla IMP-27 located on an IncQ1 plasmid found in multiple species of Enterobacterales, including P. mirabilis, Morganella morganii, and P. rettgeri, isolated from the environment of a swine operation in the United States (10).

RESULTS
Bacteria harboring bla IMP-27 . In 2018 and 2019 the National Microbiology Laboratory (NML) screened 2270 Gram-negative isolates by PCR for the most common carbapenemase gene groups, KPC, OXA-48, NDM, VIM, IMP, GES, and NMC/IMI. Twenty-eight isolates (1.2%) were positive by PCR for a bla IMP gene, including one P. rettgeri, 15 P. mirabilis, seven M. morganii, and five P. aeruginosa isolates. In 2017, of 242 P. aeruginosa and 30 Morganellaceae isolates received for routine carbapenemase PCR, four P. aeruginosa isolates and one M. morganii isolate (N17-03220) harbored an IMP gene. M. morganii N17-03220 was later found to be indistinguishable by pulsed-field gel electrophoresis (PFGE) from M. morganii N18-00103 received 56 days later (January 2018) and in fact was from the same patient, and it was no longer studied. Thus, there was a significant increase of Morganellaceae harboring bla IMP received by the NML after 2017. The 28 IMP-harboring isolates from 2018 to 2019 were from central (n = 9), western (n = 18), or eastern (n = 1) Canada and were isolated mainly from urine (n = 17), wounds (n = 4), or rectal swabs (n = 4). Whole-genome sequencing (WGS) analysis of all 2018-2019 isolates and P. mirabilis GN855 determined that all M. morganii isolates, 14 of the P. mirabilis isolates, and the P. rettgeri isolate harbored bla IMP-27 , while among the P. aeruginosa isolates, one harbored bla IMP-7 , one bla IMP-62 , and three bla IMP-26 . The bla IMP-27 gene could not be identified from the WGS data of one IMP PCR-positive P. mirabilis isolates and was presumed lost after subculture; therefore, this isolate was not further studied. Thus, among all the Morganellaceae received by the NML in 2018 to 2019 (n = 82) 26.8% (n = 22) were confirmed to harbor bla IMP-27 .
Antimicrobial susceptibility and detection of bla IMP-27 -harboring isolates. Antimicrobial susceptibilities were determined for all IMP-harboring isolates as well as a few non-carbapenemase-producing organisms (CPOs) for comparative purposes ( Table 1). As expected for Morganellaceae, most were intermediate (I) or resistant (R) to imipenem regardless of the presence/absence of IMP-27, confirming that this is not a suitable phenotype for indicating the possible presence of a carbapenemase. Gradient diffusion was poor for indicating IMP-27 presence, as most isolates were susceptible (S) to meropenem and ertapenem. By Sensititre testing, all IMP-27 P. mirabilis isolates and the P. rettgeri isolate were I or R to all carbapenems, while the non-CPOs were S to the three nonimipenem carbapenems. However, all of the M. morganii isolates were S to all nonimipenem carbapenems by Sensititre testing.
Full antibiograms were in congruence with the resistomes ( Table 2). Among the phenotypic tests (Table 2), the modified carbapenem inactivation method (mCIM) test was 100% specific and sensitive for carbapenemase presence/absence. All mCIM-positive isolates were also positive by EDTA-modified CIM (eCIM), correctly indicating the presence of a class B enzyme. The b-Carba test was 100% sensitive and specific for M. morganii and P. rettgeri, but all IMP-27-producing P. mirabilis isolates were falsely negative. The Carba-NP and Neo-Rapid Carb test, which work on the same principle, performed poorly, and all IMP-27-producing P. mirabilis isolates and the P. rettgeri isolate were falsely negative. Among IMP-27 M. morganii isolates, results for the Carba-NP and Neo-Rapid Carb tests were variable, with some exhibiting false-negative, invalid, or weakly positive results. We also tried the more expensive NG-Test CARBA 5 immunochromatographic assay on a small number of isolates, even though the package insert (ENO022CAR/Rev: 200131) does not list IMP-27 as one of the variants that can be detected by this test (Table 2). When the cells were obtained from tryptic soy agar (TSA)-blood plates (M. morganii) or Mueller-Hinton medium (P. mirabilis), all results were negative. Upon repeat testing with cells obtained from Mueller-Hinton containing 100 mg/ml ampicillin, a faint IMP-specific band was observed for the two IMP-27-harboring M. morganii isolates, though it was observed 5 to 10 min after the recommended test time of 15 min. The mCIM results indicate that IMP-27 is produced by all the bla IMP-27 -harboring isolates in the study. Nonetheless, we determined specific activity against imipenem for the isolates tested by NG-Test CARBA 5 and confirmed imipenemase activity in the IMP-27-harboring isolates, though the activities can vary by 2-to 5-fold (Table 3). Together, the results indicate that P. mirabilis is likely recalcitrant to lysis/permeabilization in the non-mCIM phenotypic tests, all of which have a cell suspension/lysis solution. For the M. morganii isolates, although results indicate that some lysis does occur, it may be suboptimal, and this, combined with low IMP-27 levels for some isolates and/or technical issues, may account for poor results in the non-mCIM phenotypic tests.
bla IMP-27 is found within a Tn7 element located in the chromosome. WGS analysis showed that the bla IMP-27 gene was located in the class 2 integron In2-71 (http:// integrall.bio.ua.pt/?), which was integrated into a Tn7 element (Fig. 1). This structure, labeled Tn7[In2-71], was inserted into the chromosome of all isolates via the attTn7 site at the 39 end of the glmS gene, the canonical bacterial Tn7 insertion site (11), and each element was flanked by direct repeats, indicating acquisition by transposition. Tn7[In2-71] elements were identified from the GenBank database (.99% identity) in P. mirabilis PM185 (accession no. NOWB01000038), P. rettgeri 106-1829X (accession no. KY847874), M. morganii 480-26370X (accession no. KY847873), and E. coli CFSAN051542 (accession no. CP020835). Sequence analysis divided the Tn7[In2-71] elements into two clades, A (n = 16) and B (n = 11), with the elements in clade A being .99.9% identical and the elements in clade B being 100% identical, but with the clades differing by 105 to 107 bp differences (Fig. 2). The vast majority of base pair differences were found in the tnsA-tnsB region indicating a region of recombination (data not shown). No plasmid replicons were identified in the P. mirabilis or the P. rettgeri isolates, whereas two M. morganii isolates harbored replicons (Table 2). Though IncQ1 plasmids have been found to harbor bla IMP-27 (10), the IncQ1 replicon in N18-00103 was found to be integrated into the chromosome and not linked to Tn7[In2-71]. Limited clonality revealed by core genome SNV analysis. We carried out core genome SNV analysis on all P. mirabilis and all M. morganii to determine strain relatedness ( Fig. 3A and B). Among the M. morganii isolates, 6 of 10 isolates are diverse, with the number of single nucleotide variants (SNVs) between them ranging from 83 to .14,000 (Fig. 3A). Four isolates clustered at 0 to 3 SNVs, but no strong epidemiological links could be uncovered between any of the four patients, though two isolates were from patients who had been in the same hospital but 470 days apart. The analysis of P. mirabilis, which included the U.S. IMP-27 isolates PM185 and PM187, showed that 10 isolates were diverse, differing by 752 to .12,800 SNVs from each other (Fig. 3B). However, the 11 P. mirabilis isolates harboring Tn7[In2-71]-B (Fig. 2) clustered together at 0 to 13 SNVs or 1 to 15 SNVs when reanalyzed separately with an internal reference and, hence, a larger core genome. Anonymized patient facilities were available for some isolates, indicating some common facilities, but the limited data make inferring direct transmission events unfeasible. Nonetheless, this cluster of closely related isolates can be postulated to have derived from a common ancestor that has spread to multiple locations in western Canada.

DISCUSSION
Morganellaceae isolates harboring bla IMP-27 have emerged in Canada since 2018. These isolates may be difficult to detect as CPOs, as they can exhibit susceptibility to carbapenems depending on which susceptibility testing method is used. The mCIM detected carbapenemase production or lack thereof among all study isolates, as did the b-Carba test for the M. morganii and P. rettgeri isolates. Though bla IMP-27 was exclusively chromosomally located here, its dissemination may be facilitated by being mirabilis isolates in this study as generated by the SNVPhyl Pipeline, which generates an alignment of high-quality valid SNVs through PhyML using the GTR1g model (15). Reference genomes used are indicated by an asterisk and were the closed genomes of M. morganii N18-00103 (CP048275) or P. mirabilis N18-00201 (CP048404) or a pseudogenome (concatenated contigs) of P. mirabilis N18-02940. SNVs or SNV ranges between isolates or groups of isolates are shown. For the main analysis of each group of the same species, boxed isolates do not harbor bla IMP-27 . For the subanalysis of the cluster of the closely related P. mirabilis isolates, each unique shape indicates a specific facility from which the bacterium was isolated. The isolates were isolated in Alberta except for the three from British Columbia (BC). harbored within a mobile Tn7 transposon. Isolates were diverse but phylogenomics revealed clones harboring bla IMP-27 have dispersed in Canada. The major limitation of this study was that isolates were voluntarily submitted to the NML, and thus, the prevalence of bla IMP-27 isolates may be underestimated. In addition, due to bla IMP family sequence variation, in-house primers and some commercial assays may yield false-negative results (12).

MATERIALS AND METHODS
Bacterial isolates. The bacteria in this study were from routine isolates voluntarily sent to the NML for carbapenemase PCR. Typically, organisms are sent because of a suspicion of carbapenemase production due to reduced susceptibility/resistance to a carbapenem and/or a positive result of a phenotypic method that indicates carbapenemase production. For the isolates that test positive, the PCR results are reported, and the carbapenemase gene is not sequenced unless by special request.