Prevalence of insertion sequence elements in plasmids relating to mgrB gene disruption causing colistin resistance in Klebsiella pneumoniae

Abstract Colistin is a last resort antibiotic for the treatment of carbapenemase producing Klebsiella pneumoniae. The disruption of the mgrB gene by insertion sequences (ISs) is a mechanism mediating colistin resistance. Plasmids encode mobilizable IS elements which integrate into the mgrB gene in K. pneumoniae causing gene inactivation and colistin resistance. The species prevalence of mgrB‐gene disrupting insertion elements ISL3 (ISKpn25), IS5 (ISKpn26), ISKpn14, and IS903B present on plasmids were assessed. IS containing plasmids were also scanned for antimicrobial resistance genes, including carbapenem resistant genes. Plasmids encoding ISs are abundant in K. pneumoniae. IS903B was found in 28 unique Inc groups, while ISKpn25 was largely carried by IncFIB(pQil) plasmids. ISKpn26 and ISKpn14 were most often found associated with IncFII(pHN7A8) plasmids. Of the 34 unique countries which contained any of the IS elements, ISKpn25 was identified from 26. ISKpn26, ISKpn14, and IS903B ISs were identified from 89.3%, 44.9%, and 23.9% plasmid samples from China. Plasmids carrying ISKpn25, ISKpn14, and ISKpn26 IS have a 4.6‐, 6.0‐, and 6.6‐fold higher carbapenemase gene count, respectively, relative to IS903B‐carrying plasmids. IS903B bearing plasmids have a 20‐, 5‐, and 5‐fold higher environmental source isolation count relative to ISKpn25, ISKpn14, and ISKpn26 bearing plasmids. ISKpn25 present on IncFIB(pQil) sourced from clinical settings is established across multiple countries, while ISKpn26, ISKpn14, and IS903B appear most often in China. Carbapenemase presence in tandem with IS elements may help promote an extensively drug resistant profile in K. pneumoniae limiting already narrow treatment options.


| INTRODUCTION
Colistin serves as a last-resort antibiotic choice for the treatment of bacterial infections caused by carbapenemase producing Klebsiella pneumoniae (CPKP) and other Gram-negative isolates. Rising colistin use has seen a corresponding increase in colistin resistance, especially during therapy and is an emerging global threat. From 2011From to 2015 Italy, the rate of colistin resistance in Klebsiella pneumoniae increased from 36% to 50% (Giani et al., 2015). Separately, in Thailand, colistin resistance has been reported at 76.1% from a sample of 280 K. pneumoniae clinical isolates collected from 2014 to 2017 (Eiamphungporn et al., 2018). Notably, colistin resistance due to disruption of the chromosomal mgrB gene in K. pneumoniae via the integration of insertion sequences (IS) has been widely reported from countries including: Lao PDR, Thailand, Nigeria, and France (Olaitan et al., 2014), Italy (Esposito et al., 2018), Greece (Giordano et al., 2018;Hamel et al., 2020;Zhu et al., 2019), Tunisia (Jaidane et al., 2018), Saudi Arabia (Zaman et al., 2018), Oman (Al-Farsi et al., 2019), Israel (Lalaoui et al., 2019), India , Taiwan (Yang et al., 2020), USA (Macesic et al., 2019), and Malaysia (Yap et al., 2020). Following membrane destabilization, colistin binds to the lipid A moiety of LPS causing the derangement of the outer membrane (Falagas et al., 2005). Colistin resistance in K. pneumoniae is mediated by the modification of the LPS through the addition of 4-amino-4-deoxy-Larabinose (L-Ara4N) to the phosphate groups of the lipid A moiety.
L-Ara4N addition to LPS attenuates the affinity of colistin to LPS targets (Helander et al., 1996). The L-Ara4N induced modification of LPS is controlled by the products of the pmrHFIJKLM operon, positively regulated by the two component PhoQ/PhoP and PmrAB systems.
MgrB, a product of the mgrB gene is a small transmembrane regulatory protein synthesized following the activation of the PhoQ/PhoP signaling cascade. MgrB interacts with the PhoQ sensor kinase exerting negative feedback on the PhoQ/PhoP system (Lippa & Goulian, 2009). Insertional inactivation of mgrB prevents the downregulation of the PhoQ/PhoP systems and represents a mechanism facilitating de novo acquired colistin resistance (Cannatelli et al., 2013).
Resistance to colistin frequently arises via the disruption of the mgrB gene by ISs in K. pneumoniae. IS-mediated mgrB gene disruption can represent a significant cause of colistin induced resistance in K.
De novo colistin resistance may occur through the transposition of ISs from plasmids into chromosomal colistin-associated genes.
Across a sample of 29 and 19 colistin resistant K. pneumoniae isolates from Italy and Greece, 2 clonally related clusters; 2 Italian ST512 isolates, and 8 Greek ST258 K. pneumoniae isolates harbored the complete copy of ISKpn25 inserted at nucleotide position 133 of the mgrB gene (Giordano et al., 2018). The same ISKpn25 was located on pKpQIL-like plasmids from these samples, indeed the ISKpn25 on pKpQIL plasmids and the ISKpn25 inserted into the mgrB gene share a 100% match between the 8154 nucleotides. Furthermore, in two ST258 and two ST512 K. pneumoniae isolates from Greece and Italy, the IS5-element derived from pKpQIL-like plasmids, was found inserted into nucleotide position 75 of the mgrB gene (Giordano et al., 2018). Nucleotide position 75 represents a hotspot for IS5 element insertion among clonally unrelated K. pneumoniae isolates (Cannatelli et al., 2013;Poirel et al., 2014). Mobilization of IS5 elements from plasmids into the mgrB gene of colistin resistant K. pneumoniae has been speculated owing to the similarity between the IS5 element inserted into mgrB and the IS5 element present on K. pneumoniae carbapenemase (KPC)-encoding and other Gram-negative bacterial plasmids (Azam et al., 2021;Cannatelli et al., 2013;Hala et al., 2019), and endogenous presence of IS5-like elements in the genome of colistin resistant K. pneumoniae (Choi & Ko, 2020).
Further evidence for the involvement of plasmids as donors for IS elements is provided by a recent cloning assay. IS elements including ISKpn26, ISKpn14, and IS903B cloned into a plasmid vector and transformed into a colistin susceptible K. pnuemoniae isolate increased the frequency of colistin resistance K. pneumoniae isolates.
Notably, for the plasmid vector carrying IS903B, colistin induced stress was responsible for IS mobilization (Yang et al., 2020).
Furthermore, a Caenorhabditis elegans killing assay model revealed nematodes fed with K. pneumoniae isolates harboring plasmids carrying ISKpn26 were associated with a significantly reduced lifespan and higher death risk during colistin treatment relative to nematodes fed with a non-IS plasmid carrying Escherichia coli and K. pneumoniae isolates, thus confirming the role of plasmids as donors for IS elements mediating colistin resistance (Yang et al., 2020).
Beyond gene insertion, IS elements can integrate into the promoter regions of chromosomal colistin-resistance associated genes.
The IS1R IS element sourced from a plasmid has been shown to integrate into the promoter region of mgrB mediating the emergence of a de novo colistin resistant phenotype (Antonelli et al., 2017), indeed IS1-like element insertion has been independently reported repeatedly in the promoter region of mgrB (Berglund et al., 2018;Haeili et al., 2017). For IS1382-like and IS1-like elements disrupting mgrB in K. pneumoniae, BLASTn searches revealed their presence only in respective isolates with disruptions in mgrB, notably absent from any other chromosomal location, thereby indicating a likely plasmid source as opposed to transposition from a chromosomal source (Jaidane et al., 2018).
Colistin resistance may arise through the transposition of ISs from source IS containing plasmids preferentially targeting specific mgrB regions for recombination. The dissemination of IS elements that transpose into the same position of the mgrB gene may represent a mechanism mediating the observed colistin resistance in both clonally related and unrelated K. pneumoniae isolates. IS elements are frequently reported in K. pneumoniae colistin-associated chromosomal resistant genes. Examples of ISKpn25, ISKpn26, ISKpn14, and IS903B inserted into the mgrB gene of K. pneumoniae are shown in Table A1. To ascertain IS element prevalence in K. pneumoniae, IS elements which have been shown to disrupt the mgrB gene; namely ISKpn25, ISKpn26, ISKpn14, and IS903B were investigated. Species IS prevalence among 1000 BLASTn hits was assessed. Separately, to support pathogen surveillance, plasmid incompatibility typing (Inc) for the top 120 BLASTn plasmid hits was performed to determine the respective plasmid incompatibility group associated with each IS element. In addition, metadata for the same plasmid samples were gathered to determine dual carbapenemase gene prevalence, plasmid size, species, country, and isolation source to elucidate the emerging clinical threat posed by plasmid bearing IS elements.
Each IS element was blasted using the NCBI nucleotide blast online tool, BLASTn. To assess IS proportional assignment between bacterial species, the max target sequences parameter was increased from the default of 100-1000 for each IS element. The description table for each BLASTn hit was downloaded and filtered for plasmids to ascertain plasmid samples encoding IS elements. For ISKpn25 which returned many low identity and coverage hits, plasmid hits above a minimum threshold; percentage identity ≥95% was used. This threshold was chosen to differentiate ISKpn25 from ISKpn26 which shares 91% nucleotide identity (https://www-is.biotoul.fr). The relative proportion of IS elements among K. pneumoniae samples versus other bacterial species was calculated and plotted.
For each IS element, the top 120 nonduplicate circularized plasmid hits harboring IS elements were downloaded, and the Fasta contigs incompatibility (Inc) typed with PlasmidFinder using a minimum identity and length of 98% (Carattoli & Hasman, 2019). Each plasmid contig carrying IS elements was also scanned for carbapenemase genes with a predicted in silico resistance phenotype to the carbapenem, meropenem and other resistant genes using ResFinder 4.0 with a minimum identity and length of 98% (Zankari et al., 2012).
Metadata pertaining to each sample, including accession number, species and isolation source, country of origin, and plasmid size was  sources. ISKpn26, ISKpn25, and ISKpn14 are however sourced from clinical sources, 20.75-, 98-, and 107-fold higher relative to IS903B.
These relationships are summarized in Table 2d.
Moreover, the distribution of all resistance genes among the four IS elements was analyzed using the χ 2 test of independence. For all four IS elements, the number of plasmids encoding either 0, ≥2, ≥5, and ≥10 resistance genes was counted. A significant difference between the distribution of the total number of resistance genes among plasmids between the four IS elements was determined, χ 2 (6, 11.0% (n = 11/120), and 5.83% (n = 7/120) for ISKpn14, ISkpn26, and ISkpn25, respectively. Furthermore, across the four IS elements, resistance to a total of 23 antibiotics was detected. Notably, across the 23 antibiotics, IS903B had the lowest predicted in silico resistance for 56.52% (n = 13/23) of the antibiotics ( Figure A1b).

| DISCUSSION
K. pneumoniae represents a key reservoir species harboring plasmids encoding IS elements which could disrupt the mgrB gene leading to colistin resistance. The high prevalence of these four IS elements from plasmids derived from K. pneumoniae provides indirect evidence for their role in generating colistin resistance. The ISKpn14 reference element inserted into the chromosomal mgrB gene in a K. pneumoniae isolate (accession: KJ129604.1) has a 100% nucleotide identity and coverage match with the ISKpn14 element found on a wide range of plasmids, indicating the role of plasmids as possible donors for ISKpn14 IS elements. This relationship is depicted in Figure A3.
Moreover, ISKpn25 (ISL3) encoding pKpQil-like plasmids have been proposed as the donor for ISKpn25 elements. A 100% identity match between the ISKpn25 found on plasmids and the ISKpn25 disrupting the chromosomal mgrB gene has been reported across two K.
pneumoniae STs: ST258, and ST512 in strains carrying both pKpQil-like plasmids and a disrupted chromosomal mgrB gene across independent investigation of colistin resistant K. pneumoniae (Cienfuegos-Gallet et al., 2017;Giordano et al., 2018). ISKpn26 and IS903B have been found inserted into the mgrB gene of K. pneumoniae (Nirwan et al., 2021;Silva et al., 2021). These IS share high identity to their respective reference IS elements identified on plasmids. Furthermore, multiple IS insertion sites in mgrB (Zaman et al., 2018), coupled with the experimentally determined inducible colistin resistance arising from bacterial cells transformed with plasmids containing IS elements (Yang et al., 2020) (Berglund et al., 2018;Yan et al., 2021;Yang et al., 2020). Whole genome sequencing (WGS) is being increasingly employed in bacterial genomics to support pathogen surveillance. The results indicate plasmid Inc groups and the country of origin for particular mgrB disrupting IS elements. Moreover, the relationship between IS elements and the cocarriage of carbapenemase genes and respective isolation source is exposed. Importantly however, other IS elements can also disrupt the mgrB gene including IS10R, ISEcp1, and ISKpn28 (Jayol et al., 2015;Yang et al., 2020;Zaman et al., 2018). Furthermore, IS elements can disrupt other chromosomal colistin-associated genes including crrCAB, giving rise to colistin resistant K. pneumoniae (Yang et al., 2020).
Functional studies involving IS elements in plasmid vectors transformed into colistin susceptible K. pneumoniae and other Enterobacteriaceae isolates will also more fully reveal the role of IS-carrying plasmids in the induction of colistin resistance. The discovery of the cocarriage of carbapenemase genes with IS elements in clinical samples indicates K. pnuemoniae strains are primed for resistance towards last-resort antibiotics, limiting already narrow therapeutic treatment options.

ACKNOWLEDGMENTS
There are no funders to declare.

CONFLICT OF INTERESTS
None declared.

ETHICS STATEMENT
None required.

DATA AVAILABILITY STATEMENT
All data are provided in this article and its appendices, as well as in supplementary material available at https://doi.org/10.5281/zenodo.