The Current Burden of Carbapenemases: Review of Significant Properties and Dissemination among Gram-Negative Bacteria.

Carbapenemases are β-lactamases belonging to different Ambler classes (A, B, D) and can be encoded by both chromosomal and plasmid-mediated genes. These enzymes represent the most potent β-lactamases, which hydrolyze a broad variety of β-lactams, including carbapenems, cephalosporins, penicillin, and aztreonam. The major issues associated with carbapenemase production are clinical due to compromising the activity of the last resort antibiotics used for treating serious infections, and epidemiological due to their dissemination into various bacteria across almost all geographic regions. Carbapenemase-producing Enterobacteriaceae have received more attention upon their first report in the early 1990s. Currently, there is increased awareness of the impact of nonfermenting bacteria, such as Acinetobacter baumannii and Pseudomonas aeruginosa, as well as other Gram-negative bacteria that are carbapenemase-producers. Outside the scope of clinical importance, carbapenemases are also detected in bacteria from environmental and zoonotic niches, which raises greater concerns over their prevalence, and the need for public health measures to control consequences of their propagation. The aims of the current review are to define and categorize the different families of carbapenemases, and to overview the main lines of their spread across different bacterial groups.


Introduction and Historical Perspective of Carbapenemases
Since the discovery of penicillin early in the previous century, Gram-negative bacteria have become proficient at evading the bactericidal activity of β-lactam antibiotics, principally through production of β-lactamases. Over the past decades, the emergence and dissemination of bacterial pathogens that are resistant to carbapenems, which are the broadest spectrum agents of the β-lactam group, has become apparent as a worldwide public health issue. The increasing prevalence of such organisms threatens to restrain treatment options after compromising carbapenems, which is often regarded as "last resort" antimicrobials in hospitals and long-term care facilities [1]. A primary mechanism of carbapenem resistance in Gram-negative bacteria is the production of acquired carbapenemases, which are β-lactamases with the widest spectrum of activity [2]. In addition to hydrolyzing carbapenems, these enzymes are active against most other members of the β-lactam group with few exceptions. The major drive behind the emergence of carbapenemases was the widespread use of carbapenems in treating serious infections caused by extended-spectrum β-lactamase (ESBL) producing-pathogens [3]. Such selection pressure on bacterial pathogens has rendered these pathogens carbapenem-resistant viewpoints, with emphasis on their molecular and genetic properties, as well as their species and geographical distribution.

Classification
Like other β-lactamases, carbapenemases can be classified according to two possible schemes: functional and molecular. The functional scheme, known as the Bush-Jacoby-Medeiros classification, is a biochemical scheme based on properties like isoelectric points, substrate profiles and inhibitor characteristics [36]. Accordingly, β-lactamases are functionally classified into groups 1-4 with many subgroups under group 2. This classification has continuously matured over the years [37] and was updated in 2010 in which several new subgroups of each of the major groups were described [38]. According to this last update, group 1 includes cephalosporinases, which are more active on cephalosporins than benzylpenicillin and are usually resistant to inhibition by clavulanic acid. They also have high affinity to aztreonam [39]. Group 2 includes the largest group of β-lactamases. In this group, subgroup 2a includes penicillinases predominantly present in Gram-positive cocci [38], while subgroup 2b includes β-lactamases that readily hydrolyze penicillin and early cephalosporins, and are strongly inhibited by clavulanic acid and tazobactam. Subgroup 2c encompasses penicillinases that are characterized functionally by the ability to hydrolyze carbenicillin or ticarcillin more than benzylpenicillin, and by being easily inhibited by clavulanic acid or tazobactam [36]. Subgroup 2d includes enzymes able to hydrolyze cloxacillin or oxacillin at a rate of 50% greater than that for benzylpenicillin and, hence, are known as OXA enzymes [38]. Subgroup 2e includes cephalosporinases that have the ability to hydrolyze extended-spectrum cephalosporins and are inhibited by clavulanic acid or tazobactam. They can be differentiated from group 1 enzymes by their poor affinity for aztreonam [40]. Subgroup 2f includes serine carbapenemases that can be inhibited by tazobactam better than by clavulanic acid [8], while group 3 includes metallo-β-lactamases distinguished by their zinc ion requirement at the active site. The last updated classification scheme did omit group 4 β-lactamases previously included in the functional classification, as they most likely would be included in one of the enzyme groups 1 to 3, until more information about them becomes available [38]. As for carbapenemases, these mainly fall under functional groups 2d, 2f, or 3 [1].
According to the more commonly used Ambler classification system, β-lactamases are categorized into molecular classes A, C, and D, which include the β-lactamases with serine at their active site, whereas molecular class B are all metalloenzymes with a zinc active site. This structural classification is based on amino acid homology but lacks detail concerning enzymatic activity [8]. In this system, carbapenemases are categorized as classes A, B, and D. Rare carbapenemases belonging to Ambler class C exist, as described below. These usually cause reduced susceptibility to carbapenems as a consequence of the low enzyme's catalytic efficiency and a permeability defect in the bacterial cell [35]. The description below shows molecular properties, inhibitors, genetic background, and major examples of each of the Ambler classes of carbapenemases. Examples of major carbapenemase groups with data on their original discovery and genetic location are listed in Table 1.

Ambler Class A Carbapenemases
and Dubai [67] revealed GES-5, while P. aeruginosa from Mexico revealed GES-20 [68]. In addition, Moubareck and Colleagues identified GES-11 in an isolate of A. baumannii in France [69], and the same enzyme was identified in nosocomial isolates of this organism from Lebanon [70]. In spite of the fact that they are plasmid mediated, GES carbapenemases do not have aptitude for efficacious dissemination like KPC [71].

Ambler Class B Carbapenemases (Metallo-β-Lactamases)
Unlike the serine-dependent carbapenemases belonging to classes A and D, class B carbapenemases are MBLs that require a heavy metal like zinc for catalysis. MBLs have a broad substrate spectrum and can catalyze hydrolysis of virtually all β-lactam antibiotics including carbapenems with the exception of monobactams, such as aztreonam [72]. Sophisticated modeling analyses show that the active site of these enzymes is defined by a peculiar set of amino acids and contains one or more bounded zinc ions that are central to the catalytic mechanism. These ions usually coordinate two water molecules necessary for hydrolysis, and the active site orients and polarizes the β-lactam bond to facilitate nucleophilic attack by zinc-bound water/hydroxides. The wide, plastic, active-site groove can accommodate most β-lactam substrates, which explains the broad spectrum of activity [73]. Because MBLs are metalloenzymes, they are resistant to the commercially available β-lactamase inhibitors but are susceptible to inhibition by metal ion chelators like ethylenediaminetetraacetic acid (EDTA) [8].
Originally, the first MBLs were ubiquitous chromosomal enzymes identified more than 50 years ago in environmental and opportunistic pathogenic bacteria such as B. cereus, Aeromonas spp., Legionella gormanii, Pseudomonas stutzeri, Shewanella spp., and S. maltophilia [8,73,74]. With the exception of S. maltophilia, these bacteria have not been frequently associated with serious nosocomial infections, which are generally opportunistic pathogens. The chromosomal metalloenzymes are not easily transferrable and have not extensively contributed to an epidemiological burden of MBLs. Perhaps one exception to such chromosomal MBLs that represents a transferrable group is that from Bacteroides fragilis. This anaerobe is relatively resistant to β-lactams, and its MBL. Designated CfiA was first genetically characterized in 1990 and is one of the most intensely studied with respect to a mechanism of catalysis and structure-function properties [75]. Insertion elements, such as IS942, IS1186, and IS4351, have been identified immediately upstream of the ribosome-binding site of the gene encoding CfiA, which provides enhanced transcriptional capabilities for the gene [76,77]. Currently, the most common MBL families include the IMP, VIM, NDM, GIM, and SIM enzymes, which are genetically located within a variety of integrons, where they have been incorporated as gene cassettes. When these integrons become associated with plasmids or transposons, transfer between bacteria is readily facilitated [78]. MBLs are currently found in different Gram-negative bacterial species, and their presence is often associated with resistance to antibiotic classes, which results in multidrug resistance and comprises treatment options [79].
The first transferrable imipenem resistance through MBLs was reported from P. aeruginosa in 1991 in Japan [13], was designated IMP-1, and its gene was located on a conjugative plasmid conferring resistance to β-lactams, gentamicin, and sulfonamide. This plasmid was transferable by conjugation to P. aeruginosa but not to Escherichia coli. Soon, the same enzyme was identified in S. marcescens in Japan as well, and the area of Southeastern Asia remained, until today, the greatest reservoir of IMP-type enzymes [73]. Several years later, the same enzyme appeared in Italy in A. baumannii, representing its first description from European countries [80]. Additionally, IMP-2 from Italy and IMP-5 from Portugal were soon described [81] as well as IMP-4 from USA [82], both IMP-1 and 2 from the Middle East [83], and IMP-7 from Australia [84], which indicates that IMP genes were not solely restricted to the Far East region. Currently, the number of IMP varieties have reached about 50. Analysis of the genetic platform of bla IMP genes has shown that most of them are embedded as gene cassettes in class I integrons, which harbors other resistance genes, such as aac (mediating resistance to aminoglycosides), bla OXA (different serine oxacillinases, described shortly), and genes conferring resistance to antiseptics, or to chloramphenicol [8,78]. Apart from detection of bla IMP in clinical isolates, zoonotic IMP-4 was described in Salmonella enterica serovar Typhimurium from companion cats in Australia [85]. Likewise, IMP-27 was isolated from environmental and fecal samples of Enterobacteriaceae recovered from swine operation farms in the US [86]. Therefore, IMP carbapenemases may play a role in a dissemination cycle that encompasses humans, animals, and the environment.
The second dominant group of MBLs are the VIM (Verona intergon-encoded MBL) category, so named due to initial discovery of VIM-1 in a P. aeruginosa isolate from Verona, Italy, in 1997 [16]. Similar to bla IMP , bla VIM was also carried on a gene cassette inserted into a class 1 integron, and, when cloned into E. coli, resulted in a significant decrease in susceptibility to a broad array of beta-lactams (ampicillin, carbenicillin, piperacillin, mezlocillin, cefotaxime, cefoxitin, ceftazidime, cefoperazone, cefepime, and carbapenems) [16]. Soon afterward, VIM-2 was isolated in Marseille, France [87] from P. aeruginosa and shared 90% similarity with VIM-1. Although VIM-1 and VIM-2 have been identified in enterobacterial species. P. aeruginosa remains the most important known reservoir of these enzymes [73]. There are more than 40 allelic variants of VIM enzymes reported so far, and they mainly belong to three phylogenetic clusters: VIM-1-like, VIM-2-like, and VIM-7-like enzymes [88]. VIM-2-like enzymes have been associated mostly with P. aeruginosa, whereas VIM-1-like enzymes, in particular, VIM-4, have been reported in Enterobacteriaceae. The latest VIM-type to be fully characterized is VIM-7, which has been characterized from a carbapenem-resistant P. aeruginosa isolate from Houston, Texas [89]. It shares only a 77% identity with VIM-1 and 74% with VIM-2, and lies on a 24-kb plasmid, which can be readily transferred into Enterobacteriaceae and other pseudomonads, and was thought to originate from a different ancestral source. After original detection in North America, VIM-7 was recently described in another report from Brazil [90]. Our group has lately detected several VIM enzymes including VIM-2, VIM-30, VIM-31, and VIM-42 from nosocomial P. aeruginosa in Dubai, United Arab Emirates [67], with VIM-2 co-existing with the Ambler class A carbapenemase GES-5. Moreover, in Lebanon, we detected VIM-2 in 16% of a nationwide collection of nosocomial P. aeruginosa that were primarily spread among different hospitals by clonal dissemination [83]. The bla VIM genes are typically embedded in class 1 integrons, which can be incorporated in either plasmids or chromosomes. Plasmids carrying bla VIM genes in Enterobacterriaceae belong, most frequently, to the IncA/C or IncN group [91].
The spectacular global spread of NDM remains one of the most worrisome antibiotic resistance events caused by MBLs. The initial enzyme, NDM-1, was originally isolated from a carbapenem-resistant K. pneumoniae recovered from urine of a Swedish patient who had recently travelled to New Delhi [31]. NDM-1 was transferrable and shared little identity with other MBLs. The most similar MBLs were VIM-1/VIM-2 with which it has only 32.4% similarity, and can hydrolyze all β-lactams except aztreonam. Compared to VIM-2, NDM-1 displays tighter binding to most cephalosporins, such as cefuroxime, cefotaxime, and cephalothin (cefalotin), and also to penicillin [92]. By mid-2010, the NDM-1 gene may have been acquired by bacteria from the Indian subcontinent and introduced to other countries including European countries and the United States through tourists travelling around the globe. Yet, the Indian subcontinent remains as the principle reservoir [93]. A second probable epicenter seems to be located in the central Balkans without a clear connection to that of India [35]. NDM-genes are dominant in K. pneumoniae and E. coli isolates with certain sequence types (for K. pneumoniae, ST11, ST14, ST15, or ST147; for E. coli, ST167, ST410, or ST617) being the most prevalent [94]. However, variants of NDM have also been found in association with A. baumannii and P. aeruginosa [95]. NDM has been detected on surveillance specimens from hospital equipment [96], in environmental samples [97], in poultry farms [98], and from injured war victims in the Middle East [99].
As of 2020, 24 NDM variants have been identified in >60 species of 11 bacterial families, and several variants have enhanced carbapenemase activity with most bla NDM -carrying plasmids belonging to limited replicon types (IncX3, IncFII, or IncC) [94]. Examples of such variants are NDM-2 [100], NDM-3 [101], NDM-4 [102], and NDM-5 [103]. While commonly used phenotypic tests cannot reliably identify NDM, immunoassays can specifically detect these enzymes [104,105], and molecular approaches remain as the reference methods [106]. NDM will remain a severe challenge in health care settings, and more studies on appropriate countermeasures are required.
In 2002, Toleman and Colleagues described a novel MBL from a P. aeruginosa isolate in Sao Paulo, Brazil, distinctly different from VIM and IMP and, accordingly, represents a new subfamily. The enzyme was termed SPM-1 (Sao Paulo MBL-1) [107]. The SPM-1 gene was detected on a transmissible plasmid, and the enzyme contains the classic MBL zinc-binding motif and shows the highest identity (35.5%) to IMP-1. SPM-1 has been repeatedly detected in South America from P. aeruginosa [108][109][110], and has bypassed nosocomial prevalence to animals where it was recently isolated from P. aeruginosa existing as a part of normal microbiota of birds [111]. Sporadic reports of SPM-1 exist for P. aeruginosa from UK [112] and Switzerland [113] and A. baumannii from Iran [114]. Although some expert opinions have suggested the probability of wider spread by SPM [115], this carbapenemase is still mainly prominent in South American hospital outbreaks. The bla SPM gene can be either chromosomal or plasmid-encoded, and is associated with the insertion sequence ISCR4 (for common region 4) [116]. These elements are usually harbored in transposons and/or plasmids, which forms mobile vesicles for horizontal transfer of captured MBL genes between bacteria [117].
Other MBL-encoding genes include GIM (German imipenemase) described in Dusseldorf, Germany, from P. aeruginosa [118] on a 22-kb nontransferable plasmid. Unlike other class B carbapenemases, GIM has two zinc ions in the active site, but appears to be a weaker enzyme [73]. GIM has been lately identified in A. baumannii from Indian patients with severe urinary tract infection, sometimes concurrently with VIM [119]. Another MBL, SIM-1 (Seoul imipenemase-1) was isolated in Korea [120] and was carried on a gene cassette inserted into a class 1 integron, found in A. baumannii strains isolated in sputum and the urine of patients with pneumonia and urinary tract infections, respectively. In 2019, the spread of bla SIM-1 to Enterobacteriaceae has been suggested after a report from China indicating a mega-plasmid harboring this gene from a clinical K. pneumoniae isolate [121]. More recently, described MBLs include DIM-1 (Dutch imipenemase-1) [122], KHM-1 (Kyorin University Hospital MBL-1) [123], and TMB-1 (Triploi MBL-1) [124] that have been reported on occasional encounters. Future investigations are needed to unravel the epidemiology and propagation potential of such MBLs.

Ambler Class D Carbapenemases (Oxacillinases)
Among the earliest β-lactamases detected, class D β-lactamases were relatively rare and always plasmid mediated [125]. They were also referred to as oxacillinases because they commonly hydrolyze isoxazolylpenicillins such as oxacillin, methicillin, and cloxacillin much faster than classical penicillin, such as benzylpenicillin, and are relatively less effective against first-generation cephalosporins. The designation, OXA, thus, refers to the preferred substrate, which is oxacillin [126]. The active site of these enzymes includes a highly conserved serine-based structure, even though the rest of the molecule shows variability in the amino acid sequences, and they are characteristically not inhibited by β-lactamase inhibitors like clavulanate, sulbactam, tazobactam, cloxacillin, or metal chelators like EDTA [1]. The emergence of such enzymes presumably coincided with the widespread introduction of flucloxacillin and methicillin for treating staphylococcal infections. The early OXA β-lactamases such as OXA-1, OXA-2, and OXA-3 were plasmid-encoded and identified in Gram-negative bacteria. They were essentially penicillanases but hydrolyzed oxacillin is better than penicillin [127]. Later, OXA-11, which is the first extended-spectrum OXA variant, was isolated from P. aeruginosa [128], and exhibited a transferrable resistance profile with enhanced ceftazidime hydrolysis. Soon, other extended-spectrum OXA enzymes appeared like OXA-13, OXA-14, OXA-15, OXA-16, OXA-17, OXA-19, OXA-28, and OXA-45, which all remained confined to P. aeruginosa, and did not seem to be spreading [8,125]. Currently, OXA enzymes with carbapenem-hydrolyzing activity mainly include, among others, the groups OXA-23-like, OXA-24/40-like, OXA-48-like, OXA-58-like, OXA-143-like, and OXA-235.
The first carbapenem-resistant OXA-type β-lactamase, OXA-23, was identified on a large plasmid of a multidrug resistant (MDR)-A. baumannii collected from the blood of a patient hospitalized at the Edinburgh Royal Infirmary, Scotland, in 1985 [29], which is the same year when imipenem was approved for clinical use. After the sequence for this enzyme was published in the year 2000 [30], several bla OXA-23-like genes (bla OXA-23 , bla OXA-102 , bla OXA-103 , bla OXA-105 , bla OXA-133 , and bla OXA-134 ) were discovered on the chromosome of Acinetobacter radioresistens. This is a commensal bacterial species that resides on the skin of hospitalized and healthy individuals, and indicates that this species may be the probable natural reservoir of these enzymes [129]. Hydrolytic activity measurements done on OXA-23-like enzymes including OXA-23, OXA-27, and OXA-146 show considerable kinetic variation. However, the enzymes are able to hydrolyze oxyiminocephalosporins, aminopenicillins, piperacillin, oxacillin, and aztreonam in addition to carbapenems [130]. Among the carbapenems, OXA-23 has a much higher turnover rate for imipenem than for meropenem, ertapenem, or doripenem [131]. Despite the relatively low turnover rates for carbapenems displayed by these enzymes, the production of OXA-23 by an A. baumannii strain is enough to increase the MIC breakpoint for considering strains to be resistant. However, when OXA-23 is produced in a strain that also expresses the AdeABC efflux pump, the MICs are significantly elevated. This indicates that, unlike some of the other OXA-type carbapenemases, strains do not require other resistance mechanisms to work in synergy with OXA-23 to be carbapenem resistant. However, high resistance levels are achieved only when there are other mechanisms present [125]. The bla OXA-23 genes are usually located on plasmids and are flanked by two copies of the insertion sequences ISAba1 in opposite directions on the transmissible transposons Tn2006 and Tn2008. They can also exist in association with the transposon Tn2007, which lacks the second copy of ISAba1, and is associated instead with one copy of ISAba4 [132]. The transposon Tn2009 has been detected in isolates from China [133,134], while Tn2006 is the most commonly observed worldwide. The insertion sequences are likely to act as strong promotors for expression of OXA-23, which enhances carbapenem resistance [135]. Besides the location on plasmids, evidence shows chromosomal insertions of bla OXA-23 associated with transposon-mediated transmission [136].
More than 10 years after the initial discovery of OXA-23, a newly isolated enzyme named OXA-24, which was chromosomally encoded, was described [137]. It increased carbapenem MIC by four-folds, hydrolyzed benzylpenicillin, and was inhibited by chloride ions. OXA-24 was subsequently named OXA-40, and additional members of the group, namely OXA-25, OXA-26, OXA-72, and OXA-160 were subsequently discovered [125]. While the OXA-24 group is documented in A. baumannii in reports from certain countries like Taiwan [138], Thailand [139], Bulgaria [140], and Lebanon [141], it was also detected in P. aeruginosa [142] in one French study. It was previously thought that bla OXA-24-like genes are mainly located on chromosomes and no mobile elements are associated. This does not definitely represent the current status of these genes. Recently, plasmids carrying OXA-24-like enzymes were detected in cattle [143] and in a domestic grey parrot presented to a veterinary clinic in Luxembourg [144]. This shows that some lineages of A. baumannii have successfully acquired OXA-24/40-encoding plasmids that emerged and started to appear in reservoirs including livestock, companion animals, and the environment.
In 2005, a carbapenem-hydrolyzing oxacillinase, OXA-58, was isolated from A. baumannii in Toulouse, France. This enzyme had only 50% amino acid similarity to other oxacillinases. It hydrolyzed penicillins, oxacillin, and imipenem but not expanded-spectrum cephalosporins. Its gene was plasmid-carried, and was found to be located within a transposon and bracketed by two copies of ISAba3 [145]. OXA-58 was reported in studies from Iran [146], Greece [147], Argentina [148], and Lebanon [149] in A. baumannii isolates. Although no more than three variants of this enzyme have been detected so far, including OXA-96, OXA-97, and OXA-164 [125], certain new observations may alter the known consensus for this enzyme. For example, although OXA-58 represents a class D carbapenemase that is extremely rare in Entereobacteriaceae, it has been reported in isolates of Proteus mirabilis in Belgium [150], in Germany [151], and, most recently, in Poland [152]. Mobilized plasmids allowing horizontal spread of the bla OXA-58 gene are suggested to allow its transfer among P. mirabilis, and the occurrence of these plasmids might suggest hidden dissemination of unknown scale and future potential. A second observation about OXA-58 comes from China, where a recent report discovered bla OXA-58 and bla NDM-1 was carried simultaneously on the same plasmid in Acinetobacter pitti, which is an emerging opportunistic nosocomial MDR pathogen [153]. This indicates the need for further investigations in the context of genomic epidemiological characteristics of OXA-58 as well as its clinical and microbiological significance on carbapenem resistance in Acinetobacter species. A third observation is the detection of OXA-58 in Acinetobacter towneri and in 9 environmental genera of bacteria from hospital sewage [154] and coastal water [155], respectively.
In 2004, a novel carbapenemase gene was detected in a carbapenem-resistant A. baumannii strain in Brazil [156]. The gene encoded a carbapenem-hydrolyzing class D beta-lactamase known as OXA-143, with 88% amino acid sequence identity with OXA-40, 63% identity with OXA-23, and 52% identity with OXA-58. It hydrolyzed penicillins, oxacillin, meropenem, and imipenem but did not expand spectrum cephalosporins. The gene was plasmid encoded but associated with neither insertion sequences nor integron structures. However, it was bracketed by similar replicase-encoding genes at both ends, which suggests acquisition through a homologous recombination process. Further identified variants of the OXA-143 group include OXA-182, OXA-231, OXA-253, and OXA-255. OXA-143 is still detected in Brazil [157,158], and, although believed that this country is the major epicenter, OXA-143 was detected in 2017 for the first time in 14% of A. baumannii isolates from teaching hospitals in the central part of Iran [159]. The variant OXA-231 was also isolated from a nosocomial A. baumannii isolate recovered from urine of a female patient, at Londrina, Brazil. This variant had reduced catalytic efficiency against carbapenems and noticeably increased the specificity for oxacillin [160].
In 2013, Higgins et al [161] reported three novel OXA enzymes, which include OXA-235 and its amino acid variants OXA-236 and OXA-237 in A. baumannii isolates from the United States and Mexico. The expression of OXA-235 resulted in reduced carbapenem susceptibility, while cephalosporin MICs were unaffected. Genetic analysis revealed that genes encoding the novel OXA variants were bracketed between two ISAba1 insertion sequences. Recently, results from the Canadian Nosocomial Infection Surveillance Program reported OXA-235 to account for almost half of all carbapenemases detected among Acinetobacter species from Canadian hospitals [162]. In a remarkable jump toward natural reservoirs, the variant OXA-278 of OXA-235 was detected in 2019 in Acinetobacter lwoffii from municipal wastewater treatment plants in Singapore [163].
Apart from the previous OXA families mostly described in Acinetobacter, OXA-48-like enzymes are particularly associated with Enterobacteriaceae, and likely designate one of the most concerning developments in carbapenem resistance in the last decade and are still globally ascending. First recovered from K. pneumoniae in Turkey [25], the OXA-48 enzyme has low-level hydrolytic activity against carbapenems, with much greater activity against imipenem than against meropenem, and results in only modest increases in MICs of the carbapenems [125]. It has very weak activity against expanded spectrum cephalosporins, and does not significantly hydrolyze ceftazidime and cefepime, but, in combination with impermeability, can lead to high-level resistance to carbapenems. OXA-48 significantly hydrolyzes only penicillin and narrow-spectrum cephalosporins. However, enteric Gram-negative bacteria with the bla OXA-48-like genes could co-harbor genes encoding ESBL (bla CTX-M , bla SHV , bla TEM ) or AmpC enzymes, or both, which confers nonsusceptibility to aztreonam, extended-spectrum cephalosporins, and carbapenem agents [1]. OXA-48 currently is widespread, not only in K. pneumoniae, but also in other Enterobacteriaceae. Turkey is reported as having the highest epidemiologic level of OXA-48, and other important reservoirs are linked to India, Middle East, and North African countries. Additionally, OXA-48 producers have been documented sporadically in several European countries, including France, Germany, Netherlands, Italy, Belgium, UK, Ireland, Slovenia, Switzerland, and Spain [33]. An "intercontinental spread" in Europe has been described, with endemic situation in certain countries [27]. In fact, the detection of OXA-48-like producers is difficult since the level of acquired resistance to carbapenems may remain very low, keeping these strains underreported. One concern for controlling the spread of OXA-48-like producers is the absence of phenotypic tests that could contribute to their laboratory recognition, since, like other class D carbapenemases, they are not inhibited by metal ion chelators or clavulanate.
The major vehicle carrying the bla OXA-48 gene has been identified. Complete sequencing of plasmid pOXA-48a carrying the gene from K. pneumoniae showed that its backbone corresponded to that of a 62.3 kb IncL/M-type plasmid in which the gene had been integrated through acquisition of the Tn1999 composite transposon harboring the upstream and downstream IS1999 insertion sequence, which acts as a promoter for bla OXA-48 gene expression [164]. Hence, the current spread of OXA-48 producers is related to the spread of such a single plasmid with a high conjugation rate among different enterobacterial isolates [165]. Nevertheless, chromosomal integration of OXA-48 was documented in E. coli from UK [166] and Egypt [167]. Several variants of OXA-48 were identified. OXA-48, OXA-181, OXA-232, OXA-204, OXA-162, OXA-163, and OXA-244, in that order, are the most common enzymes among the group [168]. Four of these, OXA-162, OXA-163, OXA-181, and OXA-232, have had their kinetic properties measured, and while OXA-163 very poorly hydrolyzes carbapenems, OXA-181 and OXA-232 appear broadly similar to OXA-48 in their activity, with OXA-232 demonstrating better hydrolysis of penicillin [125]. The OXA-181 variant, which differs from OXA-48 by four amino acid substitutions, is prominent in India and has been associated with other carbapenemase genes, including bla NDM-1 and bla VIM-5 [169]. Recently, in Tunisia, two independent research groups showed that OXA-204 has emerged in clinical E. cloacae [170], and also in Citrobacter freundii from wastewater [171]. Moreover, OXA-48-like enzymes were implicated in carbapenem resistance in wild mammals and birds from Catalonia, Spain [172] and in companion animals from Germany [173], which indicates variability of niches where such enzymes are disseminating. The respective features of each OXA-48-like variant and their respective impacts in terms of carbapenem resistance need to be continuously elucidated. Given the rapid spread of Enterobacteriaceae producing OXA-48-like enzymes in different ecosystems, and the increasing number of reservoirs for such organisms, not only in hospitals but also in the community, among animals and in the environment, the medical community should remain alert to the threats posed by these organisms. Detection should be optimized to reduce their spread.

Emerging Ambler Class C Carbapenemases
The Ambler class C β-lactamases (AmpC cephalosporinases) confer resistance to penicillin, oxyiminocephalosporins, cephamycins (cefoxitin and cefotetan), and, variably, to aztreonam [72]. An exceptional member with different properties is ACC-1, a plasmid-encoded class C β-lactamase identified in clinical isolates of K. pneumoniae, P. mirabilis, Salmonella enterica, and E. coli. Due to conformational alterations in this enzyme structure, ACC-1-producing bacteria are susceptible to cefoxitin, whereas they are resistant to oxyiminocephalosporins [174]. AmpC cephalosporinases are not significantly inhibited by other β-lactamase inhibitors such as clavulanic acid, but may be inhibited by boronic acid and cloxacillin [41]. They may be encoded by the chromosome of many Enterobacteriaceae like E. cloacae and S. marcescens, and a few non-enterobacterial organisms like P. aeruginosa. In such instances, they are inducible by antibiotics or expressed at high levels by mutation. Nevertheless, AmpC enzymes may also be acquired on transmissible plasmids, which, consequently, makes them appear in bacteria by lacking or poorly expressing a chromosomal bla AmpC gene, such as E. coli, K. pneumoniae, and P. mirabilis [39].
Studies describe only a few AmpC cepahlosporinases with carbapenemase activity. For example, plasmidic CMY-2-type, ACT-1-type, and DHA-1-type AmpC enzymes may promote the emergence of carbapenem resistance in porin-deficient clinical isolates of Enterobacteriaceae [175,176]. Very recently, Jousset et al. showed that E. cloacae can chromosomally encode an AmpC enzyme of the type ACT-28. Kinetic parameters of purified ACT-28 revealed a slightly increased imipenem hydrolysis compared to that of ACT-1 [177]. CMY-10 was the first reported carbapenemase among plasmidic class C β-lactamases, and this enzyme was also a class C ESBL with extended substrate specificity for extended-spectrum cephalosporins [178]. In 2014, ADC-68 was reported in A. baumannii from Korea as a chromosomal class C β-lactamases that possesses class C extended-spectrum β-lactamase and carbapenemase activities [179]. Many reports showed that such rare AmpC with carbapenemase activity may aggravate carbapenem resistance when coupled with outer membrane permeability and/or efflux pump overproduction [70,176].

Distribution of Carbapenemases Among Gram-Negative Pathogens
The section below describes the carbapenemase groups in Gram-negative bacteria including Enterobacteriaceae, Pseudomonas, Acinetobacter, and others. A world map showing geographical distribution and endemicity of major carbapenemases is shown in Figure 1.

Enterobacteriaceae
Members of the family Enterobacteriaceae, such as E. coli, are normal inhabitants of the intestinal tract in humans and animals, and are commonly isolated in clinical cultures. They are the causative agent of several types of infections in humans, including respiratory tract infections, urinary tract infections, and bloodstream infections in hospitalized or otherwise immunocompromised subjects. In the context of antimicrobial resistance, Enterobacteriaceae are especially important as they are a cause of community-associated as well as healthcare-associated infections, which generates a major clinical and public health challenge [180]. The overdependence on carbapenems as empiric treatment in the management of infections caused by Enterobacteriaeceae was driven by cephalosporin resistance in this family due to production of ESBLs [181]. Carbapenems proved especially useful since ESBL producers are also resistant to other antibiotic classes including aminoglycosides, tetracyclines, and fluoroquinolones [71]. Unfortunately, this has led to the emergence of carbapenem-resistant Enterobacteriaceae (CRE), defined as bacteria belonging to the Enterobacteriaceae family that have the ability to survive and grow in the presence of clinically relevant concentrations of carbapenems [182]. From a therapeutic perspective, CRE represent a threat as only a few antibiotics retain activity against them. This is due to the ability of carbapenemases to hydrolyze most other β-lactam antibiotics, and to frequent coexistence in CRE isolates of additional mechanisms of resistance against other antibiotics such as fluoroquinolones and aminoglycosides [183]. To date, carbapenems are still considered a last-line of therapy against CRE, and, although few additional options exist, concerns over their efficacy and toxicity profiles are reported. In addition, rates of resistance to these agents such as tigecycline and polymyxins are increasing [184,185]. Currently, combination therapeutic strategies for CRE infections, include high-dose tigecycline, high-dose prolonged-infusion of

Enterobacteriaceae
Members of the family Enterobacteriaceae, such as E. coli, are normal inhabitants of the intestinal tract in humans and animals, and are commonly isolated in clinical cultures. They are the causative agent of several types of infections in humans, including respiratory tract infections, urinary tract infections, and bloodstream infections in hospitalized or otherwise immunocompromised subjects. In the context of antimicrobial resistance, Enterobacteriaceae are especially important as they are a cause of community-associated as well as healthcare-associated infections, which generates a major clinical and public health challenge [180]. The overdependence on carbapenems as empiric treatment in the management of infections caused by Enterobacteriaeceae was driven by cephalosporin resistance in this family due to production of ESBLs [181]. Carbapenems proved especially useful since ESBL producers are also resistant to other antibiotic classes including aminoglycosides, tetracyclines, and fluoroquinolones [71]. Unfortunately, this has led to the emergence of carbapenem-resistant Enterobacteriaceae (CRE), defined as bacteria belonging to the Enterobacteriaceae family that have the ability to survive and grow in the presence of clinically relevant concentrations of carbapenems [182]. From a therapeutic perspective, CRE represent a threat as only a few antibiotics retain activity against them. This is due to the ability of carbapenemases to hydrolyze most other β-lactam antibiotics, and to frequent coexistence in CRE isolates of additional mechanisms of resistance against other antibiotics such as fluoroquinolones and aminoglycosides [183]. To date, carbapenems are still considered a last-line of therapy against CRE, and, although few additional options exist, concerns over their efficacy and toxicity profiles are reported. In addition, rates of resistance to these agents such as tigecycline and polymyxins are increasing [184,185]. Currently, combination therapeutic strategies for CRE infections, include high-dose tigecycline, high-dose prolonged-infusion of carbapenem, and double carbapenem therapy. Newly available combinations like ceftazidime/avibactam are active against KPC and OXA-48 producers while meropenem/vaborbactam is active against KPC-producers. Plazomicin, which is a next-generation aminoglycoside, and eravacycline, which is a tetracycline class antibacterial, have in vitro activity against CRE [186].
Enterobacteriaceae can become resistant to carbapenems by three possible mechanisms: efflux pump overactivity, porin loss or mutation, and carbapenemase production, which remains the main resistance mechanism [187]. In fact, while carbapenemases specifically target carbapenems and other ß-lactam antibiotics, efflux pump expression or porin changes are associated with multi-drug resistance, and aim to block penetration of antibiotics into the bacterial cell [188]. Regarding efflux pumps, the resistance-nodulation-division (RND) group is a major mechanism of multi-drug resistance in Enterobacteriaceae. Among the different efflux systems, the AcrAB-TolC RND system is the most common, and has been identified among carbapenem-resistant E. cloacae [189], E. coli [190], and K. pneumoniae [191]. Alteration of porin synthesis, such as deficiency of OmpK-35 and 36 in K. pneumoniae [192,193], and loss of OmpF and OmpC in E. coli [194], have been described in AmpCand carbapenemase-producing Enterobacteriaceae. Studies suggest that strains with porins mutated or downregulated typically do not have the potential for mobilization into community settings but may carry the risk of possible local proliferation within hospitals [188]. It is worth noting that CRE shows efflux activity or permeability lesions may express these mechanisms paired to production of other β-lactamases such as AmpC enzymes or ESBLs [181,193].
Nevertheless, carbapenemases are currently considered the major mechanism of resistance in CRE, even though this apparently was not the case around mid-2000s, when most carbapenemases were confined to P. aeruginosa, with only anecdotal reports in Enterobacteriaceae [73,195]. Afterward, and after the detection of the first strain of CRE in the 1990 [10], CRE had rapidly spread with different carbapenemases predominating in various geographic areas. For instance, NDM-1 is currently the main carbapenemase in India, Pakistan, and Sri Lanka, as well as in specific European countries, including Romania, Denmark, and Poland, mostly in K. pneumoniae and E. coli [71]. Morganella morgani was shown to harbor NDM-5 recently in reports from China [196], and GES-5 in reports from Brazil [197]. KPC-producing Enterobacteriaceae are endemic in the United States, Colombia, Argentina, Greece, and Italy, and cause epidemics in China and the Middle East [188]. On the other hand, OXA-48-like enzyme-producers are widespread in Turkey and surrounding countries as well as in the Middle-East and North Africa [33]. OXA-48-like producing K. pneumoniae clones have persisted in Turkey as a cause of nosocomial infections, and Turkey, according to the European Center for Disease Control and Prevention, which was reported as having the highest epidemiologic level (stage 5 "endemic situation") of these strains in 2014-2015 [198]. Today, the extensive international movement and exchange has helped OXA-48 producing Enterobacteriaceae to spread from many Middle-Eastern countries into other parts of the world.
Among CRE, and from an epidemiological implication, the international spread of KPC-producing K. pneumoniae of a single multi-locus sequence type (ST), ST258, is alarming. This clone is responsible for the rapid increase in antimicrobial resistance among K. pneumoniae strains [199]. K. pneumoniae is a major cause of hospital-acquired infections including pneumonia, bloodstream infections, urinary tract infections, and infections in newborns and the intensive care unit. In some countries, because of resistance, carbapenem antibiotics are not efficient in more than half of the patients treated for K. pneumoniae infections [200], and therapy choices are limited to colistin, polymyxin B, fosfomycin, tigecycline, and selected aminoglycosides [23]. The worldwide spread of K. pneumoniae ST258 represents a prototype of the role of epidemic plasmids in the dissemination of pathogens set for global nosocomial dominance, This is similar to the successful international spread of E.coli ST131 associated with ESBL production, especially CTX-M-15 [201]. The K. pneumoniae ST258 is described as a "high-risk clone." These clones are defined to show an enhanced ability to colonize, spread, and persist in a variety of niches. They have acquired certain adaptive traits that increase their pathogenicity and survival skills, and, at the expense of such adaptation, these strains have acquired antibiotic resistance pathways. They have the persistence and flexibility to accumulate and exchange resistance and virulence genes with other bacteria. High-risk clones have contributed to the propagation of different plasmids, genetic platforms, and resistance genes among Gram-negative bacteria [202,203]. In such a context, the K. pneumoniae ST258 pandemic has been distressing to the medical and scientific community. Originally, Kitchel and colleagues [204] showed in a nationwide study from the United States' Centers for Disease Control and Prevention in 2009, that 70% of KPC-producing K. pneumoniae across all states belonged to a single dominant strain, ST258, which was identified by both pulsed-field gel electrophoresis (PFGE) and MLST. Soon thereafter, epidemiologic surveillance from different countries showed global spread of ST258 among K. pneumoniae isolates with KPC in numerous countries such as Mexico [205], Canada [206], Brazil [207], and Ecuador [208]. Outside the Americas, ST258 was detected in Spain [209], Greece [210], Germany [211], Italy [212], Norway [213], China [214], and Korea [215], which suggests that it possesses characteristics of an international high-risk clone. Apart from such a clinical phenomenon, a recent paper from Croatia [216] described the first evidence of KPC-2-producing K. pneumoniae of ST258 in river water, where it persisted for 50 days. This confirms the ability of environmental perseverance and dissemination.
The genetic support of bla KPC in K. pneumoniae ST258 lies in a variety of plasmids, with the most commonly reported being an IncF plasmid with FII K replicons, first named pKpQIL. This was a 113-kb IncF plasmid with an FII K2 replicon containing Tn4401a, and was isolated from a KPC-3-producing K. pneumoniae [217]. Later, pKpQIL-like plasmids from strains in the United States were described, and associated with bla KPC-2 and, to a lesser extent, with bla KPC-3 . Additionally, findings demonstrated that pKpQIL plasmids are both spreading clonally in ST258 strains and transferred horizontally to different sequence types and species, which further highlights the clinical and public health concerns associated with this clone [218]. The ongoing K. pneumoniae ST258 pandemic is assumed to result from both plasmid-mediated spread that involves horizontal transmission of resistance genes between bacteria, as well as clonal expansion. Non-ST258 K. pneumoniae with bla KPC did not demonstrate global success as ST258 with bla KPC . Therefore, global dissemination and survival of K. pneumoniae ST258 are partly dependent on the combination with bla KPC on IncF plasmids with survival factors inherently present on the chromosome of this high-risk clone [200].
Overall, the worldwide genetic epidemiology of KPC-producing bacteria still shows that K. pneumoniae is the most common species and ST258 is the predominant clone. This suggests a unique fitness and selective advantage of this clone, which are far beyond simple antimicrobial resistance. It also suggests a capacity of this ST for high transmissibility, which likely requires immediate infection control actions and enhanced surveillance in favor of reducing the spread of KPC among Enterobacteriaceae.

Pseudomonas
A non-fermentative, and aerobic non-enteric Gram-negative bacterial pathogen, P. aeruginosa raises environmental, clinical, and global public health concerns due to its global presence, diverse ecological distribution, invasiveness, and life-threatening infections. It is widely distributed in the environment, and isolated from soil, organic matter, skin flora, water, plants, animal sources, moist surfaces, and medical equipment [219]. Infections with P. aeruginosa include pneumonia and sepsis, particularly in ventilated patients in intensive care units, and is still burdened with high morbidity and mortality. Additionally, it is associated with severe ocular, burn, and airway infections. It commonly infects patients with cystic fibrosis, chronic obstructive pulmonary disease (COPD), immunosuppressed organ transplant recipients, and/or those who underwent invasive medical procedures [220]. Intrinsically, P. aeruginosa is resistant to rifampin, tetracycline, chloramphenicol, trimethoprim-sulfamethoxazole, and many β-lactams. Low membrane permeability and efflux pump expression are partly responsible for such intrinsic resistance. As far as carbapenems are concerned, the Centers for Disease Control and Prevention reported in 2019 a carbapenem resistance rate of up to 12% in P. aeruginosa [221] while a PubMed search published in 2015 showed that international resistance rates of P. aeruginosa to carbapenems vary from 10% to 50% [222]. Although the rates varied from one country to another, the report estimates such strains to increase gradually due to both carbapenem use and medical applications, and the rate is estimated currently to be high enough to cause concern for public health microbiologists and infection specialists. Numerous resistance mechanisms drive carbapenem resistant phenotypes in Pseudomonas, most often including porin deficiency (especially OprD), efflux pump overactivity (mainly MexAB-OprM and MexCD-OprJ), and, less often, carbapenem-inactivating enzymes [223].
Regarding the latter mechanism, and in which lies the scope of this review, P. aeruginosa intrinsically produces chromosomal AmpC cephalosporinases, and has acquired both narrow-spectrum (such as PSE-1 and PSE-4) and broad-spectrum (PER-1, VEB-1, BEL-1, GES-1, and GES-13) β-lactamases [223]. All these enzymes do not confer resistance to carbapenems. Carbapenemases in P. aeruginosa belong to Ambler classes A, B, and D and have been extensively investigated. It is worth mentioning that very peculiar, naturally occurring, chromosomally encoded, class C enzymes that confer weak carbapenem-hydrolyzing activity have been identified in P. aeruginosa. The clinical relevance of these enzymes remains to be clarified [224].
Among class A carbapenemases, GES-2 was the first carbapenemase of the group to be identified in a strain of P. aeruginosa isolated from a patient in South Africa [65]. The gene encoding GES-2 was located on a self-transferrable plasmid, and differed from GES-1 by one amino acid substitution, which was thought to extend its substrate profile to hydrolyze imipenem. Recently, GES-5 and GES-24 carbapenemases were isolated from P. aeruginosa from long-term care facilities in South Korea [225]. GES-5 was also detected in studies from Canada [226], Indonesia [227], Dubai [67], and Japan [228]. A variant of GES-5 called GES-18 was isolated from P. aeruginosa recovered from the endotracheal aspirate of an elderly patient hospitalized in Belgium [229]. Both GES-5 and GES-18 genes are chromosomally encoded and were shown to be parts of class 1 integrons. Likewise, the chromosomal bla GES-20 carbapenemase gene was identified in a study of hospital isolates of P. aeruginosa from Mexico [68]. Apart from GES family, and, although KPC is commonly detected in Enterobacteriaceae, reports of KPC-2 in P. aeruginosa are accumulating, including data from Germany [230], Brazil [231], China [232], and Puerto Rico [233]. The bla KPC gene is carried by plasmids, which suggests the possibility of an inter-genus spread from Enterobacteriaceae into non-fermenters, even though evidence of such a spread is still unclear [224].
The first NDM-1 producing P. aeruginosa was reported in 2011 in Serbia, and this was simultaneously the first incidence of NDM-1 from the Balkan region [234]. Soon thereafter, a report described acute pyelonephritis due to NDM-1 producing P. aeruginosa in a patient in France who was previously hospitalized in Serbia [235]. Since then, additional reports of NDM-1-producing P. aeruginosa have appeared from Iraq [236], Poland [237], Singapore [238], and Malaysia [239]. As a reflection on diversity of MBLs encountered in P. aeruginosa in clinical settings, another MBL with 40% similarity to NDM-type enzymes, was isolated in Florence, Italy, from a P. aeruginosa isolate recovered from a patient with a vascular graft infection [240]. The enzyme was named FIM-1, (Florence IMipenemase), and the bla FIM-1 gene was apparently inserted into the chromosome and associated with the ISCR19 element. Most recently in 2019, and concerning MBL detection in P. aeruginosa, Boyd and colleagues from Canada [241] used whole-genome sequencing (WGS) to identify carbapenemase production in four clinical isolates of P. aeruginosa that were PCR-negative for KPC, OXA-48, NDM, VIM, IMP, GES, and NMC/IMI carbapenemase genes. The WGS analysis revealed a novel MBL gene, bla CAM-1 (Central Alberta MB-1L), chromosomally located in a 73 kb integrative element, which was not transferrable by conjugation.

Acinetobacter
The first decade of the 20th century has seen a surge in the incidence of infections due to several highly antimicrobial-resistant bacteria in hospitals worldwide. Acinetobacter species, especially A. baumannii, is one such organism that turned from an occasional respiratory pathogen into a major nosocomial one [246]. The ability of this pathogen to gain several virulence factors and to survive for prolonged periods has led to its successful emergence as an opportunistic pathogen causing bacteremia, sepsis, meningitis, and urinary tract infections. Additionally, A. baumannii has a remarkable propensity for rapid acquisition of resistance to an extensive range of antimicrobial agents. It exhibits a major resistance profile toward carbapenems and other β-lactams, which leaves clinicians with limited therapeutic options [247]. Carbapenem resistance among A. baumannii is conferred by several coexisting mechanisms including a decrease in outer membrane permeability, efflux pumps, hyperproduction of AmpC cephalosporinases, and modification of penicillin-binding proteins. However, the most prevalent mechanism of carbapenem resistance among A. baumannii is associated with carbapenemases of Ambler classes B and D. In addition, there have also been reports of resistance mediated by selected Ambler class A carbapenemases among A. baumannii strains [248].
Ambler class B enzymes have been rising in A. baumannii for the previous decade [246]. Perhaps the most concerning is NDM, which has been identified in A. baumannii since 2010 in India, where it co-existed with OXA-23 [249]. In 2011, 4 NDM-1-producing isolates were identified in a multicenter surveillance study in China [250]. In 2013, NDM-1-producing A. baumannii caused a hospital outbreak in France, which was traced to two index patients previously hospitalized in Algeria, and with no clear link to the Indian subcontinent [251]. Given the relationship between North African countries and European countries, it was then anticipated that the spread of NDM-1-producing A. baumannii would occur rapidly, as this organism is difficult to eradicate. Expectedly, similar isolates were detected in Greece [252], Denmark [253], Belgium [254], and Czech Republic [255]. NDM-1-producing A. baumannii were then considered disseminated in European countries and classified into three distinct sequence types harboring a chromosomally located bla NDM-1 gene within a Tn125 transposon [256]. NDM-1 was also isolated from A. baumannii recovered from Syrian civilians injured during the civil war [99]. NDM-2-producing A. baumannii was detected in UAE [257] and Egypt [100], which raises the suspicion that, besides the Indian subcontinent, the Middle East area may also present a reservoir for NDM-producing A. baumannii. Sporadically, about nine IMP variants and five VIM variants were detected in A. baumannii [224]. SIM-1 has been reported from A. baumannii in South Korea were it appears to be widespread [120], and has disseminated to other Acinetobacter species like A. pittii and Acinetobacter nosocomialis [258].
A. baumannii possesses naturally occurring class D β-lactamases, known as OXA-51-like enzymes, that exhibit weak carbapenemase activity. Noticeably, the corresponding genes are not expressed or only weakly expressed in most isolates. However, once overexpressed, they may subsequently be involved in reduced susceptibility to carbapenems [224]. The overexpression bla OXA-51-like genes is often driven by the upstream insertion of an ISAba1 element, which provides strong promoter sequences, and increases expression of OXA-51 by about eight-fold [135].
Besides the naturally occurring bla OXA-51 , many acquired class D carbapenemases have been detected in A. baumannii. Currently, perhaps the most important OXA-type enzyme highly prevalent in this organism is OXA-23. As of 2010, worldwide dissemination of the bla OXA-23 gene was established after a study of strains from 15 countries in different regions and on five continents [28]. The contemporary worldwide dissemination of this gene was shown to be driven by >7 MLST types associated with different genetic structures and located on either chromosomes or plasmids. The complex and dynamic spreading of bla OXA-23 will be difficult to control because this spread is not associated with a single genetic entity. The transposons Tn2006, Tn2007, and Tn2008 were identified as genetic structures harboring this gene. In Tn2006, the bla OXA-23 gene is flanked by two copies of the insertion sequence ISAba1, which are located on opposite orientations [259]. Tn2008 is similar to Tn2006 but lacks the second copy of ISAba1 and the bla OXA-23 gene is associated with one copy of ISAba4 (which differs from ISAba1) in Tn2007 [132]. While OXA-23-like enzymes are commonly associated with hospital outbreaks, and pandemic clones producing these enzymes are identified [260,261], other families like OXA-58, OXA-143, and OXA-235 are identified on a more limited or sporadic basis, as explained above in Section 5. bla OXA-23 has been detected in A. baumannii from companion animals at veterinary clinics in Germany, on plasmids with Tn2008 [262]. Therefore, resistance determinants and clonal lineages of A. baumannii strains globally emerging in humans require close molecular surveillance to monitor this interspecies spread and mitigate its effect on both human and non-human hosts.
Among carbapenemases of Ambler class A, specific GES variants identified in A. baumannii compromise carbapenem activity. Among these are GES-5 [263] detected in Saudi Arabia, GES-11 detected in France [69] and Lebanon [83], and GES-14 detected in France [264] and Kuwait [265]. KPC enzymes, belonging to Ambler class A, have mostly been identified in Enterobacteriaceae, but a report from Puerto Rico in 2010 described KPC-2, KPC-3, KPC-4, and KPC-10 in A. baumannii [266]. Soon thereafter, molecular characterization of bla KPC in A. baumannii in the same country revealed its association with the transposon Tn4401b and its integration in the chromosome by a transposition event mediated by the transposase ISEcp1 [267]. Further analysis showed that the gene is associated with a plasmid fragment derived from Enterobacteriaceae [59]. KPC-3 was detected in A. baumannii from an infected wound culture of a patient admitted to a university hospital in Portugal. The isolate accumulated carbapenem, tigecycline, and colistin resistance [60]. No other reports of KPC-producing A. baumannii exist in other countries.
Regarding the class C enzymes, intrinsic to A. baumannii, exists as a chromosomal AmpC β-lactamase, that, like OXA-51, is normally expressed only at a low level. However, it can be overexpressed as a result of upstream insertion of ISAba1 sequences, which exist in up to 13 copies per cell, and is thought to act as a moving switch to the nearby genes [181]. WGS experiments have shown the existence of multiple variants of AmpC, called Acinetobacter-derived cephalosporinases (ADCs) [268]. Although insertion of ISAba1 upstream of the genes encoding ADCs leads to a likely higher β-lactam hydrolysis rate, the clinical effect of such a factor on carbapenem resistance is not clear [224]. Nevertheless, in 2014, Jeon et al. [179] defined a novel ADC, named ADC-68. The gene encoding this enzyme was chromosomal, and analysis of the enzyme structure showed a particular conformation that was able to accommodate not only extended-spectrum cepahlosporins, but also carbapenems. Therefore, ADC-68 was the first reported enzyme among chromosomal class C β-lactamases to possess both extended-spectrum β-lactamase and carbapenemase activities.

Other Gram-Negative Organisms
Besides the above clinically significant bacteria, carbapenemases are also detected in other Gram-negative organisms both from patients and from the environment. For instance, the class A carbapenemase, PenA, was detected in Burkholderia cepacia complex isolates from cystic fibrosis patients [269]. The environment plays a major role in the maintenance and genetic exchange of resistance determinants between environmental and pathogenic bacteria. This explains why carbapenemase studies on aquatic sources (rivers, lakes, and wastewater treatment plants), livestock, and wildlife are accumulating. For example, hospital wastewater treatment plants in China were shown to contain KPC-2-producing Raoultella ornithinolytica [270], which is an aquatic Gram-negative organism with emerging importance in hospital infections associated with invasive procedures [271]. According to a Swiss study in 2019, the environmental Gram-negative organism Zhongshania aliphaticivorans, which lives in marine sediments, was shown to produce ZHO-1. This is an intrinsic carbapenemase with significant hydrolytic activity against most β-lactams including penicillins, cephalosporins, and carbapenems, with the exception of aztreonam and cefepime [272]. Another Swiss study, also in 2019, recognized the environmental Gram-negative species Pseudobacteriovorax antillogorgiicola to harbor PAN-1, which is an MBL with hydrolytic activity toward most β-lactams including carbapenems but not cefepime and aztreonam [273]. Results from the last two studies further add to the knowledge that environmental species are a reservoir of possible clinically relevant MBLs. Xin et al. reported pollution of estuarine water with bacterial genera including Rheinheimera, Stenotrophomonas, Shewanella, Raoultella, Vibrio, Pseudoalteromonas, Algoriphagus, Bowmanella, and Thalassospira, which all harbor bla OXA-58 [155] conferring resistance to carbapenems and penicillin and an ability to hydrolyze cefpirome and cephalothin but not ceftazidime, cefotaxime, or cefepime. Citrobacter freundii OXA-204 [171] Hospital wastewater E. coli KPC-2 [64] Raoltella ornitholitica KPC-2 [270] Acinetobacter towneri OXA-58 [154] Coastal water A. towneri OXA-58 [155] Municipal water Acinetobacter lwoffii OXA-235 and OXA-278 [163] River water Klebsiella pneumoniae KPC-2 [216] Estuarine water

Conclusions
Almost three decades after their original discovery in Gram-negative bacteria, carbapenemases linger to create a definite growing threat to public health, and carbapenemase-encoding genes are already widespread in many parts of the world. Awareness of the prevalence and incidence of carbapenemases is crucial in preventing their spread and selection of appropriate prevention and containment options. Perhaps particularly alarming is the fact that carbapenemases are not only restricted to hospital isolates where they compromise advanced medicine. They have been continuously circulating among hospitals, long-term care facilities, community, animals, and the environment. In a period of widespread international travel, tourism, population migration, and patient transfer to receive medical care, the association between a specific resistance mechanism and a given region or country may change, which creates an urgent need for routine local, national, and global surveillance of carbapenemases. Furthermore, in light of dissemination of carbapenemases among both fermenters and non-fermenters, the complexity of their genetic transfer, and heterogeneity of their genetic backgrounds, ongoing research on their molecular epidemiology is imperative.