Class C β-Lactamases: Molecular Characteristics

SUMMARY

Class C β-lactamases or cephalosporinases can be classified into two functional groups (1, 1e) with considerable molecular variability (≤20% sequence identity). These enzymes are mostly encoded by chromosomal and inducible genes and are widespread among bacteria, including Proteobacteria in particular. Molecular identification is based principally on three catalytic motifs (⁶⁴SXSK, ¹⁵⁰YXN, ³¹⁵KTG), but more than 70 conserved amino-acid residues (≥90%) have been identified, many close to these catalytic motifs. Nevertheless, the identification of a tiny, phylogenetically distant cluster (including enzymes from the genera Legionella, Bradyrhizobium, and Parachlamydia) has raised questions about the possible existence of a C2 subclass of β-lactamases, previously identified as serine hydrolases. In a context of the clinical emergence of extended-spectrum AmpC β-lactamases (ESACs), the genetic modifications observed in vivo and in vitro (point mutations, insertions, or deletions) during the evolution of these enzymes have mostly involved the Ω- and H-10/R2-loops, which vary considerably between genera, and, in some cases, the conserved triplet ¹⁵⁰YXN. Furthermore, the conserved deletion of several amino-acid residues in opportunistic pathogenic species of Acinetobacter, such as A. baumannii, A. calcoaceticus, A. pittii and A. nosocomialis (deletion of residues 304–306), and in Hafnia alvei and H. paralvei (deletion of residues 289–290), provides support for the notion of natural ESACs. The emergence of higher levels of resistance to β-lactams, including carbapenems, and to inhibitors such as avibactam is a reality, as the enzymes responsible are subject to complex regulation encompassing several other genes (ampR, ampD, ampG, etc.). Combinations of resistance mechanisms may therefore be at work, including overproduction or change in permeability, with the loss of porins and/or activation of efflux systems.

INTRODUCTION

β-lactamases remain an important natural or acquired mechanism of resistance to β-lactam antibiotics. New enzymes of this type, belonging to the molecular classes defined by Ambler and then completed by several authors (1–3), have been regularly discovered since the 1980s. The class C β-lactamases (BLCs), also known as AmpC or cephalosporinases, have a long history marked by the gradual loss of efficacy for the treatment of many bacterial infections, due initially to their large inactivation spectrum, including penicillins, the first cephalosporins (e.g., cephalothin), and cephamycins (e.g., cefoxitin), together with the general absence of an inhibitory effect of clavulanic acid, sulbactam, and tazobactam (4–6). The next problem encountered was the emergence of constitutive or overproduced mutants, eventually overcome by the development of oxyiminocephalosporins, such as cefotaxime and ceftazidime (7). However, plasmid-borne cephalosporinases were subsequently discovered, particularly in species without chromosomal ampC genes (e.g., Klebsiella pneumoniae, Salmonella enterica, and Proteus mirabilis) and in Escherichia coli, which possesses an intrinsic ampC gene usually not expressed. These enzymes are derived from chromosomally encoded enzymes specific to other species, such as Enterobacter cloacae and Citrobacter freundii, and their discovery raised new fears, allayed by the discovery of cefepime and cefpirome (4, 8). However, a new step in resistance development was then detected, with the discovery of extended-spectrum β-lactamases AmpC (ESAC), mutants or variants with an extended inactivation spectrum for oxyiminocephalosporins (ceftazidime, cefotaxime, cefepime, and cefpirome) due to the mutation of certain sites, in the R2-loop (9, 10), for example, by substitution, deletion, or insertion. Carbapenem resistance mediated by a combination of mechanisms, including a constitutive species-specific cephalosporinase and porin loss, has emerged more recently (11, 12). Finally, the recent development of novel enzyme inhibitors, such as avibactam, has elicited considerable medical interest due to its ability to inhibit serine β-lactamases (classes A, C, and D). However, this has already led to the selection of clinical mutants resistant to ceftazidime–avibactam combinations, most of them bearing deletions of various sizes in the Ω-loop region of AmpC (13, 14), similarly to β-lactamases from other classes (e.g., KPC-3).

Various methods for detecting resistant clinical isolates have been proposed, particularly for resistance to oxyiminocephalosporins and carbapenems. These methods include phenotypic tests, enzymatic methods based on hydrolysis, immunochromatographic assays, and molecular tests designed to test for the presence of particular genes, encoding class C β-lactamases, for example (15). Whole-genome sequencing (WGS) is a very useful approach for the precise identification of mechanisms of resistance to β-lactams, and has provided a large body of sequence data. Nevertheless, genotype-to-phenotype extrapolations are not straightforward, due to the variability of gene expression and polymorphisms linked to silent mutations at diverse sites, depending on the β-lactamase considered (16). Improvements in our knowledge should make it possible to improve analyses of the resistance mechanisms detected, particularly against β-lactams, in the future. These mechanisms are numerous and differ considerably between species. The current classification of β-lactamases into four molecular classes is based on motifs involved in binding and hydrolysis, such as ⁷⁰SXXK, ¹³⁰SDN, and ²³⁴KTG for class A (17), and on diverse residues involved in determining affinity, either increasing the inactivation spectrum (e.g., ESBLs) or decreasing it (e.g., IRT) (18–20). A more detailed comparative analysis of the primary structure of a large number of proteins would facilitate the classification of enzymes into groups of clusters displaying common structural features, as for the enzymes of class A (21). BLCs seem to have a lower level of structural diversity, but the abundance of data now available in databases (e.g., Beta-Lactamase DataBase [BLDB, http://bldb.eu/] [22], Bacterial Antimicrobial Resistance Reference Gene Database [https://www.ncbi.nlm.nih.gov/bioproject/313047] [23], CARD [https://card.mcmaster.ca/] [24]) suggests that a more detailed analytical approach, based on several thousand sequences, would now be justified.

When we began this analysis in 2019, several numbering schemes had been proposed for class C β-lactamases, which were unified in a standardized structure-based numbering scheme published in 2020 (25). Accurate comparisons of protein sequences have improved structural classification, by providing a clearer identification of polymorphisms by species, a more precise identification of the bacterium updated in line with the continual changes in taxonomy, a better understanding of the residues or zones involved in the possible extension of the inactivation spectrum by species or bacterial group, and with respect to both substrates and enzyme inhibitors, such as avibactam. The ACC-type β-lactamases appeared to be plasmid-encoded enzymes originating from Hafnia alvei with an unusual susceptibility pattern characterized by resistance to expanded-spectrum cephalosporins such as ceftazidime, cefotaxime, and, sometimes, cefpirome, and by susceptibility to cefepime and inhibition by cefoxitin (see the section on Natural ESACs for more details).

PHYLOGENETIC COMPARISON

The serine β-lactamases have been divided into three classes (A, C, and D) based on sequence similarity (1, 2, 26). A protein structure-based phylogeny clearly distinguished between these classes (27–29). The amino acid-based phylogenetic tree based on 3,943 class C sequences reveals large differences between the major groups (Acinetobacter, Aeromonadales, Burkholderiales, Enterobacterales, Pseudomonadales, Rhizobiales), each of which displays at least 24% amino-acid sequence identity (Fig. 1).

FIG 1.

(A) Phylogram for representative and putative class C β-lactamases, compared with β-lactamases from classes A, B, and D. (B) Focused view on the phylogram of class C β-lactamases. The protein sequences of representative enzymes are listed in references 33, 47, 54, and 64. The sequences were filtered using CD-HIT (https://github.com/weizhongli/cdhit) at 90% sequence identity, then aligned with Clustal Omega (319). The tree was constructed using RAxML (320), and the phylogram generated using FigTree (version 1.4.3). The tree was unrooted.

Highly conserved and major clusters, including many variants displaying >75% sequence identity with each other, were found for the following genera: Escherichia/Shigella, Citrobacter, Enterobacter, Hafnia, Klebsiella, Morganella, Serratia. Other clusters (Acinetobacter, Aeromonas, Burkholderia, Erwinia, Pseudomonas, Rhizobium, Yersinia, etc.) are less conserved, with a percent amino-acid identity between sequences of less than 50%. These findings highlight the need for more accurate taxonomic approaches in the future. Sequences from the Proteobacteria were particularly prevalent, but BLCs were widely distributed between bacterial groups, with the exception of Gram-positive bacteria, in which they were rare (Table 1) (28, 29). Finally, a separate cluster including several genera (e.g., Bradyrhizobium, Legionella, Parachlamydia) was identified and found to contain generic serine hydrolases and some carboxylesterases VIII with weak β-lactamase activity (Fig. 1).

TABLE 1.

Overview of bacteria producing class C β-lactamases

Class	Order	Genus
Alphaproteobacteria	Rhizobiales	Agrobacterium, Bosea, Bradyrhizobium, Inorhizobium, Mesorhizobium, Methylobacterium, Microvirga, Ochrobactrum, Phyllobacterium, Pseudorhodoplanes, Rhizobium
	Rhodobacterales	Rhodobacter, Ruegeria, Silicibacter, Sulfitobacter
	Rhodospirillales	Dongia
Betaproteobacteria	Burkholderiales	Achromobacter, Bordetella, Burkholderia, Caballeronia, Collimonas, Cupriavidus, Herbaspirillum, Janthinobacterium, Massilia, Noviherbaspirillum, Pandoraea, Paraburkholderia
	Neisseriales	Chromobacterium, Laribacter, Snodgrassella
	Rhodocyclales	Thauera
Gammaproteobacteria	Aeromonadales	Aeromonas
	Alteromonadales	Shewanella
	Cellvibrionales	Microbulbifer
	Enterobacterales	Budvicia, Buttiauxella, Cedecea, Citrobacter, Cronobacter, Edwardsiella, Enterobacter, Erwinia, Escherichia:Shigella, Ewingella, Hafnia, Klebsiella, Lelliottia, Morganella, Pantoea, Photorhabdus, Pluralibacter, Pragia, Providencia, Regiella, Rouxiella, Serratia, Siccibacter, Xenorhabdus, Yersinia
	Legionellales	Legionella
	Oceanospirillales	Aidingimonas, Chromohalobacter, Halomonas, Salinicola
	Pseudomonadales	Acinetobacter, Pseudomonas, Psychrobacter
	Vibrionales	Vibrio
	Xanthomonadales	Dyella, Lysobacter, Xanthomonas
Deltaproteobacteria	Myxococcales	Myxococcus
Terrabacteria	Actinobacteria	Mycobacterium
	Negativicutes	Pelosinus
FCB group	Chitinophagales	Sediminibacterium
	Cytophagales	Dyadobacter, Emticicia, Siphonobacter
	Flavobacteriales	Chryseobacterium
	Sphingobacteriales	Sphingobacterium
PVC group	Chlamydiales	Chlamydia
	Parachlamydiales	Parachlamydia
Unclassified		Dependentiae

PRIMARY STRUCTURE/SEQUENCE ANALYSIS

Highly Conserved Motifs and Residues

A standardized amino-acid numbering scheme was recently developed, through collaboration, for class C β-lactamases (25). This numbering scheme has greatly facilitated molecular comparisons between class C enzymes. Indeed, “SANC” (structural alignment-based numbering of class C β-lactamases) preserves the usual numbering of the major catalytic residues (⁶⁴S, ⁶⁷K, ¹⁵⁰Y, and ³¹⁵K). Three highly conserved motifs are currently used to characterize this molecular class (⁶⁴SXSK, ¹⁵⁰YXN, and ³¹⁵KTG) (17, 28), but other conserved residues have been characterized and can be used for the more accurate identification of AmpC enzymes (Tables 2 and 3). They include 37 residues strictly conserved (100%) and 33 residues highly conserved (90–97%) in a data set of 32 representative class C β-lactamases examined in a previous study (25). Proline (P) and aromatic (F, W, Y) residues are overrepresented at these positions compared to class A β-lactamases. Indeed, prolines are present at positions 18, 26, 94, 118, 122, 140, 192, 213, 277, 330, and 345 for class C β-lactamases, but only at positions 107, 183, and 226 for class A β-lactamases. Similarly, strictly conserved aromatic residues are present at positions 60, 138, 150, 199, 203, 221, 271, 322, 325, and 328 for class C β-lactamases, but only at positions 66, 210, and 229 for class A β-lactamases. If aromatic amino acids (F, Y and W) are considered together, as a single category, additional conserved positions emerge for BLCs: 43, 69, 134, 135, 170, 188, 233, 260, 276, and 344 (Table 3).

TABLE 2.

Representative class C β-lactamases

Bla	Origin of namec	Accession no.	Genomic localizationa	Organism	No. of residues	References
ACC-1	Ambler Class C-1	AJ133121	P	Klebsiella pneumoniae	386	321
ACT-1	AmpC Type	U58495	P	Klebsiella pneumoniae	381	322
ABA-1 (ADC-2)	Acinetobacter baumannii	AY177427	IS	Oligella urethralis	383	323
ABAC-1 (ADC-3)	Acinetobacter baumannii Class C	AY178995	Chr	Acinetobacter baumannii	383	323
ADC-1	Acinetobacter-derived cephalosporinase	AJ009979	Chr	Acinetobacter baumannii	383	247
AQU-1	Aeromonas aquariorum	AB765393	Chr	Aeromonas dhakensis	380	249
AsbA1	Aeromonas so bria	U10250	In	Aeromonas jandaei	381	324
BIL-1 (CMY-2-like)	Name of patient (Bilal)	X74512	P	Escherichia coli	383	325, 326
BlaE	Gene name	AY442183	Chr	Mycobacterium smegmatis	380	327, 328
BUT-1	Buttiauxella sp.	AJ415568	Chr	Buttiauxella sp.	383	329
CAV-1	Aeromonas caviae	AF462690	Chr	Aeromonas caviae	382	44
CDA-1	Cedecea davisae	KJ650399	Chr	Cedecea davisae	382	330
CepH	Cephalosporinase hydrophila	AJ276030	Chr	Aeromonas hydrophila	382	41
CepS	Cephalosporinase sobria	X80277	Chr	Aeromonas sobria	382	331
CFE-1	Citrobacter fr eundii	AB107899	P	Escherichia coli	381	208
CHR-1	Chromohalobacter sp.	AB070219	Chr	Chromohalobacter sp.	396	169, 332
CMA-1	Cronobacter malonaticus	KF640251	Chr	Cronobacter malonaticus	375	333
CMH-1	Chi Mei Hospital	JQ673557	P	Enterobacter cloacae	381	334
CMY-1	Active on cephamycins	X92508	P	Klebsiella pneumoniae	382	233
CMY-2	Active on cephamycins	X91840	P	Klebsiella pneumoniae	381	335
CSA-1	Cronobacter sakazakii	KF623543	Chr	Cronobacter sakazakii	375	333
DHA-1	Dhahran (Saudi Arabia)	Y16410	P	Salmonella enteritidis	379	336, 337
Ear-1	Enterobacter ae rogenes	AJ544162	Chr	Enterobacter aerogenes	381	338
EDC-1	Edwardsiella AmpC	EF467366	Chr	Edwardsiella tarda	386	–
ENT-1	Buttiauxella agrestis CF01Ent1	AJ489827	Chr	Buttiauxella agrestis	390	339
ERH-1	Erwinia rhapontici	AY288518	Chr	Erwinia rhapontici	379	340
FOX-1	Active on cefoxitin	X77455	P	Klebsiella pneumoniae	382	341
IDC-1	Integron derived cephalosporinase	MN985649	In	sediment metagenome	395	31
K12 (EC-1)	Escherichia coli K12	J01611	Chr	Escherichia coli	377	342, 343
LAT-1	Active on latamoxef	X78117	P	Klebsiella pneumoniae	381	344, 345
LHK-1	Laribacter hongkongensis	AY632070	Chr	Laribacter hongkongensis	388	346
LRA10-1	β-lactam resistance from Alaska	EU408357	–	uncultured bacteria (soil)	375	30
LRA13-1b	β-lactam resistance from Alaska	EU408352	–	uncultured bacteria (soil)	609b	30
LRA18-1	β-lactam resistance from Alaska	EU408355	–	uncultured bacteria (soil)	386	30
LYL-1	Lysobacter lactamgenus	X56660	Chr	Lysobacter lactamgenus	385	347
MIR-1	Miriam hospital	M37839	P	Klebsiella pneumoniae	381	348, 349
MOX-1	Active on moxalactam	D13304	P	Klebsiella pneumoniae	382	232, 350
OCH-1	Ochrobactrum anthropi	AJ401618	Chr	Ochrobactrum anthropi	390	351
P99 (ACT-89)	Enterobacter hormaechei P99	X07274	Chr	Enterobacter hormaechei	397	352
PAO-1 (PDC-1)	Pseudomonas aeruginosa (Pseudomonas-derived cephalosporinase)	AY083595	Chr	Pseudomonas aeruginosa	397	353, 354
PAC-1	Pseudomonas aeruginosa Class C	KY285014	Tn	Pseudomonas aeruginosa	381	75
SLC-1	Serratia liquefaciens Class C	DQ022079	–	uncultured bacteria (soil)	379	76
PSI-1	Psychrobacter immobilis	X83586	Chr	Psychrobacter immobilis	401	355, 356
RHO-1	Rhodobacter sphaeroides	CP000144	Chr	Rhodobacter sphaeroides	380	357
SR50 (SRT-1-like)	Serratia marcescens SR50, Serratia resistant	X52964	Chr	Serratia marcescens	376	358
SST-1	Susceptible strain	AB008455	Chr	Serratia marcescens	378	359
TRU-1	Formerly Aeromonas tructi	EU046614	Chr	Aeromonas enteropelogenes	382	360
YEC-1	Yersinia enterocolitica cephalosporinase	X63149	Chr	Yersinia enterocolitica	388	361
YRC-1	Yersinia ruckeri cephalosporinase	DQ185144	Chr	Yersinia ruckeri	383	362

^aChr, chromosome; In, integron; P, plasmid; Tn, transposon; IS, insertion sequence; –, unknown.

^bFusion between two β-lactamases (class C and class D).

^cThe underlined letters form the name of the β-lactamase.

TABLE 3.

Conserved residues in class C β-lactamases

Positiona	Residue (>90 % conserved)	Secondary structureb	% conservedb
18	P	H1	94
26	P	S	97
27	G	E	100
29	A	E	97
36	G	T	97
43	F/Y/W	E	100
44	G	E	100
54	V		91
58	T		100
60	F	E	100
61	E	E	100
63	G	G	100
64	S	G	100
66	S	H2	100
67	K	H2	100
71	G/A	H2	100
73	L	H2	94
77	A	H2	91
94	P	G	94
96	L	G	100
109	L	H3	97
110	A/G	H3	100
111	T	T	100
113	T/S		100
115	G	S	100
116	G		96
118	P		94
119	L	S	97
122	P		100
123	D/E	T	100
134	F/Y/W	H4	100
135	Y/F	H4	97
138	W		97
140	P		100
145	G	T	97
148	R	E	100
150	Y		100
152	N	H5	100
155	I	H5	91
156	G	H5	97
159	G	H5	100
170	F/Y	H6	100
187	T/S	Ω	100
188	Y/W/F	Ω	100
191	V	Ω	97
192	P	Ω	97
199	Y	Ω	97
200	A	Ω	100
202	G	Ω	100
203	Y	Ω	100
210	R/H	Ω	100
211	V	Ω	91
213	P	Ω	94
214	G	Ω	94
221	Y	Ω H7	100
222	G	Ω	100
224	K	Ω	94
229	D	H8	100
260	Y/W/F	E	97
267	Q	E	100
269	L	S	97
271	W		100
272	E	E	100
277	P	S	97
286	G	H10 R2	97
315	K	E	100
316	T	E	100
317	G	E	100
319	T		97
321	G	S	97
322	F		100
325	Y	E	100
328	F	E	97
330	P	E	100
335	G/A	E	100
337	V	E	91
339	L	E	100
340	A	E	97
341	N	S	100
345	P		97
349	R	H11	100
353	A	H11	100

^aAccording to the “SANC” class C β-lactamases numbering scheme described in reference 25. Residues in boldface type are involved in the catalytic mechanism and/or in substrate binding.

^bFrom the alignment of 32 representative class C β-lactamases examined in reference 25. H, α-Helix; S/T, bend or turn; E/B, β-strand or β-bridge; G, 3₁₀-Helix; Ω, Ω-loop; R2, R2-loop.

In a representative set of 50 typical AmpC (see Table 2), the total number of amino acids in these enzymes ranged from 358 (P. fluorescens TAE4) to 397 (PDC-1) with a mean value of 383 ± 5.2, and the mean of molecular mass was estimated at 41.64 ± 0.58 kDa. The number of amino acids may be much higher for enzymes with fused domains (29, 30). AmpC enzymes typically have alkaline isoelectric points (pI), with a mean pI of 8.48 ± 1.22 for this representative set. Nevertheless, low predicted pI values were obtained for some enzymes (4.37 for CHR-1, 5.27 for BlaE, 4.59 for RHO-1, and 5.59 for IDC-1) (4, 31).

Enterobacterales

The Enterobacterales (EB) are one of the major groups emerging from the phylogram in Fig. 1. This group was clearly separated into two major clusters based on the analysis of 413 protein sequences: EB1, including Buttiauxella, Cedecea, Citrobacter, Edwardsiella, Erwinia, Escherichia, Enterobacter, Klebsiella aerogenes, Lelliottia, Morganella, Pluralibacter, Serratia fonticola, Yersinia, and Xenorhabdus; and EB2, including Budvicia, Cronobacter, Erwinia, Hafnia, Pantoea, Photorhabdus, Pragia, Providencia, Regiella, Rouxiella, Serratia, and Siccibacter (32). Most species generally cluster together in the same genus or cluster, but, surprisingly, several species (e.g., Erwinia teleogrylli, S. fonticola and Y. ruckeri) were located at some distance from their main clusters. The following conserved residues distinguished between the EB1 and EB2 groups: D47K, V65L, G71A, V72T, G74A, W101F, Y112H, Q120F, S154G, F158L, W201Q, A255T, R258G, G270M, and H314N. A deletion in position 116 and an insertion in position 204a were also identified in EB2 (25, 33). These groups displayed a high degree of diversity (e.g., 55–100% sequence identity for the whole EB1 group). Consensus sequences (CS) were therefore determined for the main species.

For Escherichia species (E. coli/Shigella, E. albertii, and E. fergusonii), sequence identity ranged from 88% to 100% and 11 consensus sequences were obtained from 367 protein sequences. The E. coli sensu stricto population is now considered to have a strong phylogenetic structure, with the identification of at least 12 phylogroups (A, B1, B2, C, D, E, F, G, H, clade 1, clade III, and clade IV) (34).

Several conserved residues were found to be specific for phylogroups A, A1, and B1 (²³⁵R, ²³⁸M, ²³⁹N, ²⁴¹R, ²⁴⁵N), whereas others were specific for phylogroups B2 and D (²³⁵Q, ²³⁸L, ²³⁹K, ²⁴¹L, ²⁴⁵N/T) (16). All four clusters (A–D) carry RMNRE residues in the 235–245 region (33), and this key feature was observed in clinical isolates able to generate extended-spectrum AmpC β-lactamases (ESACs), as described above. The ²⁸⁷SGN triplet is present in the sequence of such enzymes (35). Phylogroup B2 contains the specific conserved residues ¹⁷⁵K, ¹⁹³S, ²⁸²I, ²⁸⁸D, ²⁹⁶R, and ³⁰⁰P, and is separated from phylogroup D by the residues ¹⁸⁵T and ²⁴⁴T, in particular.

Finally, 78 protein sequences from uncultured bacteria obtained by metagenomic analyses on human fecal and environmental samples, originally found in the NCBI database as class C CMY-LAT-MOX-ACT-MIR-FOX β-lactamases, belong to this major cluster. They have been assigned to E. coli, Shigella, E. fergusonii, or E. albertii, and are characterized by a typical triplet (⁶HSE) (36–38).

One particular feature of interest is that all the protein sequences obtained from the genus Hafnia (H. alvei, H. paralvei) carry a two-residue deletion (residues 289–290) in H-10 and the R2-loop. The plasmid-encoded enzymes ACC-1 and ACC-4 are more than 99.4% identical to the AmpC of H. paralvei, a newly identified bacterial species (39, 40).

Aeromonas

The second major group identified included 318 protein sequences from Aeromonas, a genus for which many new species have been discovered over the last 2 decades. This is also the group in which the first inducible cephalosporinases (AsbA1, CAV-1, CepH, CepS) were detected, in A. hydophila (41), A. jandei (42), A. sobria (43), and A. caviae (44). The group is highly diverse, with a sequence identity ranging from 50% to 100%. Thirty-six species from this group have been identified, including about 20 species pathogenic to humans, such as A. caviae, A. dhakensis, A. veronii, and A. hydrophila (45). Phylogenetic analysis identified two CS for A. caviae, A. dhakensis, A. hydrophila, and A. salmonicida, with sequence identities of 80% to 100%, suggesting that new species may be discovered in the future (46, 47). Some of these species displayed a particular primary structure modification: the deletion of two or three residues (positions 301 to 303) in the R2-loop, as in A. caviae and in A. dhakensis (25).

Acinetobacter

Genus Acinetobacter has a long history of taxonomic changes, but is dominated by A. baumannii, the principal genomic species, which plays a major role in nosocomial infections, particularly in intensive care units (48, 49). This major group is heterogeneous, with 500 protein sequences and 26 species, and with sequence identities ranging from 42% to 100% (50). The denomination ADC for “Acinetobacter-derived cephalosporinase,” followed by a number to distinguish between individual enzymes, was proposed because of the large numbers of A. baumannii variants or members of the Abc complex (51) (http://bldb.eu/BLDB.php?prot=C#ADC). In a previous analysis of 103 genomes, the bla_ADC-like gene was found to be present in 13 validly named species, eight genomic species, and six taxa (52). Three major clusters were identified: one cluster including the A. baumannii–A. calcoaceticus or Abc complex and a number of other species: A. lactucae (formerly known as A. dijkshoorniae), A. nosocomialis, A. oleivorans, A. pittii, and A. seifertii; a second cluster containing nine species: A. apis, A. albensis, A. bereziniae, A. bohemicus, A. celticus, A. guillouiae, A. johnsonii, A. kyonggiensis, and A. rudis; and a third cluster including a number of other species: A. baylyi, A. beijerinckii, A. gyllenbergii, A. haemolyticus, A. junii, A. proteolyticus, A. parvus, A. soli, A. ursingii, A. tjernbergiae, and A. venetianus (50). A large number of protein sequences were initially misidentified, for two reasons. The first was the substantial taxonomic modifications that had taken place, and the second was the preferential use of standard biochemical methods and automated systems or devices widely used in clinical bacteriology laboratories, such as API 20NE, Vitek 2, Phoenix, Biolog, and MicroScan WalkAway, potentially leading to incorrect identification. MALDI-TOF can facilitate correct identification of the members of the Abc complex to species level, provided that an accurate database is used (15, 49). Accurate identification of the clinically important members of the Abc complex (A. baumannii, A. pittii, A. nosocomialis, A. seifertti and A. dijkshoorniae) and A. calcoaceticus, which is considered pathogenic, is possible only by molecular methods such as DNA-DNA hybridization (gold standard method) and DNA sequence-based analysis on various types of DNA sequences (16S rRNA, rpoB, 16S-23S intergenic spacer) (15). The 16S rRNA sequencing method is highly reliable at genus level, but discriminates poorly between species. By contrast, rpoB genes are highly variable and considered appropriate for Acinetobacter species identification. The multiplex PCR based on species-specific genes, such as gyrB, is simple, rapid (results obtained within 2 h), and reproducible, but limited to major species: A. baumannii, A. nosocomialis, A. pittii and A. calcoaceticus.

One structural feature distinguishing these clusters was a conserved deletion of three residues in the R2-loop (residues 304–306) for the first two clusters (53, 54). The third cluster, including A. baylyi producing ADC-8 with a low susceptibility pattern, particularly for cephalosporins, did not display this deletion (54, 55). ADC-1 and ADC-68, naturally produced by A. baumannii, carry this deletion of three residues in the R2-loop (residues 304–306) and were considered to be extended-spectrum AmpC (ESACs). Such enzymes typically hydrolyze penicillins and narrow- and expanded-spectrum cephalosporins and aztreonam, but not cefepime. Acquired resistance to cefepime or cefpirome has been reported to be associated with amino-acid substitutions—R148Q (close to YXN), V211A (within the Ω-loop), and N287S (in helix H-10)—also conferring higher levels of resistance (4 to 64 times higher) to ceftazidime and cefotaxime (56, 57). Ceftazidime resistance resulted from overproduction of the ADC β-lactamase and the provision of strong promoter sequences by the insertion sequence ISAba1 (56, 58).

In conclusion, the complexity of this bacterial group suggests that other species are likely to be identified, particularly given the very small size of some of the clusters identified.

Pseudomonas

A phylogenetic analysis of 582 class C protein sequences from Pseudomonas yielded a broad distribution, with at least 36 clusters. Sequence identity varied considerably, from 42% to 100%, between the 33 species producing class C β-lactamases from this group. Heterogeneity levels were particularly high for some species, such as P. putida, for which at least 11 clusters were observed, and for P. fluorescens, suggesting misidentification, as reported by subsequent comparative studies of genomes (59–64). Indeed, the Pseudomonas group is among the most diverse in terms of the species it contains, but, with the exception of P. aeruginosa, for which 430 sequences have been analyzed, its taxonomy is still under revision.

P. aeruginosa is one of the principal pathogens isolated from immunosuppressed patients and cases of hospital-acquired infections associated with multiresistance (65). It also causes chronic lung infections in patients with cystic fibrosis (66). The effective treatment of P. aeruginosa infections is challenging, because several mechanisms of resistance to antibiotics, including β-lactamases and efflux pump overexpression, have evolved in this species (67). An inducible chromosomal AmpC-type enzyme was identified, with a wild-type inactivation spectrum including various β-lactams, such as aminopenicillins, the oldest cephalosporins, and cephamycins. Following its overproduction due to mutations altering the peptidoglycan recycling process, this cephalosporinase is a major source of resistance to ticarcillin, piperacillin, ceftazidime, and aztreonam (68, 69). However, additional missense mutations have extended its inactivation spectrum to cefepime and cefpirome. Many variants, also observed as natural polymorphisms, have been named according to a nomenclature specifically developed for P. aeruginosa, as PDCs (Pseudomonas-derived cephalosporinases) (68, 69). A phylogenetic analysis of 430 sequences displaying 87% to 100% identity yielded two clusters (C1, C2). The main cluster (C1), containing 416 protein sequences, had a consensus sequence with several variants (70). The second cluster (C2) was very small, comprising only 14 sequences. The polymorphism observed in cluster C1 had already been described in 44 antibiotic-susceptible strains with amino-acid substitutions at positions 53, 71, 79, 149, 178, and 365 (69). However, other variants were observed, in particular in positions 121, 153, 174, 178, 213, 216, 219, and 293, which served as mutation hot spots, thus extending the substrate specificity of PDC β-lactamases (see the section on ESACs).

Plasmid-encoded Enzymes

Chromosomal class C β-lactamases predominate, but plasmid-mediated enzymes are nevertheless important, and the protein sequences of 256 such enzymes were examined. Since their emergence in 1989, several types of plasmid-encoded AmpCs and their variants have been identified: ACC, ACT, CFE, CMY-2/BIL/LAT, DHA, FOX, MIR, and MOX/CMY-1 (4, 8, 22). CMY-2 is the most common plasmid-encoded enzyme family, detected worldwide in isolates that do not naturally produce a cephalosporinase (P. mirabilis, K. pneumoniae, and S. enterica, in particular), and also in E. coli, which has a weak expression of the natural cephalosporinase. This family displays little variability, with only about 8% of amino acids varying in a comparison of 92 sequences, including BIL-1 and LAT-1. Other families varied in terms of the number of protein sequences and percent identity: ACC (2 and 99%), ACT (10 and 86%), CFE (2 and 92%), DHA (15 and 97%), FOX (13 and 94%), MIR (5 and 99%), and MOX/CMY-1 (10 and 75%). Since the 1990s, many modifications have been made to bacterial taxonomy, with the definition of new bacterial species, particularly in the genera Citrobacter, Enterobacter, and Aeromonas. Finally, phylogenetic analysis confirmed several changes in the identification of the progenitors for chromosomal and plasmid-encoded cephalosporinases (25, 71–74) (Table 4). The rare class C β-lactamases SLC-1 and PAC-1, encoded by a gene in a chromosome-inserted Tn1721-like transposon, probably originate from an enterobacterium, but the origins of other enzymes (integron-encoded IDC-types, Alaska soil metagenomes LRA10-1, LRA13-1, LRA18-1) remain unknown (30, 31, 75, 76).

TABLE 4.

Class C β-lactamases and species-specific progenitors

Enzyme/straina	First identification	Updated identification	%b	References
ACC-1*c	Hafnia alvei	Hafnia paralvei	99.7	321, this review
ACT-1*c	Enterobacter cloacae	Enterobacter asburiae	98.4	322, this review
Aer-1	Enterobacter aerogenes	Klebsiella aerogenes	98.9	338, 363
AQU-1	Aeromonas aquariorum	Aeromonas dhakensis	99.2	249, 364
AsbA1	Aeromonas sobria	Aeromonas jandaei	94.5	324, this review
BIL-1*	Citrobacter freundii	Citrobacter portucalensis	96.8	326, this review
BUT-1	Buttiauxella sp.	Scandinavium goeteborgense	99.0	329, 365
BUT-2	Buttiauxella agrestis	Scandinavium goeteborgense	99.5	339, 365
CAV-1	Aeromonas caviae	Aeromonas allosaccharophila	97.1	44, 71
CFE-1*	Citrobacter freundii	Citrobacter europaeus	99.2	208, this review
CFE-2*	Citrobacter freundii	Citrobacter werkmanii	95.0	366, this review
CMY-1*c	Aeromonas hydrophila	Aeromonas sanarellii	95.3	233, this review
CMY-2*	Citrobacter freundii	Citrobacter portucalensis	98.4	335, this review
EDC-1	Edwardsiella tarda	Edwardsiella piscicida	99.1	This review
FOX-1*	Aeromonas caviae	Aeromonas allosaccharophila	94-98	71, 341
LAT-1*	Citrobacter freundii	Citrobacter portucalensis	97.4	345, this review
MIR-1*c	Enterobacter cloacae	Enterobacter roggenkampii	99.7	25, 352
MOX-1*	Aeromonas hydrophila	Aeromonas sanarellii	94.5	72, 232
MOX-2*	Aeromonas sp.	Aeromonas caviae	98.9	72, 234
MOX-9*	Aeromonas caviae	Aeromonas media	98.0	72, 367
P99 (ACT-89)	Enterobacter cloacae	Enterobacter hormaechei	97.9	25, 352, this review
TRU-1	Aeromonas tructi	Aeromonas enteropelogenes	97.1	360, this review

^aPlasmid-encoded enzymes are labeled by an asterisk.

^bPercentage of identity with consensus sequence.

^cOther plasmid-encoded types: ACC-4 (99.5% for H. paralvei); ACT-3, ACT-6, ACT-8, and ACT-10 (≥97.6% for E. asburiae); CMY-17, CMY-55, CMY-132, and CMY-161 (≥ 98.1% for A. sanarellii); MIR-4 (99.4% for E. roggenkampii). Their respective phylogenies can be found in references 32, 46, 50, and 63; and 73 and 74.

Serine Hydrolase, a Class C β-lactamase or a Carboxylesterase VIII?

BLCs are usually genus-specific and chromosomally encoded, found mostly in Gram-negative bacteria, and more specifically in Alpha-, Beta- and Gammaproteobacteria. However, several clusters (e.g., Legionella, Bradyrhizobium, Parachlamydia) are clearly divergent, with low levels of sequence identity (between 10 and 20%) (28) (Fig. 1). This early divergence is related to a limited number of highly conserved motifs or residues, most reported in Table 3: ⁴³Y/G, ⁵⁴V/T, ⁵⁸T, ⁶⁰F/E, ⁶³A/G, ⁶⁴SXXK, ⁹⁴P, ⁹⁶L, ¹⁰⁹L, ¹¹¹T, ¹³⁸W, ¹⁴⁰P, ¹⁴⁵G, ¹⁴⁸R/Y, ¹⁵⁰YXN/H, ¹⁸⁷T, ²⁰²G/Y, ²²⁹D, ²⁷¹W, ³¹⁵KXG, ³³⁰P, ³³⁵G, ³³⁷V, ³³⁹L, ³⁴¹N, ³⁵⁷L, and ³⁶⁰L. Several bacteria from Bacteroidetes and Firmicutes (e.g., Pelosinus fermentans, Sediminibacterium salmoneum) also carry enzymes with some of these highly conserved residues and motifs: ²⁷G, ⁴⁴G, ⁵⁸T, ⁶⁰F, ⁶¹E/F, ⁶³G, ⁶⁴SXXK, ⁶⁷K, ⁹⁴P, ¹¹⁵G, ¹⁴⁴G, ¹⁵⁰YXN, ²⁰⁰A, ²⁰²G, ²²⁰A/S, ²²²G/S, ²²⁹D, ³¹⁵K/HT/SG, ³¹⁹T, ³²¹GF, ³³⁷V, and ³⁴¹N. These enzymes are generally identified as serine hydrolases in databases, but with no additional bacteriological and enzymatic details that could be used to determine whether or not they should be considered as class C β-lactamases.

One family of enzymes (microbial carboxylesterases VIII, EC 3.1.1.1) mostly identified in metagenomic libraries from various environmental samples, including EstC, EstM-N1/N2, Est22, EstU1, and EstSRT1, appears to be phylogenetically and structurally related to BLCs (Fig. 1). These enzymes generally hydrolyze nitrocefin (EstC), cephalothin (Est22), cephaloridine, cefazolin (EstU1), and oxyiminocephalosporins (EstSRT1) (77–83). The EstB and Est22 enzymes selectively deacetylate cephalosporin-based substrates leaving the amide bond of the β-lactam ring intact (82, 83), while for EstU1 it was not clear from the HPLC data if the observed pattern against cephalosporin substrates was due to deacetylation or amide bond hydrolysis of the β-lactam ring (77). The catalytic efficiencies of EstSRT1 for cephalothin, cefotaxime, and cefepime are similar to that of EstU1 for cefazolin (77). These enzymes bear the active site residues ¹⁰⁰S, ¹⁰³K, and ²¹⁸Y (⁶⁴S, ⁶⁷K, and ¹⁵⁰Y according to SANC numbering), as well as other residues highly conserved in BLCs (Table 3). They have an overall structure consisting of a mixed α/β domain and a small helical domain, similar to that of class C β-lactamases (80, 84–86).

The classification of these carboxylesterases as BLCs does not appear to be appropriate, given the total absence of β-lactamase activity for some of these enzymes, their low levels of in vitro antibacterial activity against β-lactams, including cephalosporins, their primary structure, and their low levels of sequence identity to genuine BLCs. Nevertheless, these findings suggest that the promiscuous β-lactamase activity of some of these family VIII esterases may have evolved from BLCs or vice versa, and that some may have closer evolutionary relationships to BLCs (79).

SECONDARY AND TERTIARY STRUCTURES

The first crystallographic structure of an AmpC was reported in 1994, for the covalent complex between a phosphonate transition-state analog and the E. hormaechei P99 (formerly known as E. cloacae P99) cephalosporinase, which belongs to the ACT family (87). Many structures from different families of class C β-lactamases have since been published, including ACC (88, 89), ACT (87, 90–102), ADC (53, 103–108), CMH (109), CMY (100, 110–114), EC (111, 115–152), FOX (153–155), MOX (156, 157), PDC (158–167) and TRU (168) enzymes, together with the halophilic β-lactamase (HaBLA) from Chromohalobacter sp. 560 (169) and the class C β-lactamase from a psychrophilic organism, Pseudomonas fluorescens (170). More than 200 structures of class C β-lactamases have been described, and an updated list of these structures can be found in the Beta-Lactamase DataBase (BLDB; http://bldb.eu/S-BLDB.php) (22).

The typical three-dimensional structure of class C β-lactamase contains two mixed α/β domains: one composed of nine antiparallel β-sheets and three α-helices (H1, H10 and H11) on one side, and the other, composed of three small antiparallel β-sheets and eight α-helices (H2 to H9) on the other side (Fig. 2). Two additional structural elements important for the interaction with substrates, Ω-loop and R2-loop, are located between H6 and H8 (residues 189–225), and close to H10 (residues 288–309), respectively (see reference 25 for a detailed analysis of residues involved in these regions). Structurally, class C β-lactamases belong to the Beta-lactamase PFAM family (PF00144) and share the same overall fold with class A and class D β-lactamases, penicillin-binding proteins (PBPs), and family VIII carboxylesterases.

FIG 2.

Representative three-dimensional structure for class C β-lactamases. The structure of E. hormaechei P99 (formerly known as E. cloacae P99; PDB code 1BLS) (87) is colored in orange (α-helixes) and purple (β-strands). The Ω- and R2-loops are colored in green and blue, respectively. The most conserved residues (see Table 3) are represented as sticks and colored in cyan. The numbering of residues follows the SANC nomenclature (25).

REGULATION AND EXPRESSION OF CLASS C β-LACTAMASES

In some Enterobacterales (other than E. coli) and P. aeruginosa, ampC gene expression is weak and inducible. This induction involves several proteins (AmpR, AmpD, and AmpG) and two muropeptides (4, 171–174). Nevertheless, the molecular mechanisms mediating AmpC overproduction in P. aeruginosa appear to be more complex than initially thought, as other proteins, such as AmpDh2, AmpDh3, and DacB (PBP4), have since been identified as involved in these mechanisms (175–177).

Genetic Context and Regulation

The ampR and ampC genes are linked. AmpR is encoded by a gene located immediately upstream from ampC and acts as a transcriptional activator, binding to the intercistronic region upstream from the ampC gene promoter. The ampD gene encodes an N-acetyl-anhydromuramyl-L-alanine amidase involved in recycling the products of peptidoglycan catabolism, including, in particular, the 1,6-anhydro-N-acetylmuramyl-tripeptide (anhNAM-tripeptide). AmpD releases the tripeptide, which is then recycled for synthesis of the UDP-NAM pentapeptide (uridine 5′-pyrophosphoryl-N-acetylmuramic acid-pentapeptide), for integration into the neo-peptidoglycan. There is therefore a permanent balance between these two components in the cytoplasm.

Under normal conditions, AmpR associates with the UDM-NAM pentapeptide, and the resulting complex binds to the intercistronic region, thereby preventing transcription from both the ampR and ampC promoters. However, the presence of excess anhNAM-tripeptide alters this binding and promotes an increase in ampC transcription. Two clinical situations can result in an excessive increase in anhNAM-tripeptide levels: substantial degradation of the peptidoglycan due to the presence of β-lactams in the periplasmic space, and changes in the AmpD amidase preventing the recycling of this component.

In E. cloacae complex and C. freundii, the main cause of ampC overexpression is represented by amino-acid substitutions in AmpD, resulting in the constitutive production of large amounts of AmpC and increasing resistance to cefotaxime and ceftazidime. Several resistant clinical isolates or in vitro mutants displaying AmpC overproduction have been identified in the Enterobacterales, including C. freundii and E. cloacae with various amino-acid substitutions (e.g., W7G, H34A, S37R, Y63F, R80H, G82C, A94P/V, Y102, E116A, L117R, R108 D121G, D127G, N150I, H154N, A158D, K162H/Q, D164E/A, W171, A172L) capable of triggering constitutive β-lactamase production (178–182). In some cases, the ampD gene is truncated by a premature stop codon or an IS1 insertion, or an IS4321 element may be inserted into the promoter (178, 183, 184).

In P. aeruginosa, acquired resistance to amino- and ureido-penicillins, cephamycins, and, to a lesser extent, to oxyiminocephalosporins (ceftazidime, cefotaxime, ceftriaxone) and aztreonam, is very often related to overproduction of the PDC β-lactamase, which is also linked to virulence (185). Several variants with ampD mutations displaying resistance in vivo and in vitro have been reported (186–189). As in the Enterobacterales, the principal genetic mechanisms involved missense mutations creating amino-acid substitutions (A84G, D61Y, G148A, A136V, G148A, S175L) or insertion sequences (IS1669).

The LysR-type transcriptional regulator AmpR was initially identified as involved in regulating the inducible class C β-lactamases produced by various Gram-negative rods, such as C. freundii, E. cloacae, S. marcescens, and P. aeruginosa. However, recent results indicate that in P. aeruginosa AmpR regulates the expression of many genes involved in other pathways, such as quinolone resistance, quorum sensing, and associated virulence phenotypes (190). Phylogenetic analyses revealed the presence of AmpR homologs in many Alpha-, Beta-, and Gammaproteobacteria, and several highly conserved residues were found in C. freundii, E. cloacae, and P. aeruginosa (190, 191). AmpR loss leads to susceptibility to β-lactams, whereas constitutively high levels of ampC expression were reported for clinical isolates and in vitro strains with mutations resulting in AmpR amino-acid substitutions: C. freundii (S35F, G102D/E, D135A, Y264N) (4, 192, 193), E. cloacae (T84I, R86C, D135N/V, E274K) (178, 194), and P. aeruginosa (A12R, D135N, G154R, G237A) (186, 195–197). Transposon mutagenesis of P. aeruginosa strain PAO-1 and complementation experiments have generated two mutants with stronger β-lactam resistance due to transposition into two new genes (mpl, nuoN) (198).

AmpG, an intrinsic membrane protein, displays permease activity mediating the transport of muropeptides from the periplasm to the cytoplasm, such transport being essential for the induction of class C β-lactamases (199–201). Deletion of the gene encoding this protein in P. aeruginosa results in a lower level of bacterial resistance to ampicillin. AmpG proteins are widespread and highly variable in Gram-negative bacteria (e.g., Enterobacterales, A. baumannii, P. aeruginosa). Genetic experiments showed that ampG genes from E. coli and A. baumannii can complement AmpG function in P. aeruginosa (200). In this species, the site-directed mutagenesis of some highly conserved AmpG residues (G29A, G29V, A129V, A197S, and A197D mutants) results in a loss of resistance to ampicillin; ampG mRNA levels were found to be normal for two other mutants (A129T and A129D), but the proteins encoded had much lower levels of activity (200).

AmpC Overproduction in E. coli

In E. coli, in the absence of the ampR gene, the expression of ampC is usually weak, under the control of a naturally nonefficient promoter (4). In this species, acquired resistance to old or narrow-spectrum cephalosporins, cephamycins (e.g., cefoxitin), or even oxyiminocephalosporins, is related to ampC overexpression due to the selection of a stronger promoter by mutation, or, less often, by insertion (202, 203). The most important factors responsible for strengthening the ampC promoter are mutations creating a consensus -35 box (TTGACA) by T/A transversion at position -32 or C/T transition at position -42, and base-pair insertions increasing the distance between the -35 and -10 boxes to 17 or 18 bp (4, 203, 204). Mutations have also been identified within the attenuator region and the -10 box, but such mutations have little effect on ampC expression. Similar promoter mutations have been reported in strains from livestock, such as pigs or calves (205, 206).

AmpC Overproduction in A. baumannii

Overexpression of the chromosomal ampC gene in A. baumannii was observed following the acquisition of an IS element, mostly ISAba1, resulting in a strong promoter and resistance to expanded-spectrum cephalosporins (cefotaxime, ceftazidime) and aztreonam (207).

Transferable β-lactamases

Most plasmid-encoded class C β-lactamases mobilized by transposons and insertion sequence elements, with the exception of the ACT-1, CFE-1, and DHA-types, are in a “derepressed” state (31, 208–210). The genes encoding the ACC-1 and CMY-2 types are transcribed under the control of a strong promoter in the mobile element (ISEcp1 or IS26) (211, 212). The high level of bla_MIR-1 gene expression is due to the presence of a more efficient hybrid promoter upstream from the natural promoter (213). All these genes are often associated with integrons inserted in close proximity and forming composite structures with other transposable elements, including a gene cassette in a class 1 integron characterized by rapid spread (31). Gene cassettes are under the control of a constitutive promoter specific for the gene cassette array. Their expression level is determined not only by promoter strength, but also by distance from the promoter (214). The removal of ampC gene cassettes from the integron may constitute a less costly control mechanism than the continuous overproduction of many plasmid-borne class C enzymes (215). However, three plasmid-encoded enzyme types with an ampR gene (ACT-1, CFE-1, DHA-1) and a genetic organization identical to those of chromosomal enzymes rarely cause resistance to oxyiminocephalosporins despite constitutively high levels of AmpC activity due to an amino-acid substitution (D135A) (190, 193).

In conclusion, given the complexity of the regulation of ampC expression, the variability of these genes in databases and the small number of clinical examples of acquired resistance with the identification of hot spots for at least ampR, ampD, and ampG, it is not currently appropriate to analyze such mutants by genomic procedures, particularly in the absence of the initial susceptible clinical isolate. Classical methodologies are preferable for the detection of AmpC β-lactamase overproduction. These methods include phenotypic approaches assessing synergy between a cephamycin or an oxyiminocephalosporin and an inhibitor, such as cloxacillin or boronic acid (4). Nevertheless, the determination of β-lactamase activity with or without imipenem or cefoxitin induction, together with determinations of the mRNA levels for these genes by real-time RT-PCR, is probably much more accurate (69, 200, 216–218).

ACQUIRED EXTENDED-SPECTRUM CEPHALOSPORINASES (ESACs)

In Gram-negative bacteria, such as the Enterobacterales, various genetic events may increase the level of resistance, particularly to oxyiminocephalosporins, such as ceftazidime, cefotaxime, and, to a lesser extent, cefepime and cefpirome. The evolution of such resistance led to a new denomination: ESAC, for extended spectrum AmpC β-lactamases (219). The first ESAC was identified in an E. cloacae isolate (GC1) in Japan in 1992, based on the duplication of three amino acids at positions 208–210 (Ω-loop) and constitutive production (10).

In E. coli, AmpC is constitutively produced in small amounts in the absence of the ampR gene, under the control of a weak promoter (4). Within this species, acquired resistance to old or narrow-spectrum cephalosporins (e.g., cefoxitin) is related to the overproduction of AmpC due to the selection of a strong promoter generated by mutation, or, less often, by insertion (202, 203). A second step in the development of resistance involved the acquisition of plasmid-encoded AmpCs with a higher level of resistance to ceftazidime and aztreonam, for example (8). Finally, an even higher level of resistance (e.g., to cefepime or cefpirome) was developed following the emergence of an ESAC at low prevalence (≤1%) in human clinical isolates and even isolates from cattle (16, 220–222).

The genetic determinism of ESACs, leading to an increase in the catalytic efficiency of these enzymes against extended-spectrum cephalosporins, is based principally on missense mutations resulting in amino-acid substitutions, with structural modifications to the R1 (Ω-loop between residues 189 and 225) or R2 (H10 helix between residues 280 and 292 and/or R2-loop between residues 286 and 310) binding sites. Insertions or duplications in H10 or the R2 loop were also observed (Table 5).

TABLE 5.

Localization of ESAC-associated mutations in chromosomal or plasmid-encoded class C β-lactamases, the exact position of each mutation and the specific regions associated with each cluster

Bla or straina	Location	First numbering	Updated numberingb	References
E. coli
MEV	H10-helix	S282 duplication	idem	368
ECB33	H10-helix	I283 duplication	I284	223
HKY28	H10-helix	286GSD deletion	idem	225
EC16	H10-helix	S287C	idem	16
EC13	H10-helix	S287N	idem	16, 35
8009162	H10-helix, R2-loop	A292V	idem	221
EC80	H10-helix, R2-loop	L293P	idem	369
BER	R2-loop	293AA insertion	idem	370
7014517	R2-loop	295ALA insertion	idem	221
EC15	R2-loop	H296P	idem	16, 240
EC14	R2-loop	V298L	idem	16, 240
KL	H11-helix	V350F	idem	224
Citrobacter
CHA	close to YSN	R148H	idem + Q196H	35, 371
CMY-107	Ω-loop	Y199C	idem	372
CMY-27	Ω-loop	W221C	W201C	373
CMY-30	Ω-loop	V231G	V211G	374
CMY-42	Ω-loop	V231S	V211S	260, 375
CMY-95	Ω-loop	V211A	idem	369
CMY-32	Ω-loop	G214E	idem	376
CMY-54	Ω-loop	217EL insertion	216aEL insertion	377
GN346	Ω-loop	E219K	idem	378
CMY-136	Ω-loop	Y221H	idem	113
CMY-172	R2-loop	290KVA deletion+ N346I	idem	295
CMY-2	R2-loop	A292P or L293P	idem	379
CMY-33	R2-loop	293LA deletion	idem	235, 380
CMY-44	R2-loop	293LAAL deletion	idem	235
CMY-69	R2-loop	A295P	A294P	381
CMY-99	R2-loop	P306T	idem	382
CMY-37	R2-loop	L316I	L296I	383
Enterobacter
LN04004SS1	Ω-loop	213K-226G deletion	idem	384
GC1	Ω-loop	208AVR duplication	idem	10, 97
CHE	R2-loop	289SKVALA294 deletion	idem	9
Ent630	R2-loop	292-293 AL deletion	idem	294
P99 (ACT-89)	R2-loop	L293P	idem	385
MHN	R2-loop	V298E	idem	386
LN04004SS1	H11-helix	N366H	N346H	384
K. aerogenes
EA6/13/17/20	H3-helix	Q90H, W101C, L107Y	idem	387
Ea595	H10-helix, R2-loop	V291G	idem	388
Ear2	R2-loop	L293P	idem	338
S. marcescens
520R	H2-helix	T64I	T70I	389
SRT-1	Ω-loop	E213K	E219K	359
ES46, ES71	Ω-loop	E235K	E219K	390
SMSA	Ω-loop	S220Y	idem	391
HD	R2-loop	287MNGT deletion	293MNGT deletion	392
Hafnia
ACC-4g	Ω-loop	V211G + 289-290 deletion	idem	393, this review
ACC-1g	R2-loop	289-290 deletion	idem	25, 250, this review
ACC-2	R2-loop	289-290 deletionc	idem	250, this review
Aeromonas
CMY-9	R2-loop	E85D + 299-301 deletion	E61D + 301-303 deletion	394 – 396
CMY-19	R2-loop	I292S + 299-301 deletion	I292S + 301-303 deletion	236, 395, 396
CMY-1	R2-loop	299-301 deletion	301-303 deletiond	110, 236, 397
MOX-1	R2-loop	303-305 deletion	301-303 deletiond	156
MOX-2	R2-loop	303-305 deletion	301-303 deletione	234, this review
MOX-13	R2-loop	303-305 deletion + N346I	301-303 deletione	398, this review
CMY-10	R2-loop, H11-helix	E85D + 299-301 deletion	E61D + 301-303 deletion	110, 237, 395, 397
A. baumannii
ADC-56	H5-helix	R148Q	idem	57
ADC-53	Ω-loop	V208A	V211A	56
ADC-33	Ω-loop	P210R + 215A duplication	P213R + 218aA duplication	248
ADC-51	R2-loop	N283S	N287S	56
ADC-1	R2-loop	304-306 deletion	idem deletionf	53
ADC-68	Ω-loop	G220D + R320G	G217D + R321G	108
P. aeruginosa
PDC-222	H2-helix	T96I	T70I	230
PDC-78	H2-H3-loop, Ω-loop	R100H + G216R	R100H + R215G	69
PDC-82	H3-H4-loop	F121L + M175L	F121L + M174L	69, 229
PDC-73	H5-helix	P154L	P153L	69
PDC-82	H6-helix	M175L	M174L	69
PDC-50	Ω-loop	V213A	V211A	69
PDC-74	Ω-loop	G216R	G215R	69
PDC-86	Ω-loop, H7-helix	E221K	E219K	69
PDC-80	Ω-loop, H7-helix	E221G	E219G	69
PDC-85	Ω-loop, H7-helix	Y223H	Y221H	69
PDC-223	Ω-loop	229G-247E deletion	202G-219E deletion	230
PDC-221	Ω-loop	E247K	E219K	69
PDC-88	H10-helix, R2-loop	290TP deletion	289TP deletion	69
PDC-89	H10-helix, R2-loop	290TPM deletion	289TPM deletion	69
PDC-91	H10-helix, R2-loop	290TPMA deletion	289TPMA deletion	69
PDC-44	R2-loop	L294P	L293P	69
PDC-92	R2-loop	294LQ deletion	293LQ deletion	69
PDC-76	H11-helix	N347I	N346I	69

^aStrain names are underlined.

^bAccording to the Standard Numbering Scheme (25).

^cAll AmpC sequences of H. paralvei and H. alvei examined have this deletion.

^dThis cluster includes CMY-1, MOX-1, CMY-8, CMY-10, CMY-11, CMY-19, and MOX-14.

^eThese enzymes were located in different clusters (72).

^fThis deletion was observed for all A. baumannii sequences, and particularly for ADC-7 (51).

^gPlasmid-encoded.

Most ESACs belong to phylogenetic group A, but others belong to phylogenetic group B1 and are characterized by the presence of the following conserved residues: ²³⁵R, ²³⁸M, ²³⁹N, ²⁴¹R, ²⁴⁵D, ²⁸¹S, ²⁸⁸G, and ²⁹⁶H (16, 35, 221, 223). Nevertheless, two isolates with different residues in these positions have been reported (224). A single E. coli strain (HKY28) isolated from urine in Japan was found to produce an ESAC with a tripeptide deletion (²⁸⁶GS²⁸⁹D) altering binding to extended-spectrum cephalosporins, but also to sulbactam and tazobactam (225) (Table 5). Reversion of the deletion through a nine-base insertion restored the typical inhibitor-resistant phenotype of class C enzymes and decreased the level of resistance to cefepime and cefpirome. Finally, other positions affecting the ESAC phenotype were identified by the selection of mutants in vitro or by site-directed mutagenesis, resulting in a greater structural flexibility or affinity for extended-spectrum cephalosporins (141, 226).

Many BLCs (e.g., those of Pseudomonas, Citrobacter, Enterobacter, Morganella, and Serratia) are inducible, opening up possibilities for higher levels of production, resulting in stronger resistance to aminopenicillins, old cephalosporins, and ceftazidime or cefotaxime, and even to cefepime if the mutated enzyme is overproduced (227). Additionally, a significant proportion of strains has constitutive overproduction of class C enzymes, often after selection with β-lactam antibiotics. However, the risk of transitions between susceptibility and intermediate resistance (S/I) and between susceptibility and resistance (S/R) to cefepime is species-dependent. It is particularly high for the E. cloacae complex (66.3%), non-negligible for H. alvei (36.4%), moderate for Citrobacter and Proteus (18.1–21.9%) and S. marcescens (12.3%), low for K. aerogenes (1.1%), and inexistent for M. morganii (0%) (228). As summarized in Table 5, expansion of the inactivation spectrum results from missense mutations at several mutation hot spots (around triplet ¹⁵⁰YAN, Ω-loop, H10-helix, and R2-loop), from duplications or deletions of 2–4 amino acids in the H10 helix (position 293), or insertions of amino acids into the Ω-loop. In some cases, two such events may be combined (Table 5). Various examples of increased catalytic efficiency (k_cat/K_m) of these enzymes, particularly against extended-spectrum cephalosporins (ceftazidime, cefotaxime, cefepime), are illustrated in Table 6.

TABLE 6.

ESAC and effects on kinetic constants for groups of enzymes with different phenotypes^a

Bla or strainb	Mutation	Ceftazidime (CAZ)			Cefotaxime (CTX)			Cefepime (FEP)			References
		kcat (s−1)	Km (μM)	kcat/Km (μM · s−1)	kcat (s−1)	Km (μM)	kcat/Km (μM · s−1)	kcat (s−1)	Km (μM)	kcat/Km (μM · s−1)
CMY-2-like
CMY-2		0.004	0.5	0.008c	0.007	0.001	7				371
CMY-2	R148H	0.67	0.6	1.12c	NM	0.003	NM				371
CMY-2		0.01	0.02	0.5	<0.01	0.005	<2				372, 399
CMY-107		0.14	0.15	0.9	0.8	0.075	10.7				372
CMY-30		0.4	0.14	2.9	1.7	0.3	5.7				399
CMY-42		0.5	0.3	1.7	0.2	0.08	2.5c				375
CMY-2		0.005	0.15	0.033	0.007	0.001	7	0.37	412	9 · 10−4	113, 296, 371
CMY-136		6.26	2360	0.003	4.71	20	0.24	1.79	3588	5 · 10−4	113
CMY-2					NM	1.8	NM	NM	108.1	NM	376
CMY-32					0.9	4.05	0.22	NM	988.9	NM	376
ACC
ACC-2		0.03	5.2	0.006c	0.02	19	0.001	3.6	147	0.024	250
ACC-4		1.5	15	0.1	2.7	9.4	0.29	0.14	73	0.002	393
SRT
SerS		<0.5	50d	<0.01	2	4	0.5	<0.5	100d	<0.005c	391
SerR	S220Y	520	570	0.9	800	980	0.8	330	1000	0.33	391
FOX
FOX-3		0.273	1.18	0.231	0.081	0.076	1.06	ND	ND	3.32 · 10−3	241
FOX-8		2.6 · 10−3	0.382	6.8 · 10−3	18.2 · 10−3	10.95 · 10−3	1.66c	ND	ND	1.36 · 10−3	241
FOX-4		1.33	13	10	1.33	0.23	5.69	11.01	1071	0.01	245
FOX-4	306GNSΔ	0.84	6.44	0.13	0.24	0.087	2.75	2.5	103	0.02	245
MOX-1/CMY-1-like
CMY-9		1.8	560	3.2 · 10−3	0.27	0.28	0.96	NM	950	ND	236
CMY-19		0.085	3.7	0.023	0.33	31	0.011	1.8	630	2.9 · 10−3	236
CMY-8		0.091	48	1.9 · 10−3	0.36	2.3	0.16				394
CMY-9		0.53	120	4.4 · 10−3	0.48	3.4	0.14				394
MOX-1		ND	311	ND				ND	211	ND	232
CMY-1					0.01	0.015	0.67				400
CMY-10		5	33.9	0.15							110
ACT
ACT-89 (P99)		6.1 · 10−3	18.4	3.2 · 10−4							110
ACT-89 (P99)		<1	20d	ND	0.5	0.5d	1	1	15d	0.067	9
CHE	289-294Δ	<1	1d	ND	0.5	0.05d	10	2	3d	0.67	9
ACT-89 (P99)		0.065	28	2.3 · 10−3							401
ACT-89 (P99)	L293C	0.041	7	5.9 · 10−3							401
ACT-89 (P99)		0.013	15	8.7 · 10−4				0.5	100	4.7 · 10−3	385
ACT-89 (P99)	L293P	0.10	10	0.01				3.1	24	0.13	385
Ear
Ear		ND	16d	ND	0.15	>500	ND	0.4	126	0.003	338
Ear2		ND	9.8d	ND	0.15	10d	0.015	0.4	9.1	0.044	338
ADC
ADC-1		0.7	16.0	0.044	0.16	0.5	0.32				53, 108
ADC-68		1.66	147.7	0.01	18.5	117.5	0.16				108
ADC-11		0.01	10	0.001	0.2	2.5d	0.1	1	1800	5.5 · 10−4	248
ADC-33		4	30	0.13	1	0.5d	2	10	1300	7.7 · 10−3	248
ADC-1		1.255	265	4.7 · 10−3							207
ADC-5		0.011	232	4.7 · 10−5							207
ADC-5	P167S	2.5 · 10−3	120	2.1 · 10−5							207
ADC-5	P167S/D242G/ Q163K/G342R	1.235	90	0.014							207
ADC-30		0.05	1.39	0.04	0.18	0.51	0.32				57
ADC-56		0.1	1.42	0.07	0.27	1	0.27	0.2	17.17	0.011	57
PDC
PDC-1		0.004	20	2 · 10−4	0.02	6	3.3 · 10−3	0.08	800	1 · 10−4	68
PDC-2		0.01	20	5 · 10−4	0.15	5	0.03	2	850	2.4 · 10−3	68
PDC-3		0.02	35	5.7 · 10−4	0.15	8	0.019	2	1300	1.5 · 10−3	68
PDC-5		0.015	30	5 · 10−4	0.1	5	0.02	2.5	1700	1.5 · 10−3	68
PDC-5		0.01	7.3	1.3 · 10−3	0.07	0.14	0.5	>0.15	>250	6 · 10−4	402
PDC-5	N346Y	0.06	19	3.2 · 10−3	0.2	1.2	0.17	>0.17	>300	5.5 · 10−4	402

^aNM, k_cat not measurable; ND, not determined.

^bStrain names are underlined.

^cValue computed from k_cat and K_m, which is different from the value reported in the original paper.

^dK_i values (μM) were determined instead of K_m values, using cephalothin as a reporter substrate.

In P. aeruginosa, acquired resistance to amino- and ureidopenicillins, cephamycins, and, at low levels, to oxyiminocephalosporins (ceftazidime, cefotaxime, ceftriaxone) and monobactams (aztreonam), is frequently related to overproduction of the AmpC β-lactamase (14, 68, 185, 187, 229). ESACs emerged in P. aeruginosa several years after Enterobacterales, and such enzymes were identified mainly from clinical sources, but also from in vitro studies. They differ from the wild-type AmpC of P. aeruginosa by various amino-acid substitutions, deletions, or insertions in four regions in the vicinity of the active site: the Ω-loop, the H10-helix, the H2-helix, and the C-terminal end of the protein (69, 229, 230) (Table 5).

Genus Aeromonas may have been the origin of several chromosomal (e.g., MOX-3, MOX-10) or plasmid-encoded β-lactamases (e.g., MOX-1, MOX-2, CMY-1, CMY-8, CMY-19) identified on several occasions from clinical isolates, mostly in Enterobacterales (e.g., K. pneumoniae and E. coli). The first plasmid-encoded enzyme, MOX-1, was identified in a K. pneumoniae isolate with a high level of resistance to various broad-spectrum β-lactams, including moxalactam, flomoxef, ceftizoxime, cefotaxime, and ceftazidime (231). According to its kinetic parameters, cephalothin was its ideal substrate, and it had good activity against benzylpenicillin, but poor activity against cloxacillin and piperacillin. Moxalactam and cefoxitin were also hydrolyzed, but ceftazidime and cefepime were poor substrates, with very high K_m values (Table 6). Finally, aztreonam was found to inhibit MOX-1 (232). The X-ray crystallographic structure of MOX-1 suggested that residues 303–306 show a significant structural flexibility, possibly underlying the unique substrate profile of this enzyme, which can hydrolyze penicillins, cephalothin, expanded-spectrum cephalosporins, cefepime, and moxalactam (156). The position of the H10-helix in both MOX-1 and CMY-10 is shifted further away from the catalytic serine residue compared with the AmpCs from E. coli, E. hormaechei P99 (formerly known as E. cloacae P99), and E. cloacae GC1, whereas that of the AmpC from P. aeruginosa occupies an intermediate position. Such structural features may underlie the extended substrate profile of MOX-1 and CMY-10, which are also considered to be ESBLs, but have never been called ESACs (110, 156). Finally, the susceptibility profile of these enzymes is characterized by a high level of hydrolysis for cephamycins, oxacephems (e.g., cefoxitin), and moxalactam, increased resistance to cefotaxime compared with ceftazidime and aztreonam, and resistance to cefpirome and/or cefepime (233–239). The deletion of three residues in the R2-loop (positions 301–303) appears to be responsible for the expanded spectrum activity of CMY-10, and further mutation around this deletion in the P99 enzyme extended its substrate spectrum by widening the R2 region (Table 5) (110). Globally, all chromosomal (e.g., MOX-3, MOX-10, MOX-11) or plasmid-encoded (e.g., MOX-1, MOX-2, MOX-14, CMY-1, CMY-8, CMY-8b, CMY-9, CMY-10, CMY-11, CMY-19) β-lactamases from this family feature the sequence ²⁸⁹AKVILEANPTAAΔΔΔPRESG³⁰⁹S in the conserved R2 region (25). An increase in resistance to cefepime was also achieved by an additional amino-acid substitution (I292S) in the H10-helix region, as observed in CMY-11 and CMY-19 relative to CMY-9 (236) (Table 5).

The other enzymes from Aeromonas include a cluster of plasmid-encoded FOX-type enzymes without the amino-acid deletion at positions 301–303. These enzymes have a susceptibility profile characterized by a higher degree of resistance to cefoxitin (the origin of the family name, FOX) and to ceftazidime compared with cefotaxime (240–244). No ESACs have been identified among FOX variants (22). A tripeptide deletion (²⁸⁶GN²⁸⁸S) in the R2-loop of E. coli HKY28 led to an extended hydrolysis spectrum, but the same deletion by site-directed mutagenesis in FOX-4 did not increase catalytic efficiency for ceftazidime, cefotaxime, or cefepime, despite large differences in K_m and k_cat values (Table 6). A decrease in the MIC for cefoxitin was obtained in both E. coli HKY28 and FOX-4 Δ286-288, together with a slight increase in susceptibility to clavulanate, sulbactam, and tazobactam (245).

Structural studies have confirmed the importance of the Ω- and R2-loops for modulating the catalytic activity of class C β-lactamases. In CMY-10, a three-amino acid deletion in the R2-loop appears to underlie the extended spectrum activity (110), whereas a two-amino-acid deletion in the R2-loop of CMH-family AmpC enzymes from E. cloacae Ent385 leads to reduced susceptibility to ceftazidime-avibactam and cefiderocol (109). Similarly, the Y221H mutation in CMY-136 induces an important change in the confirmation of the Ω-loop, widening the active site cavity and conferring resistance to many β-lactams and combinations, including ceftolozane-tazobactam (113).

NATURAL ESACs

The term “naturally occurring ESAC” has been proposed for the enzymes in E. coli isolates with increased hydrolysis of oxyiminocephalosporins, including cefepime and cefpirome (MICs greater than or equal to 16 μg/mL and 0.5 μg/mL for ceftazidime and cefepime, respectively, without a positive synergy test with clavulanic acid), and resistance to amoxicillin and to amoxicillin-clavulanic acid (16). As for class A ESBLs, which are sometimes encoded by chromosomal genes (e.g., OXY-types for K. oxytoca) (246), both chromosomal and natural ESACs have been identified in A. baumannii (53), in which these β-lactamases are the principal source of resistance (49). The genes encoding ADC-type β-lactamases are non-inducible and generally expressed at low levels, resulting in resistance to penicillins and narrow-spectrum cephalosporins (51, 247). However, overexpression may be observed following the acquisition of an IS element, principally ISAba1, providing a strong promoter, and this may lead to resistance to expanded-spectrum cephalosporins (cefotaxime, ceftazidime) and aztreonam (207). All ADC-type β-lactamases (about 450 protein sequences examined) are characterized by a deletion of three residues (positions 304–306) in the R2-loop that enhances their catalytic efficiency against these clinically important drugs compared with enzymes without this deletion (53, 54, 108). Clinical variants with higher levels of resistance (4- to 64-fold increase) have been detected, with V211A (Ω-loop) or N287S (H10-helix) substitutions (56) (Table 5). Moreover, acquired resistance to cefepime or cefpirome and increased hydrolysis efficiency (ESAC) were induced by the P210R substitution and duplication of A215 in the Ω-loop (ADC-33) or by the substitution R148Q in the P2-loop (ADC-56) (57, 248). ADC-68 has seven amino-acid substitutions compared with ADC-1, one of which (³²¹G) is located in the C-loop and another two (¹⁹²A and ²¹⁷D) in the Ω-loop. The overall structures of ADC-68 and ADC-1 are conserved, but there are marked structural differences in the Ω-loop and the C-loop. In particular, residues ²¹⁷D and ³²¹G in ADC-68 make a major contribution to the structural differences between these two ADC-type β-lactamases (108).

There is a two-amino acid deletion (residues 301–302) in the R2-loop of AmpC from two species, A. dhakensis and A. caviae (25), but an equivalent deletion in AQU-1 does not seem to affect the usual β-lactam resistance phenotype. However, cefotaxime monotherapy should be used with caution for severe A. dhakensis infections, because such treatment could lead to the selection of variants with constitutively high levels of β-lactamase production. Nevertheless, cefepime susceptibility was conserved in all cases studied (249).

A specific structural alteration to the R2-loop (deletion of two amino acids in positions 289–290) was observed in all Hafnia protein sequences, and this feature is unique among Enterobacterales (25, 33, 88) (Table 5). Furthermore, ACC-1 confers an unusual and unique pattern of susceptibility to β-lactams, with higher MICs for ceftazidime and cefotaxime than for cefoxitin (4, 240). In addition to the two-amino-acid deletion in positions 289–290 mentioned above, ACC-1 features two other remarkable structural alterations, one in the Ω-loop (²¹³ME instead of ²¹³PG) and another along the active-site rim (¹²⁰F instead of ¹²⁰Q) (25, 88). Moreover, ACC-2, a cephalosporinase encoded by an inducible chromosomal gene, displays marked hydrolysis activity against cefpirome in particular (250), in opposition to the widely accepted view that cefpirome is resistant to hydrolysis by class C enzymes, even when they are overproduced (251). ACC-2 is also strongly inhibited by cefoxitin (250).

In conclusion, these natural ESACs constitute a threat because of their resistance to cefepime and/or cefpirome, or even to carbapenems if they are overproduced, combined with porin loss and/or efflux pump activation.

CARBAPENEM RESISTANCE

Carbapenem resistance has become a serious threat to public health (252). It has spread worldwide in Gram-negative bacteria, and is continuing to increase due to the production of plasmid-encoded carbapenemases by Enterobacterales and by nonfermenting organisms (mainly A. baumannii and P. aeruginosa) (253, 254). The main mechanism underlying carbapenem resistance is the production of highly transmissible carbapenemases, such as KPC-types (class A), IMP-, NDM-, or VIM-types (class B), and OXA-types (class D). Carbapenems, which diffuse efficiently across the native outer membrane of Gram-negative bacteria, are poor substrates for class C β-lactamases (4). However, they may act as inhibitors in some cases, due to the high affinity of AmpCs for these antibiotics, suggesting that resistance may be acquired by a mechanism known and sometimes described as “trapping” (255), in which the cephalosporinase has a good affinity for the substrate combined with a very slow hydrolysis (256–259).

Nevertheless, several clinical failures and in vitro studies have been published concerning chromosomally encoded and transmissible class C β-lactamases with weak carbapenem hydrolysis activity, suggesting poor carbapenemase activity (253, 260, 261). Such resistance was always obtained from the combination of a high production of β-lactamase (a hyperproducing mutant of a chromosomally-encoded enzyme or a plasmid-mediated enzyme) with another mutation-driven mechanism, such as porin loss (OmpC, OmpF, OmpG, OmpK35, OmpK36) or efflux overexpression, which was essential for carbapenem resistance, preferentially against ertapenem, then imipenem, and finally meropenem (240, 262) (Table 7). The prevalence of such nonenzymatic mechanisms is low and differs between bacteria, with relatively few occurrences among Enterobacterales (mostly in K. pneumoniae, but also in E. coli, P. mirabilis, and S. enterica).

TABLE 7.

Chromosomal or plasmid-encoded class C β-lactamases: mechanisms of acquired resistance to carbapenems

				Fold-increase MICsc					Reference
Bla or straina	Species	Locationb	Mechanisms	CAZ	FEP	IMP	ERT	MER
MEV	E. coli	Chr	ESAC + OmpC decrease + OmpF loss	>128	>16	8	>16	8	368
ACC-1	E. coli	P	+ OmpC/OmpF loss	>64	64	4	>128	4	240
ACT-1	E. coli	P	+ OmpC/OmpF loss	>256	>16	128	>512	32	240
CMY-2	E. coli	P	+ OmpC/OmpF loss	>256	>16	256	>512	256	240
DHA-1	E. coli	P	+ OmpC/OmpF loss	>512	64	16	256	8	240
FOX-1	E. coli	P	+ OmpC/OmpF loss	>512	>128	4	>256	16	240
CMY-2	E. coli	P	Overproduction + OmpC/OmpF loss	>256	>16	>64	>256	>32	403
CMY-2	E. coli	P	Overproduction + OmpC insertion IS1			>64		>256	260
CMY-2	E. coli	P	Overproduction + OmpC/OmpF loss	>256	>128	>256		>512	258
EC14	E. coli	Chr	ESAC + OmpC/OmpF loss	>256	>128	16	>32	1	404
CMY-4	S. enterica	P	+ OmpF loss	>512	>64	>16			405
ACT-1	K. pneumoniae	P	+ Omp42-kDa loss	>128	8-32	8->16			322
ACT-1	K. pneumoniae	P	+ OmpK35/36 insertion + PhoE decrease	64-256		128		256	406
DHA-1	K. pneumoniae	P	+ OmpK36 loss		>128	>32		>32	407
DHA-1	K. pneumoniae	P	+ OmpK35/36 loss + AcrAB/OqxAB			32	>32		277
ACC-1	K. pneumoniae	P	+ OmpK35/36 loss	>128	16	8	32	16	262
FOX-1	K. pneumoniae	P	+ OmpK35/36 loss	>512	32	64	128	64	262
MOX-1	K. pneumoniae	P	+ OmpK35/36 loss	64	8	32	32	32	262
EA-Z	K. aerogenes	Chr	+ Omp40-kDa loss			16			408
E15	K. aerogenes	Chr	+ OmpK35/36 loss	>256		>16	>16		409
E11	E. cloacae	Chr	Overproduction + OmpK35/36 loss	>256		>16	>16		409
144	E. cloacae	Chr	Overproduction + porin loss	64		16		>128	410
213	P. rettgeri	Chr	Overproduction + porin loss	4		8		>128	410
ACT-28	E. kobei	Chr	Overproduction + OmpC-like protein	>256	>64	>128	>512	>512	264
A-1	A. baumannii	Chr	Overproduction + Omp33/36 kDa decrease	32		>64			265
A(4) or B(4)	A. baumannii	Chr	Overproduction + Omp37/44/47 kDa decrease	>32	>32	8-16		>16	411
Paeβ-04	P. aeruginosa	Chr	Overproduction ESAC + mexB increase	32	16	16		16	275
3-D8	P. aeruginosa	Chr	Overproduction + mexB increase + OprD loss + dacB	128	64	32		>256	273
AM339	P. aeruginosa	Chr	Overproduction + mexA and mexC and mexX increase + OprD loss	>32	>128	>16	>16	>16	216

^aStrain names are underlined.

^bChr, chromosome; P, plasmid.

^cCAZ, ceftazidime; FEP, cefepime; IMP, imipenem; ERT, ertapenem; MER, meropenem.

Within the genus Enterobacter, carbapenem resistance is primarily due to the hyperproduction of chromosomal ampC associated with defective penetration of the carbapenemases in the bacterium (263). However, E. kobei from the E. cloacae complex group naturally produces an ACT-28-type AmpC β-lactamase that hydrolyzes imipenem more efficiently (264) and several plasmid-encoded AmpCs show intrinsically low carbapenemase activity (CMY-10, and to a lesser extent, CMY-2 and ACT-1) (240, 261).

Although in A. baumannii carbapenem resistance is mainly related to class B (VIM-, IMP-, and NDM-type) and class D (OXA-type) carbapenemases, the combination of overproduced class C β-lactamase and porin loss was also proposed for this species, which produces a natural ESAC with a deletion in the R2-loop and a low binding affinity for imipenem (53, 108, 265–267). No carbapenemase activity was detected in crude extracts, for example, so it was assumed that porin loss (CarO, Omp22-33, Omp33-36, Omp37, Omp43, Omp44, and Omp47) was the most likely mechanism (253, 265, 268–270).

Carbapenems (e.g., imipenem and meropenem) are important drugs to treat P. aeruginosa infections. Carbapenem resistance is mediated by various mechanisms, including the production of a carbapenemase (mostly class B enzymes), the overproduction of efflux pumps (mostly MexA-MexB), the overproduction of class C enzymes (e.g., PDC-type β-lactamases), and decreases in porin expression (mostly OprD), with combinations of these mechanisms observed in many cases (69, 216, 271, 272). In P. aeruginosa, transmissible class B carbapenemases or MBLs are considered to be the most clinically relevant source of resistance. Class D (oxacillinases) enzymes are also frequently encountered, and sometimes even class C β-lactamases, which are not real carbapenemases and have a low carbapenem hydrolysis capacity. The combination of PDC overproduction with a decrease in outer membrane permeability (OprD) and/or the overexpression of efflux systems (MexA-MexB-OprM) has been shown to lead to carbapenem resistance (69, 216, 271, 273–275) (Table 7). Various studies on clinical isolates have highlighted the difficulty of clearly defining the contribution of each mechanism, as resistance is generally a multifactorial phenomenon. However, most imipenem- and meropenem-resistant isolates display high levels of PDC-type enzymes and increased efflux pump expression (216).

RESISTANCE TO INHIBITORS

Early Inhibitors

The first inhibitors to be developed (e.g., clavulanate, sulbactam, tazobactam) were inactive against a majority of class C β-lactamases, but some older combinations have been revamped (e.g., ceftolozane-tazobactam) because ceftolozane has a reduced affinity for PDC enzymes and is not, therefore, hydrolyzed (276). Ceftolozane also has a high potency against P. aeruginosa when used alone. Resistance may be acquired during treatment and is mediated principally by transmissible β-lactamases (e.g., ESBLs and serine carbapenemases, such as KPCs). Several missense mutations or deletions in PDC genes have also been reported in P. aeruginosa isolates displaying overexpression of these genes and in studies performed in vitro, but the underlying mechanisms of resistance (such as combination with porin deficiency and overexpression of various efflux pumps) remain poorly defined (69, 229, 230, 277–282).

Avibactam

Novel non-β-lactam (diazabicyclooctane or DBO) β-lactamase inhibitors, such as avibactam, with a high affinity for a wide range of Ambler class A (e.g., ESBLs and KPC-types), class C, and some D β-lactamases (e.g., OXA-48), have been developed and showed encouraging results in clinical trials (276, 283). Avibactam is currently available in combination with ceftazidime, and is being used to combat multidrug-resistant Gram-negative bacterial infections, which are on the rise worldwide, including carbapenem-resistant Enterobacterales in particular. These inhibitors are not active against class B metallo-β-lactamases (e.g., NDM, VIM), and resistance to the ceftazidime–avibactam combination or combinations of avibactam with other antibiotics have been reported in patients treated since 2015 and in screens in vitro (283–289). This type of resistance has yet to be fully elucidated, but several possible underlying mechanisms have been reported. The most frequently encountered resistance mechanism is enzymatic, due to mutations identified mostly in transmissible class A (e.g., KPC and ESBLs) and some class D (e.g., OXA-48) β-lactamases. Other combined mechanisms have been reported, including membrane impermeability and efflux (277, 290), and there have been several reports of PBP3 mutations leading to ceftazidime-avibactam resistance (291–293). Finally, the overexpression of chromosomally- or plasmid-encoded AmpCs, with or without mutations and even in combination with porin deficiency, appears to make only a small contribution to resistance (13, 14, 160). Avibactam inhibits the E. cloacae CHE ESAC less efficiently than other E. cloacae AmpC proteins, due to a subtle rearrangement of the binding site (13). In vitro studies with P. aeruginosa showed that high-level resistance to combinations of avibactam with ceftazidime or aztreonam occurs with low frequency. In all cases, the mutant displayed alterations to the chromosomal ampC gene, mostly deletions of various sizes (5 to 19 residues) in the Ω-loop region of AmpC (14). Eight highly conserved residues made key contributions to binding interactions with avibactam (⁶⁴S, ⁶⁷K, ¹²⁰Q, ¹⁵⁰Y, ¹⁵²N, ³¹⁵K, ³¹⁶T, and ³⁴⁶N). The PAC-1 enzyme, recently identified in four P. aeruginosa isolates and encoded by a gene on a chromosome-inserted Tn1721-like transposon, mediates very high levels of resistance to β-lactams, such as ceftazidime, cefepime, and ceftolozane alone or in combination with avibactam or tazobactam (75). Resistance to the ceftazidime–avibactam combination has been attributed to a deletion of two amino acids in the R2 loop of the AmpC β-lactamase produced independently by Enterobacter, which simultaneously causes resistance to cefepime and carbapenems and reduced susceptibility to cefiderocol, a novel siderophore cephalosporin (294).

Six carbapenem-resistant clinical isolates of K. pneumoniae in China were recently shown to produce KPC-2, a class A carbapenemase, and CMY-172, a new CMY-2-like class C β-lactamase (295). CMY-172 mediates high levels of resistance to β-lactams, including ceftazidime, cefotaxime, and cefepime, and to the ceftazidime–avibactam combination. This resistance results from a major modification (²⁹⁰KVA deletion) to the R2-loop (ESAC) and an amino-acid substitution, N346I (Table 5). The variant retains all the highly conserved residues for binding to avibactam (⁶⁴S, ⁶⁷K, ¹²⁰Q, ¹⁵⁰Y, ¹⁵²N, ³¹⁵K, ³¹⁶T) except for ³⁴⁶N (162). In addition, the residue in position 346, which is well conserved among AmpC-type enzymes, modulates the hydrolysis spectrum of cephalosporinases (296). More generally, in vitro resistance to ceftazidime-avibactam and aztreonam-avibactam was examined among Enterobacterales, showing that resistance associated with changes to β-lactamases was seen only for mutants of AmpC. The mutants R148H, G156R/D, N346Y, or small deletions at positions 289–294, were obtained with ceftazidime-avibactam, whereas with aztreonam-avibactam the Y150C or N346Y substitutions were observed (297).

Relebactam

The newer imipenem-relebactam combination recently approved by the FDA has antimicrobial activity against Enterobacterales and P. aeruginosa strains producing class A and C β-lactamases, through the strong inhibition of KPC and AmpC enzymes (276, 298). Isolates resistant to this combination, due to porin loss, have been reported in Enterobacteriaceae (299, 300). There are in vivo and in vitro examples of resistance to this combination in P. aeruginosa mediated by class C enzymes, but resistance was again multifactorial, dependent on OprD loss, cephalosporinase overproduction, and mutations of the genes encoding the MexAB-OprM efflux system pump and the peptidoglycan recycling machinery (276, 301, 302).

Vaborbactam

Meropenem-vaborbactam is a fixed-dose combination of a carbapenem antibiotic and a novel boronic acid-based β-lactamase inhibitor. It has in vitro activity against Enterobacterales (e.g., K. pneumoniae) producing class A (e.g., ESBL, KPC), class C (mostly plasmid-encoded enzymes), and also some class D enzymes (103, 105, 121, 303). The impact of vaborbactam on class C enzymes appears to be limited, but resistance may nevertheless result from combinations of mechanisms, including the overexpression of ESAC mutants after cefepime treatment, associated with OmpK35 and OmpK36 porin deficiencies due to the insertion of an IS903-like element, as recently reported in a clinical isolate of E. hormaechei (294). This isolate was multidrug-resistant, with resistance even against ceftazidime-avibactam, meropenem-vaborbactam, and imipenem-relebactam combinations. This resistance should serve as an important warning that cephalosporins and carbapenems should be used in a judicious manner, whether used alone or in combination with these novel inhibitors.

NEW CEPHALOSPORINS

Cefiderocol

Cefiderocol is an advanced injectable siderophore cephalosporin active against MDR and XDR Gram-negative rods, including producers of various β-lactamases, including members of Enterobacterales, P. aeruginosa, A. baumannii, Stenotrophomonas maltophila, B. cepacia, and B. pseudomallei (304–307). Through its catechol siderophore moiety, cefiderocol exploits iron-transport systems via a Trojan-horse strategy, navigating the bacterial periplasm and evading various β-lactamases, including ESBLs, KPC (class A), NDM, IMP, VIM (class B), and various class C and D enzymes (OXA-23, OXA-24, OXA-48, and OXA-51), as well as other mechanisms of resistance (308–310). Resistance to cefiderocol may be due to a two-amino-acid deletion in the R2 loop of the AmpC of E. cloacae complex (109, 294), but can also be acquired through the loss of energy-transducing proteins or catecholate receptors, such as PirA, PiuA, PiuD, and TonB in Enterobacterales, P. aeruginosa and A. baumannii (309, 311–313).

CONCLUSIONS

BLCs have a highly variable primary structure, with minimum sequence identity levels as low as 20% (28). However, the three principal catalytic motifs that are characteristic for this molecular class (⁶⁴SXSK, ¹⁵⁰YXN, and ³¹⁵KTG) are highly conserved, together with at least 70 other residues displaying >90% conservation. Some of these other residues, including ⁶⁰F, ⁶¹E, ⁶³G, ¹⁴⁵G, ¹⁴⁸R, ³¹⁸T, ³²¹G, ³²²F, ³²⁵Y, and ³²⁸F, are located around the conserved catalytic motifs. Generally, with the exception of a few sequences corresponding to serine hydrolases often identified by BLAST analysis, structural classification should be based on the notion of “genus.” Genera contain highly variable numbers of species, and future taxonomic reorganizations are possible, as already observed for the genus Erwinia, for example, with the discovery of E. teleogrylli, or for the genus Pseudomonas, with definition of the species P. putida and P. fluorescens.

The considerable increase in the number of protein sequences present in databases for several species has made it possible to confirm the existence of a certain degree of polymorphism, the large number of clusters observed (more than 10 for E. coli/Shigella) making it possible to define consensus sequences for each cluster or phylogroup (16). By contrast, enzymes of the DHA- and FOX-types, encoded by plasmid-borne genes, and the species from which they originate, M. morganii and A. allosaccharophila, respectively, display little polymorphism.

With the development of high-throughput sequencing, this notion of ampC gene polymorphism complicates the detection of variants responsible for a significant increase in the MICs of β-lactams, particularly for oxyiminocephalosporins (e.g., cefotaxime, ceftazidime, cefepime, and cefpirome), monobactams (e.g., aztreonam), and carbapenems. However, most infections are identified in hospital inpatients, from whom several isolates of the same species can be obtained, therefore facilitating the detection of such variants. Nevertheless, in the absence of an isolate obtained at the start of treatment, certain variants, such as insertions, duplications, or deletions, appear to be easy to detect, whereas others, such as the substitution of a single residue in a hot spot, may be much harder to identify. The diverse examples of ESAC-producing strains observed in clinical practice demonstrate that genetic modifications similar to those described above essentially affect three regions (around the ¹⁵⁰YXN motif, the Ω-loop, and the H10-helix/R2-loop). Surprisingly, these regions vary to different extents in different species, with a limited number of conserved residues. The notion of a “consensus sequence” should therefore favor the possible detection of ESACs in the framework of increasingly frequent genotype/phenotype analyses (Fig. 3). Such diversity at the heart of species, particularly for the H10-helix/R2-loop region, could account for the considerable variation of MICs between substrates, or the diversity of the β-lactam resistance and kinetic constants reported. Wild-type bacterial species have variable MICs for β-lactams (https://mic.eucast.org/), suggesting that combinations of other, non-enzymatic resistance mechanisms may be at work.

FIG 3.

Consensus partial protein sequences of species or their progenitors susceptible to expand their spectrum of inactivation. For E. coli, the consensus sequence was calculated from protein sequences of clusters A, B, C and D (32, 46, 50, 63). Residue boxed in yellow indicates 100% conserved. Underlined positions indicate at least two different residues (polymorphism).

It should not be forgotten that the chromosomal ampC gene is subject to regulation, and thus is dependent on other genes, particularly as it is inducible in most of the bacterial species producing the corresponding enzyme. Diverse genetic events can, therefore, occur in the intercistronic space, generating a strong promoter, or in other genes, such as ampR, ampD, and ampG, to generate mutants displaying derepression or overexpression. The analysis would not really be complete without an examination of several systems involved in the transfer of β-lactams across the cell wall, such as porins (e.g., OmpC, OmpF, OmpK35, OmpK36), and efflux systems (e.g., AcrAB-TolC, MexAB-OprM) (293, 314–318).

For medical biologists, the identification of a β-lactamase from one of the four known molecular classes is based on the presence of at least three motifs (17). The accumulation of a large number of protein sequences in current databases, with the development of WGS in particular, has made it possible to compare almost 4,000 protein sequences from class C β-lactamases, facilitating classification within clusters or taxa. This class is as diverse as class A, with about 20% identity between sequences, but its genetic footprint may be enriched in fewer than 80 highly conserved residues, justifying the exclusion of certain carboxylesterases with β-lactamase activity. In addition, the existence of major taxa in bacterial genera, such as Citrobacter, Enterobacter, Acinetobacter, and Aeromonas, and of recent and more long-standing taxonomic reorganizations, provides an indispensable aid to diagnosis at species level, particularly in light of the potential failings of the phenotypic methods currently used in laboratories. This type of analysis, based on primary structure, makes it possible to propose one or several consensus sequences for a species, potentially improving the characterization of possible polymorphisms of the β-lactamase. In the absence of the susceptible strain of the bacterium at the start of infection, prior knowledge of the diverse genetic modifications likely to lead to an extended spectrum of inactivation, which vary between bacterial species, should make it possible to detect these modifications (substitutions, duplications, insertions, deletions). The recent adaptation of a numbering system for this molecular class of enzymes has led to comparisons of protein sequences and emergence of the notion of a “natural ESAC.” Finally, medical biologists should not ignore possible interactions with other genes likely to lead to hyperproduction, which are particularly important for this molecular class.

Biographies

Alain Philippon is a veterinarian (Ecole Nationale Vétérinaire d’Alfort, 1963) graduated in biochemistry (University of Orsay, 1965) and bacteriology-immunology (Institut Pasteur, 1966) who prepared his Ph. D. thesis at the Commissariat à l’Energie Atomique (CEA). Initially trained on Brucella and experimental bovine brucellosis at Nouzilly (Institut National de la Recherche Agronomique/INRA) between 1966 and 1970, he started to analyze bacterial resistance to antibiotics, mostly to β-lactams (susceptibility patterns, ESBL, and plasmid-encoded AmpC) at CHU Cochin, Paris, France, and in collaboration with François Le Goffic and Roger Labia (ENS Ulm) in 1972. As professor emeritus of Microbiology at the University of Paris Descartes and previous head of the Bacteriology Laboratory at Hopital Cochin, he published more than 200 scientific papers and mentored 20 doctoral and postdoctoral researchers. He was also codirector of a course on medical bacteriology at the Institut Pasteur of Paris for a decade.

Guillaume Arlet is a doctor (René Descartes University, Paris, 1979), graduate in Medical Biology in 1984, in Biochemistry and Molecular Biology in 1986 (Denis Diderot University, Paris), and Bacteriology (Institut Pasteur, 1986). He did his PhD (University Paris-Sud, 1989–1992) under the supervision of Alain Philippon. Associate professor at Saint-Louis Hospital and Denis Diderot University (1991–1999), he became professor at Pierre & Marie Curie University (now Sorbonne University), and head of the medical bacteriology department, first at Tenon Hospital (1999–2012) and then within the Eastern Parisian Hospitals Group (2012–2018). He headed the teaching department of Clinical Bacteriology (2007–2018). He joined in 2014 the Center for Immunology and Infectious Diseases (INSERM U1135) at Sorbonne Université. He supervised ten PhD students and published more than 210 scientific papers mainly on β-lactamases (ESBL, AmpC β-lactamases, carbapenemases), their genetic supports, and in relation to virulence of their bacterial hosts. He is currently professor emeritus.

Roger Labia graduated from the world-renowned Ecole Polytechnique in Paris, France (1962 to 1964), with specialization in mathematics, physics, and chemistry. In relation to his high and early interest in chemical and biological problems, he joined scientific research, first at the Pasteur Institute (starting in 1965). In 1969, he received a PhD, studying the organic synthesis of natural compounds. Subsequently, he spent one postdoctoral year at Ottawa University, Canada (1969 to 1970), where he began studying biochemistry and bacteriology. Back in France, he started a research program on antibiotics, including their mode of action and mechanisms of resistance. This allowed him to develop multiple collaborations with chemists and bacteriologists from France and other countries. He has published more than 320 scientific papers, mostly in high-impact international journals. He has been involved in teaching and directed about 50 theses.

Bogdan I. Iorga received in 2001 a PhD in molecular chemistry from the Ecole Polytechnique, France. He is currently CNRS Research Director and heads the Molecular Modeling and Structural Crystallography team at the Institut de Chimie des Substances Naturelles (CNRS, Université Paris-Saclay, Gif-sur-Yvette, France). He is author of one book, six book chapters, four patents, and more than 100 publications. He is the developer of the Beta-Lactamase DataBase (BLDB, http://bldb.eu) and is involved in several French and European projects related to the antimicrobial resistance. His main research interests include the design of biologically active compounds, the study of structure–function relationships in different classes of β-lactamases, and the development of innovative protocols in molecular modeling. His recent work focuses on the development of tools for in silico prediction of antibiotic susceptibility from genomic data using machine learning and deep learning approaches.