Network Analysis of Sequence-Function Relationships and Exploration of Sequence Space of TEM β-Lactamases

Understanding the effects of sub-inhibitory antibiotic concentrations on the development of β-lactamase resistance based on quantile regression analysis

AIMS

Quantile regression is an alternate type of regression analysis that has been shown to have numerous advantages over standard linear regression. Unlike linear regression, which uses the mean to fit a linear model, quantile regression uses a data set's quantiles (or percentiles), which leads to a more comprehensive analysis of the data. However, while relatively common in other scientific fields such as economic and environmental modeling, it is infrequently used to understand biological and microbiological systems.

METHODS AND RESULTS

We analyzed a set of bacterial growth rates using quantile regression analysis to better understand the effects of antibiotics on bacterial fitness. Using a bacterial model system containing 16 variant genotypes of the TEM β-lactamase enzyme, we compared our quantile regression analysis to a previously published study that uses the Tukey's range test, or Tukey honestly significantly difference (HSD) test. We find that trends in the distribution of bacterial growth rate data, as viewed through the lens of quantile regression, can distinguish between novel genotypes and ones that have been clinically isolated from patients. Quantile regression also identified certain combinations of genotypes and antibiotics that resulted in bacterial populations growing faster as the antibiotic concentration increased-the opposite of what was expected. These analyses can provide new insights into the relationships between enzymatic efficacy and antibiotic concentration.

CONCLUSIONS

Quantile regression analysis enhances our understanding of the impacts of sublethal antibiotic concentrations on enzymatic (TEM β-lactamase) efficacy and bacterial fitness. We illustrate that quantile regression analysis can link patterns in growth rates with clinically relevant mutations and provides an understanding of how increasing sub-lethal antibiotic concentrations, like those found in our modern environment, can affect bacterial growth rates, and provide insight into the genetic basis for varied resistance.

Screening of global microbiomes implies ecological boundaries impacting the distribution and dissemination of clinically relevant antimicrobial resistance genes

Understanding the myriad pathways by which antimicrobial-resistance genes (ARGs) spread across biomes is necessary to counteract the global menace of antimicrobial resistance. We screened 17939 assembled metagenomic samples covering 21 biomes, differing in sequencing quality and depth, unevenly across 46 countries, 6 continents, and 14 years (2005-2019) for clinically crucial ARGs, mobile colistin resistance (mcr), carbapenem resistance (CR), and (extended-spectrum) beta-lactamase (ESBL and BL) genes. These ARGs were most frequent in human gut, oral and skin biomes, followed by anthropogenic (wastewater, bioreactor, compost, food), and natural biomes (freshwater, marine, sediment). Mcr-9 was the most prevalent mcr gene, spatially and temporally; blaOXA-233 and blaTEM-1 were the most prevalent CR and BL/ESBL genes, but blaGES-2 and blaTEM-116 showed the widest distribution. Redundancy analysis and Bayesian analysis showed ARG distribution was non-random and best-explained by potential host genera and biomes, followed by collection year, anthropogenic factors and collection countries. Preferential ARG occurrence, and potential transmission, between characteristically similar biomes indicate strong ecological boundaries. Our results provide a high-resolution global map of ARG distribution and importantly, identify checkpoint biomes wherein interventions aimed at disrupting ARGs dissemination are likely to be most effective in reducing dissemination and in the long term, the ARG global burden.

Crystal structures of the molecular class A β-lactamase TEM-171 and its complexes with tazobactam

The resistance of bacteria to β-lactam antibiotics is primarily caused by the production of β-lactamases. Here, novel crystal structures of the native β-lactamase TEM-171 and two complexes with the widely used inhibitor tazobactam are presented, alongside complementary data from UV spectroscopy and fluorescence quenching. The six chemically identical β-lactamase molecules in the crystallographic asymmetric unit displayed different degrees of disorder. The tazobactam intermediate was covalently bound to the catalytic Ser70 in the trans-enamine configuration. While the conformation of tazobactam in the first complex resembled that in published β-lactamase–tazobactam structures, in the second complex, which was obtained after longer soaking of the native crystals in the inhibitor solution, a new and previously unreported tazobactam conformation was observed. It is proposed that the two complexes correspond to different stages along the deacylation path of the acyl-enzyme intermediate. The results provide a novel structural basis for the rational design of new β-lactamase inhibitors.

Allogenous Selection of Mutational Collateral Resistance: Old Drugs Select for New Resistance Within Antibiotic Families

Allogeneous selection occurs when an antibiotic selects for resistance to more advanced members of the same family. The mechanisms of allogenous selection are (a) collateral expansion, when the antibiotic expands the gene and gene-containing bacterial populations favoring the emergence of other mutations, inactivating the more advanced antibiotics; (b) collateral selection, when the old antibiotic selects its own resistance but also resistance to more modern drugs; (c) collateral hyper-resistance, when resistance to the old antibiotic selects in higher degree for populations resistant to other antibiotics of the family than to itself; and (d) collateral evolution, when the simultaneous or sequential use of antibiotics of the same family selects for new mutational combinations with novel phenotypes in this family, generally with higher activity (higher inactivation of the antibiotic substrates) or broader spectrum (more antibiotics of the family are inactivated). Note that in some cases, collateral selection derives from collateral evolution. In this article, examples of allogenous selection are provided for the major families of antibiotics. Improvements in minimal inhibitory concentrations with the newest drugs do not necessarily exclude “old” antibiotics of the same family of retaining some selective power for resistance to the newest agents. If this were true, the use of older members of the same drug family would facilitate the emergence of mutational resistance to the younger drugs of the family, which is frequently based on previously established resistance traits. The extensive use of old drugs (particularly in low-income countries and in farming) might be significant for the emergence and selection of resistance to the novel members of the family, becoming a growing source of variation and selection of resistance to the whole family. In terms of future research, it could be advisable to focus antimicrobial drug discovery more on the identification of new targets and new (unique) classes of antimicrobial agents, than on the perpetual chemical exploitation of classic existing ones.

Ancestral sequences of a large promiscuous enzyme family correspond to bridges in sequence space in a network representation

Evolutionary relationships of protein families can be characterized either by networks or by trees. Whereas trees allow for hierarchical grouping and reconstruction of the most likely ancestral sequences, networks lack a time axis but allow for thresholds of pairwise sequence identity to be chosen and, therefore, the clustering of family members with presumably more similar functions. Here, we use the large family of arylsulfatases and phosphonate monoester hydrolases to investigate similarities, strengths and weaknesses in tree and network representations. For varying thresholds of pairwise sequence identity, values of betweenness centrality and clustering coefficients were derived for nodes of the reconstructed ancestors to measure the propensity to act as a bridge in a network. Based on these properties, ancestral protein sequences emerge as bridges in protein sequence networks. Interestingly, many ancestral protein sequences appear close to extant sequences. Therefore, reconstructed ancestor sequences might also be interpreted as yet-to-be-identified homologues. The concept of ancestor reconstruction is compared to consensus sequences, too. It was found that hub sequences in a network, e.g. reconstructed ancestral sequences that are connected to many neighbouring sequences, share closer similarity with derived consensus sequences. Therefore, some reconstructed ancestor sequences can also be interpreted as consensus sequences.

Extended-spectrum β-lactamases: an update on their characteristics, epidemiology and detection

Extended-spectrum β-lactamase (ESBL)-producing Gram-negative pathogens are a major cause of resistance to expanded-spectrum β-lactam antibiotics. Since their discovery in the early 1980s, they have spread worldwide and an are now endemic in Enterobacterales isolated from both hospital-associated and community-acquired infections. As a result, they are a global public health concern. In the past, TEM- and SHV-type ESBLs were the predominant families of ESBLs. Today CTX-M-type enzymes are the most commonly found ESBL type with the CTX-M-15 variant dominating worldwide, followed in prevalence by CTX-M-14, and CTX-M-27 is emerging in certain parts of the world. The genes encoding ESBLs are often found on plasmids and harboured within transposons or insertion sequences, which has enabled their spread. In addition, the population of ESBL-producing Escherichia coli is dominated globally by a highly virulent and successful clone belonging to ST131. Today, there are many diagnostic tools available to the clinical microbiology laboratory and include both phenotypic and genotypic tests to detect β-lactamases. Unfortunately, when ESBLs are not identified in a timely manner, appropriate antimicrobial therapy is frequently delayed, resulting in poor clinical outcomes. Several analyses of clinical trials have shown mixed results with regards to whether a carbapenem must be used to treat serious infections caused by ESBLs or whether some of the older β-lactam-β-lactamase combinations such as piperacillin/tazobactam are appropriate. Some of the newer combinations such as ceftazidime/avibactam have demonstrated efficacy in patients. ESBL-producing Gram-negative pathogens will continue to be major contributor to antimicrobial resistance worldwide. It is essential that we remain vigilant about identifying them both in patient isolates and through surveillance studies.

A clinically important, plasmid-borne antibiotic resistance gene (β-lactamase TEM-116) present in desert soils

The exhaustive use of antibiotics in humans, animal farming and other agricultural practices has resulted in the frequent appearance of antibiotic resistant bacteria in human-impacted habitats. However, antibiotic resistance in natural (less-impacted) habitats is less understood. Using shotgun metagenomics we analysed soils from relatively low anthropogenic impact sites across the Namib Desert. We report the presence of a clinically significant extended spectrum β-lactamase (TEM-116), on a ColE1-like plasmid also carrying a metal resistance gene (arsC). The co-occurrence of resistance to antimicrobial drugs and metals encoded on a single mobile genetic element increases the probability of dissemination of these resistance determinants and the potential selection of multiple resistance mechanisms. In addition, the presence of a P7 entero-bacteriophage on the same plasmid, may represent a new vehicle for the propagation of TEM-116 in these soil communities. These findings highlight the role of the environment in the One Health initiative.

OXA-1 β-lactamase and non-susceptibility to penicillin/β-lactamase inhibitor combinations among ESBL-producing Escherichia coli

Background

ESBL-producing Escherichia coli have expanded globally since the turn of the century and present a major public health issue. Their in vitro susceptibility to penicillin/inhibitor combinations is variable, and clinical use of these combinations against ESBL producers remains controversial. We hypothesized that this variability related to co-production of OXA-1 penicillinase.

Methods

During a national study we collected 293 ESBL-producing E. coli from bacteraemias, determined MICs by BSAC agar dilution, and undertook genomic sequencing with Illumina methodology.

Results

The collection was dominated by ST131 (n = 188 isolates, 64.2%) and blaCTX-M-15 (present in 229 isolates, 78.2%); over half the isolates (159/293, 54.3%) were ST131 with blaCTX-M-15. blaOXA-1 was found in 149 ESBL producers (50.9%) and blaTEM-1/191 in 137 (46.8%). Irrespective of whether all isolates were considered, or ST131 alone, there were strong associations (P < 0.001) between co-carriage of blaOXA-1 and reduced susceptibility to penicillin/inhibitor combinations, whereas there was no significant association with co-carriage of blaTEM-1/191. For piperacillin/tazobactam the modal MIC rose from 2 mg/L in the absence of blaOXA-1 to 8 or 16 mg/L in its presence; for co-amoxiclav the shift was smaller, from 4 or 8 to 16 mg/L, but crossed the breakpoint. blaOXA-1 was strongly associated with co-carriage also of aac(6')-Ib-cr, which compromises amikacin and tobramycin.

Conclusions

Co-carriage of OXA-1, a penicillinase with weak affinity for inhibitors, is a major correlate of resistance to piperacillin/tazobactam and co-amoxiclav in E. coli and is commonly associated with co-carriage of aac(6')-Ib-cr, which narrows aminoglycoside options.

The Structural Dynamics of Engineered β-Lactamases Vary Broadly on Three Timescales yet Sustain Native Function

Understanding the principles of protein dynamics will help guide engineering of protein function: altering protein motions may be a barrier to success or may be an enabling tool for protein engineering. The impact of dynamics on protein function is typically reported over a fraction of the full scope of motional timescales. If motional patterns vary significantly at different timescales, then only by monitoring motions broadly will we understand the impact of protein dynamics on engineering functional proteins. Using an integrative approach combining experimental and in silico methodologies, we elucidate protein dynamics over the entire span of fast to slow timescales (ps to ms) for a laboratory-engineered system composed of five interrelated β-lactamases: two natural homologs and three laboratory-recombined variants. Fast (ps-ns) and intermediate (ns-µs) dynamics were mostly conserved. However, slow motions (µs-ms) were few and conserved in the natural homologs yet were numerous and widely dispersed in their recombinants. Nonetheless, modified slow dynamics were functionally tolerated. Crystallographic B-factors from high-resolution X-ray structures were partly predictive of the conserved motions but not of the new slow motions captured in our solution studies. Our inspection of protein dynamics over a continuous range of timescales vividly illustrates the complexity of dynamic impacts of protein engineering as well as the functional tolerance of an engineered enzyme system to new slow motions.

Past and Present Perspectives on β-Lactamases

β-Lactamases, the major resistance determinant for β-lactam antibiotics in Gram-negative bacteria, are ancient enzymes whose origins can be traced back millions of years ago. These well-studied enzymes, currently numbering almost 2,800 unique proteins, initially emerged from environmental sources, most likely to protect a producing bacterium from attack by naturally occurring β-lactams. Their ancestors were presumably penicillin-binding proteins that share sequence homology with β-lactamases possessing an active-site serine. Metallo-β-lactamases also exist, with one or two catalytically functional zinc ions. Although penicillinases in Gram-positive bacteria were reported shortly after penicillin was introduced clinically, transmissible β-lactamases that could hydrolyze recently approved cephalosporins, monobactams, and carbapenems later became important in Gram-negative pathogens. Nomenclature is based on one of two major systems. Originally, functional classifications were used, based on substrate and inhibitor profiles. A later scheme classifies β-lactamases according to amino acid sequences, resulting in class A, B, C, and D enzymes. A more recent nomenclature combines the molecular and biochemical classifications into 17 functional groups that describe most β-lactamases. Some of the most problematic enzymes in the clinical community include extended-spectrum β-lactamases (ESBLs) and the serine and metallo-carbapenemases, all of which are at least partially addressed with new β-lactamase inhibitor combinations. New enzyme variants continue to be described, partly because of the ease of obtaining sequence data from whole-genome sequencing studies. Often, these new enzymes are devoid of any phenotypic descriptions, making it more difficult for clinicians and antibiotic researchers to address new challenges that may be posed by unusual β-lactamases.

The scale-free nature of protein sequence space

The sequence space of five protein superfamilies was investigated by constructing sequence networks. The nodes represent individual sequences, and two nodes are connected by an edge if the global sequence identity of two sequences exceeds a threshold. The networks were characterized by their degree distribution (number of nodes with a given number of neighbors) and by their fractal network dimension. Although the five protein families differed in sequence length, fold, and domain arrangement, their network properties were similar. The fractal network dimension Df was distance-dependent: a high dimension for single and double mutants (Df = 4.0), which dropped to Df = 0.7-1.0 at 90% sequence identity, and increased to Df = 3.5-4.5 below 70% sequence identity. The distance dependency of the network dimension is consistent with evolutionary constraints for functional proteins. While random single and double mutations often result in a functional protein, the accumulation of more than ten mutations is dominated by epistasis. The networks of the five protein families were highly inhomogeneous with few highly connected communities ("hub sequences") and a large number of smaller and less connected communities. The degree distributions followed a power-law distribution with similar scaling exponents close to 1. Because the hub sequences have a large number of functional neighbors, they are expected to be robust toward possible deleterious effects of mutations. Because of their robustness, hub sequences have the potential of high innovability, with additional mutations readily inducing new functions. Therefore, they form hotspots of evolution and are promising candidates as starting points for directed evolution experiments in biotechnology.

PCR-Based Analysis of ColE1 Plasmids in Clinical Isolates and Metagenomic Samples Reveals Their Importance as Gene Capture Platforms

ColE1 plasmids are important vehicles for the spread of antibiotic resistance in the Enterobacteriaceae and Pasteurellaceae families of bacteria. Their monitoring is essential, as they harbor important resistant determinants in humans, animals and the environment. In this work, we have analyzed ColE1 replicons using bioinformatic and experimental approaches. First, we carried out a computational study examining the structure of different ColE1 plasmids deposited in databases. Bioinformatic analysis of these ColE1 replicons revealed a mosaic genetic structure consisting of a host-adapted conserved region responsible for the housekeeping functions of the plasmid, and a variable region encoding a wide variety of genes, including multiple antibiotic resistance determinants. From this exhaustive computational analysis we developed a new PCR-based technique, targeting a specific sequence in the conserved region, for the screening, capture and sequencing of these small plasmids, either specific for Enterobacteriaceae or specific for Pasteurellaceae. To validate this PCR-based system, we tested various collections of isolates from both bacterial families, finding that ColE1 replicons were not only highly prevalent in antibiotic-resistant isolates, but also present in susceptible bacteria. In Pasteurellaceae, ColE1 plasmids carried almost exclusively antibiotic resistance genes. In Enterobacteriaceae, these plasmids encoded a large range of traits, including not only antibiotic resistance determinants, but also a wide variety of genes, showing the huge genetic plasticity of these small replicons. Finally, we also used a metagenomic approach in order to validate this technique, performing this PCR system using total DNA extractions from fecal samples from poultry, turkeys, pigs and humans. Using Illumina sequencing of the PCR products we identified a great diversity of genes encoded by ColE1 replicons, including different antibiotic resistance determinants, supporting the previous results achieved with the collections of bacterial isolates. In addition, we detected cryptic ColE1 plasmids in both families with no known genes in their variable region, which we have named sentinel plasmids. In conclusion, in this work we present a useful genetic tool for the detection and analysis of ColE1 plasmids, and confirm their important role in the dissemination of antibiotic resistance, especially in the Pasteurellaceae family of bacteria.

Mutual influence of secondary and key drug-resistance mutations on catalytic properties and thermal stability of TEM-type β-lactamases

Highly mutable β-lactamases are responsible for the ability of Gram-negative bacteria to resist β-lactam antibiotics. Using site-directed mutagenesis technique, we have produced in vitro a number of recombinant analogs of naturally occurring TEM-type β-lactamases, bearing the secondary substitution Q39K and key mutations related to the extended-spectrum (E104K, R164S) and inhibitor-resistant (M69V) β-lactamases. The mutation Q39K alone was found to be neutral and hardly affected the catalytic properties of β-lactamases. However, in combination with the key mutations, this substitution resulted in decreased KM values towards hydrolysis of a chromogenic substrate, CENTA. The ability of enzymes to restore catalytic activity after exposure to elevated temperature has been examined. All double and triple mutants of β-lactamase TEM-1 bearing the Q39K substitution showed lower thermal stability compared with the enzyme with Q39 intact. A sharp decrease in the stability was observed when Q39K was combined with E104K and M69V. The key R164S substitution demonstrated unusual ability to resist thermal inactivation. Computer analysis of the structure and molecular dynamics of β-lactamase TEM-1 revealed a network of hydrogen bonds from the residues Q39 and K32, related to the N-terminal α-helix, towards the residues R244 and G236, located in the vicinity of the enzyme's catalytic site. Replacement of Q39 by lysine in combination with the key drug resistance mutations may be responsible for loss of protein thermal stability and elevated mobility of its secondary structure elements. This effect on the activity of β-lactamases can be used as a new potential target for inhibiting the enzyme.

Percolation in protein sequence space

The currently known protein sequences are not distributed equally in sequence space, but cluster into families. Analyzing the cluster size distribution gives a glimpse of the large and unknown extant protein sequence space, which has been explored during evolution. For six protein superfamilies with different fold and function, the cluster size distributions followed a power law with slopes between 2.4 and 3.3, which represent upper limits to the cluster distribution of extant sequences. The power law distribution of cluster sizes is in accordance with percolation theory and strongly supports connectedness of extant sequence space. Percolation of extant sequence space has three major consequences: (1) It transforms our view of sequence space as a highly connected network where each sequence has multiple neighbors, and each pair of sequences is connected by many different paths. A high degree of connectedness is a necessary condition of efficient evolution, because it overcomes the possible blockage by sign epistasis and reciprocal sign epistasis. (2) The Fisher exponent is an indicator of connectedness and saturation of sequence space of each protein superfamily. (3) All clusters are expected to be connected by extant sequences that become apparent as a higher portion of extant sequence space becomes known. Being linked to biochemically distinct homologous families, bridging sequences are promising enzyme candidates for applications in biotechnology because they are expected to have substrate ambiguity or catalytic promiscuity.