Logo: to the web site of Uppsala University

Background

Heteroresistance (HR) is a significant type of antibiotic resistance observed for several bacterial species and antibiotic classes where a susceptible main population contains small subpopulations of resistant cells. Mathematical models, animal experiments and clinical studies associate HR with treatment failure. Currently used susceptibility tests do not detect heteroresistance reliably, which can result in misclassification of heteroresistant isolates as susceptible which might lead to treatment failure. Here we examined if whole genome sequence (WGS) data and machine learning (ML) can be used to detect bacterial HR.

Methods

We classified 467 Escherichia coli clinical isolates as HR or non-HR to the often used β-lactam/inhibitor combination piperacillin-tazobactam using pre-screening and Population Analysis Profiling tests. We sequenced the isolates, assembled the whole genomes and created a set of predictors based on current knowledge of HR mechanisms. Then we trained several machine learning models on 80% of this data set aiming to detect HR isolates. We compared performance of the best ML models on the remaining 20% of the data set with a baseline model based solely on the presence of β-lactamase genes. Furthermore, we sequenced the resistant sub-populations in order to analyse the genetic mechanisms underlying HR.

Findings

The best ML model achieved 100% sensitivity and 84.6% specificity, outperforming the baseline model. The strongest predictors of HR were the total number of β-lactamase genes, β-lactamase gene variants and presence of IS elements flanking them. Genetic analysis of HR strains confirmed that HR is caused by an increased copy number of resistance genes via gene amplification or plasmid copy number increase. This aligns with the ML model's findings, reinforcing the hypothesis that this mechanism underlies HR in Gram-negative bacteria.

Interpretation

We demonstrate that a combination of WGS and ML can identify HR in bacteria with perfect sensitivity and high specificity. This improved detection would allow for better-informed treatment decisions and potentially reduce the occurrence of treatment failures associated with HR.

Interactions between eukaryotic hosts and their bacterial symbionts drive key ecological and evolutionary processes, from regulating ecosystems to the evolution of complex molecular machines and processes. Over time, endosymbionts generally evolve reduced genomes, and their relationship with their host tends to stabilize. However, host–bacteria relationships may be heavily influenced by environmental changes. Here, we review these effects on one of the most ancient and diverse endosymbiotic groups, formed by—among others—Legionellales, Francisellaceae, and Piscirickettsiaceae. This group is referred to as Deep-branching Intracellular Gammaproteobacteria (DIG), whose last common ancestor presumably emerged about 2 Ga ago. We show that DIGs are globally distributed, but generally at very low abundance, and are mainly identified in aquatic biomes. Most DIGs harbour a type IVB secretion system, critical for host-adaptation, but its structure and composition vary. Finally, we review the different types of microbial interactions that can occur in diverse environments, with direct or indirect effects on DIG populations. The increased use of omics technologies on environmental samples will allow a better understanding of host–bacterial interactions and help unravel the definition of DIGs as a group from an ecological, molecular, and evolutionary perspective.

The recently isolated bacterium "Candidatus Uabimicrobium amorphum" is the only known prokaryote that can engulf other bacterial cells. Its proteome contains a high fraction of proteins involved in signal transduction systems, which is a feature normally associated with multicellularity in eukaryotes. Here, we present a protein-based phylogeny which shows that "Ca. Uabimicrobium amorphum" represents an early diverging lineage that clusters with the Saltatorellus clade within the phylum Planctomycetota. A gene flux analysis indicated a gain of 126 protein families for signal transduction functions in "Ca. Uabimicrobium amorphum", of which 66 families contained eukaryotic-like Serine/Threonine kinases with Pkinase domains. In total, we predicted 525 functional Serine/Threonine kinases in "Ca. Uabimicrobium amorphum", which represent 8% of the proteome and is the highest fraction of Serine/Threonine kinases in a bacterial proteome. The majority of Serine/Threonine kinases in this species are membrane proteins and 30% contain long, tandem arrays of WD40 or TPR domains. The pKinase domain was predicted to be located in the cytoplasm, while the WD40 and TPR domains were predicted to be located in the periplasm. Such domain combinations were also identified in the Serine/Threonine kinases of other species in the Planctomycetota, although in much lower abundances. A phylogenetic analysis of the Serine/Threonine kinases in the Planctomycetota inferred from the Pkinase domain alone provided support for lineage-specific expansions of the Serine/Threonine kinases in "Ca. Uabimicrobium amorphum". The results imply that expansions of eukaryotic-like signal transduction systems are not restricted to multicellular organisms, but have occurred in parallel in prokaryotes with predatory lifestyles and phagocytotic-like behaviors.

Legionella pneumophila are host-adapted bacteria that infect and reproduce primarily in amoeboid protists. Using similar infection mechanisms, they infect human macrophages, and cause Legionnaires’ disease, an atypical pneumonia, and the milder Pontiac fever. We hypothesized that, despite the similarities in infection mechanisms, the hosts are different enough that there exist high-selective value mutations that would dramatically increase the fitness of Legionella inside the human host. By comparing a large number of isolates from independent infections, we identified two genes, mutated in three unrelated patients, despite the short duration of the incubation period (2–14 days). One is a gene coding for an outer membrane protein (OMP) belonging to the OmpP1/FadL family. The other is a gene coding for an EAL-domain-containing protein involved in cyclic-di-GMP regulation, which in turn modulates flagellar activity. The clinical strain, carrying the mutated EAL-domain-containing homologue, grows faster in macrophages than the wild-type strain, and thus appears to be better adapted to the human host. As human-to-human transmission is very rare, fixation of these mutations into the population and spread into the environment is unlikely. Therefore, parallel evolution – here mutations in the same genes observed in independent human infections – could point to adaptations to the accidental human host. These results suggest that despite the ability of L. pneumophila to infect, replicate in and exit from macrophages, its human-specific adaptations are unlikely to be fixed in the population.

The profusion of sequenced genomes across the bacterial and archeal domains offers unprecedented possibilities for phylogenetic and comparative genomic analyses. In general, phylogenetic reconstruction is improved by the use of more data. However, including all available data is (i) not computationally tractable, and (ii) prone to biases, as the abundance of genomes is very unequally distributed over the biological diversity. Thus, in most cases, subsampling taxa to build a phylogeny is necessary. Currently, though, there is no available software to perform that handily. Here we present TADA, a taxonomic-aware dataset selection workflow that allows sampling across user-defined portions of the prokaryotic diversity with variable granularity, while setting constraints on genome quality and balance between branches.

Midichloria spp. are intracellular bacterial symbionts of ticks. Representatives of this genus colonise mitochondria in the cells of their hosts. To shed light on this unique interaction we evaluated the presence of an intramitochondrial localization for three Midichloria in the respective tick host species and generated eight high-quality draft genomes and one closed genome, showing that this trait is non-monophyletic, either due to losses or multiple acquisitions. Comparative genomics supports the first hypothesis, as the genomes of non-mitochondrial symbionts are reduced subsets of those capable of colonising the organelles. We detect genomic signatures of mitochondrial tropism, including the differential presence of type IV secretion system and flagellum, which could allow the secretion of unique effectors and/or direct interaction with mitochondria. Other genes, including adhesion molecules, proteins involved in actin polymerisation, cell wall and outer membrane proteins, are only present in mitochondrial symbionts. The bacteria could use these to manipulate host structures, including mitochondrial membranes, to fuse with the organelles or manipulate the mitochondrial network.

Protein translation is a foundational attribute of all living cells. The translation function carried out by the ribosome critically depends on an assortment of protein interaction partners, collectively referred to as the translation machinery. Various studies suggest that the diversification of the translation machinery occurred prior to the last universal common ancestor, yet it is unclear whether the predecessors of the extant translation machinery factors were functionally distinct from their modern counterparts. Here we reconstructed the shared ancestral trajectory and subsequent evolution of essential translation factor GTPases, elongation factor EF-Tu (aEF-1A/eEF-1A), and initiation factor IF2 (aIF5B/eIF5B). Based upon their similar functions and structural homologies, it has been proposed that EF-Tu and IF2 emerged from an ancient common ancestor. We generated the phylogenetic tree of IF2 and EF-Tu proteins and reconstructed ancestral sequences corresponding to the deepest nodes in their shared evolutionary history, including the last common IF2 and EF-Tu ancestor. By identifying the residue and domain substitutions, as well as structural changes along the phylogenetic history, we developed an evolutionary scenario for the origins, divergence and functional refinement of EF-Tu and IF2 proteins. Our analyses suggest that the common ancestor of IF2 and EF-Tu was an IF2-like GTPase protein. Given the central importance of the translation machinery to all cellular life, its earliest evolutionary constraints and trajectories are key to characterizing the universal constraints and capabilities of cellular evolution.

Bacteria adapting to living in a host cell caused the most salient events in the evolution of eukaryotes, namely the seminal fusion with an archaeon, and the emergence of both mitochondrion and chloroplast. A bacterial clade that may hold the key to understanding these events is the deep-branching gammaproteobacterial order Legionellales—containing among others Coxiella and Legionella—of which all known members grow inside eukaryotic cells. Here, by analyzing 35 novel Legionellales genomes mainly acquired through metagenomics, we show that this group is much more diverse than previously thought, and that key host-adaptation events took place very early in its evolution. Crucial virulence factors like the Type IVB secretion (Dot/Icm) system and two shared effector proteins were gained in the last Legionellales common ancestor (LLCA). Many metabolic gene families were lost in LLCA and its immediate descendants, including functions directly and indirectly related to molybdenum metabolism. On the other hand, genome sizes increased in the ancestors of the Legionella genus. We estimate that LLCA lived approximately 1.89 Ga, probably predating the last eukaryotic common ancestor by approximately 0.4–1.0 Gy. These elements strongly indicate that host adaptation arose only once in Legionellales, and that these bacteria were using advanced molecular machinery to exploit and manipulate host cells early in eukaryogenesis.

Homologous recombination is an important mechanism directly involved in the repair, organization, and evolution of prokaryotic and eukaryotic chromosomes. We developed a system, based on two genetic cassettes, that allows the measurement of recombinational repair rates between different locations on the chromosome. Using this system, we analyzed 81 different positional combinations throughout the chromosome to answer the question of how the position and orientation of sequences affect intrachromosomal homologous recombination. Our results show that recombination was possible between any two locations tested in this study and that recombinational repair rates varied by just above an order of magnitude. The observed differences in rate do not correlate with distance between the recombination cassettes or with distance from the origin of replication but could be explained if each location contributes individually to the recombination event. The relative levels of accessibility for recombination vary 5-fold between the various cassette locations, and we found that the nucleoid structure of the chromosome may be the major factor influencing the recombinational accessibility of each chromosomal site. Furthermore, we found that the orientation of the recombination cassettes had a significant impact on recombination. Recombinational repair rates for the cassettes inserted as direct repeats are, on average, 2.2-fold higher than those for the same sets inserted as inverted repeats. These results suggest that the bacterial chromosome is not homogenous with regard to homologous recombination, with regions that are more or less accessible, and that the orientation of genes affects recombination rates. IMPORTANCE Bacterial chromosomes frequently carry multiple copies of genes at separate chromosomal locations. In Salmonella, these include the 7 rrn operons and the duplicate tuf genes. Genes within these families coevolve by homologous recombination, but it is not obvious whether their rates of recombination reflect general rates of intrachromosomal recombination or are an evolved property particularly associated with these conserved genes and locations. Using a novel experimental system, we show that recombination is possible between all tested pairs of locations at rates that vary by just above 1 order of magnitude. Differences in rate do not correlate with distance between the sites or distance to the origin of replication but may be explained if each location contributes individually to the recombination event. Our results suggest the existence of bacterial chromosomal domains that are differentially available for recombination and that gene orientation affects recombination rates.