An atlas of metabolites driving chemotaxis in prokaryotes

Abstract

Chemicals inducing chemotaxis have been characterised for over 60 years across hundreds of publications. Without any synthesis of these scattered results, our current understanding of the molecules affecting prokaryotic behaviours is fragmented. Here, we examined 341 publications to assemble a comprehensive database of prokaryotic chemoeffectors, compiling the effect (attractant, repellent or neutral) of 926 chemicals previously tested and the chemotactic behaviour of 394 strains. Our analysis reveals that (i) not all chemical classes trigger chemotaxis equally, in particular, amino acids and benzenoids are much stronger attractants than carbohydrates; (ii) over one-quarter of attractants tested are not used for growth but solely act as chemotactic signals; (iii) the prokaryote's origin matters, as terrestrial strains respond to 50% more chemicals than those originating from human or marine biomes; (iv) repellents affect cell behaviour at concentrations 10-fold higher than attractants; (v) the effect of large molecules and the behaviour of bacteria other than Proteobacteria have been largely overlooked. Taken together, our findings provide a unifying view of the chemical characteristics that affect prokaryotic behaviours globally.

Fig. 1. Overview of the chemicals used in attraction assays.

A Chemical classification of the tested compounds. Colours denote organic chemical super-classes while the inner and outer rings represent classes and sub-classes, respectively. B Distribution profile of the chemical used in attraction assays based on their molecular weight. Each bar represents a 10 g mol⁻¹ range, except the last bar, which includes all chemicals with a molecular weight exceeding 1000 g mol⁻¹. The red dots indicate the proportion (%) of compounds identified as attractants in a given range and the red trend line was generated using a local regression method. Chemicals were identified as attractants if they acted as such on at least half of the prokaryotic strains tested. C Proportion of strains attracted toward a given chemical according to its dipole moment (top) or isotropic polarizability volume (bottom). Linear regressions are displayed for amino acids and organic oxygen compounds and the Pearson correlation coefficient ρ (one-sided) is indicated together with the p-value.

Fig. 2. The effect of chemical classification on chemoattraction.

A Number of chemicals per super-class (left) and main sub-class (right). Only sub-classes containing at least 10 tested compounds are shown. B Compounds never (top) or always (>97%; bottom) found as attractants in the collected assays. Only compounds whose effect was assessed with at least five different strains belonging to at least three distinct genera among at least three studies are reported. A red colour gradient was applied to reflect the number of genera tested. Compounds are coloured based on their classification at the super-class level. Red: organic oxygen compounds; Orange: organic acids; Yellow: benzenoids. The abbreviation “der.” stands for “derivatives”. The number of prokaryotic strains tested with each compound is indicated per biome. MW: Molecular weight; H: Human/Animal; T: Terrestrial; F: Freshwater; M: Marine; P: Polluted; N/A: Not available.

Fig. 3. Origin and identity of the prokaryotes used in attraction assays.

A Number of tested chemicals per prokaryotic genus. Genera are coloured by biome. The number of publications that performed chemotaxis assays with each genus is indicated at the end of the respective bar. Pie charts representing strain distribution depending on biome (left) and taxonomy (class level; right) are shown. Asterisks denote archaeal genera/classes. Unknown taxonomy and classes containing less than three tested strains (Acidithiobacillia, Actinomycetia, Cyanophyceae, Flavobacteriia, Methanococci*, Methanomicrobia* and Oligoflexia) are referred to as “Other classes”. B Venn diagram showing the number of chemicals tested with strains from only one biome or with strains from multiple biomes. The total number of chemicals tested per biome is indicated in brackets. C Super-class profile of the chemicals tested with strains from each biome. Each time a chemical was tested with a different strain, it was counted as a separate occurrence. Super-classes containing less than 10 occurrences in all biomes are included in “Other and unknown classification”.

Fig. 4. Response to attractant across different prokaryotic biomes.

A Proportion (%) of chemicals attracting a given strain. Only strains tested with at least 10 chemicals are included (n = 23, 44, 3, 14 and 5 for Human / Animal, Terrestrial, Freshwater, Marine and Polluted environments, respectively). The box plots represent the first quartile, median, third quartile, and minimum and maximum values (i.e., whiskers). The white crosses denote the average values. ANOVA followed by post hoc Tukey HSD test (one-sided) were conducted, the asterisk denotes significant differences (p = 0.01). B Chemicals were scattered at the super-class level. Dot colour represents the proportion (%) of chemicals within a super-class attracting each prokaryotic strain. Strains are organised depending on their isolation biome. Only the strains that were tested with at least six chemicals of one super-class are represented. Dot size is proportional to the number of chemicals within each super-class tested with each strain. C Dot colour represents the proportion (%) of prokaryotic strains attracted towards different chemicals. Only chemicals that were tested at least three times in at least two out of the biomes “Human/Animal”, “Terrestrial” and “Marine” are represented. Dot size is proportional to the number of tested strains.

Fig. 5. General features of the repulsion assays.

A Number of unique chemicals tested on strains from each biome. B Venn diagram of chemicals used in chemotaxis assays. Bars indicate the organic chemical classification at the sub-class level. C Number of chemicals per sub-class. Black: tested chemicals; Red: chemicals identified as attractant with at least one strain; Blue: chemicals identified as attractant with at least 50% of the strains tested. D Threshold chemical concentrations (M) assessed for attractants and repellents. The box plots represent the first quartile, median, third quartile, and minimum and maximum values (i.e., whiskers). The asterisk denotes a significant difference (t-test, one sided, p = 0.007).