(C) PLOS One
This story was originally published by PLOS One and is unaltered.
 . . . . . . . . . .



Experimental evolution partially restores functionality of bacterial chemotaxis network with reduced number of components [1]

['Manika Kargeti', 'Max Planck Institute For Terrestrial Microbiology', 'Center For Synthetic Microbiology', 'Synmikro', 'Marburg', 'Irina Kalita', 'Sarah Hoch', 'Maryia Ratnikava', 'Wenhao Xu', 'Bin Ni']

Date: 2025-08 

The chemotaxis signaling pathway, which enables bacteria to follow chemical gradients in their environment, is highly conserved among motile bacteria. It is assumed that Escherichia coli contains the minimal and non-redundant set of protein activities that are necessary for bacterial chemotaxis and nearly universally conserved among bacterial chemotaxis pathways. These include stimulus sensing, signal transduction towards the flagellar motor, and adaptation-based temporal comparisons of the environment. In this study, we show that functionality of the chemotaxis signaling pathway lacking some of its proteins can be partially regained by subjecting E. coli strains to experimental evolution under selection for chemotactic spreading in porous medium. While the core signaling components are indeed essential for the pathway function, the absence of auxiliary pathway proteins required for adaptation and desensitization could be compensated by specific sets of mutations affecting the other pathway components. Further characterization of the evolved strain lacking the adaptation enzyme CheR suggested that this strain utilizes an alternative mechanism of biased drift in chemical gradients, which does not rely on short-term adaptation that is normally considered a prerequisite for bacterial chemotaxis. Although the efficiency of this alternative mechanism remains below the one that can be achieved by the original memory-based chemotaxis strategy of E. coli, it can mediate chemotaxis not only in porous medium but also in liquid. Thus, even short-term experimental evolution of microorganisms can result in the appearance of behavioral strategies that are qualitatively different from those used by parental organisms.

Chemotactic behavior of motile bacteria in environmental gradients is one of the most-studied models for signal transduction and information processing in biology. The chemotaxis pathway of the gut bacterium Escherichia coli has been assumed to possess the minimal set of activities that are necessary to mediate bacterial navigation in gradients. Here we demonstrate that the short-term experimental laboratory evolution could rewire the signaling network to restore the ability to follow chemical gradients in the absence of individual components that were previously considered essential. Subsequent characterization revealed that these bacteria evolved an alternative strategy that is markedly different from the established paradigm of bacterial chemotaxis and yet exhibits a comparable efficiency to that of the non-evolved wildtype cells. This demonstrates the surprising evolvability of bacterial signaling and behavior, with evolution over just a few hundred generations resulting in the appearance of a qualitatively different behavioral strategy.

To test the essentiality of E. coli pathway proteins, here we subjected a set of E. coli deletion strains to experimental evolution under selection for chemotaxis-driven spreading in soft agar over several hundred generations. Experimental evolution, also known as adaptive laboratory evolution, is a powerful approach for investigating how individual proteins and gene regulatory networks adapt under defined selection pressure [ 19 , 20 ]. Recently, this approach has been used to examine the evolvability of genetic regulation under selection for motility [ 21 ] and the underlying cost-benefit tradeoffs between motility and growth [ 22 – 24 ]. We show that the absence of auxiliary chemotaxis proteins can be reproducibly compensated for by short-term adaptive evolution, albeit to various extent, while the core signaling functions remain essential. Importantly, the evolved strains not only regained the ability to spread in soft agar but also demonstrated biased drift in chemical gradients in soft agar and in liquid, indicating their capability to perform true chemotaxis. Analysis of the evolved ∆cheR cells revealed that their restored chemotaxis does not require short-term adaptation at the receptor level, indicating emergence of an alternative behavioral strategy.

Despite their apparent inability to perform chemotaxis, early studies indicated that E. coli strains lacking both CheR and CheB activities may exhibit some degree of tactic behavior [ 13 , 15 ]. This was further supported by the emergence of spontaneous (pseudo)revertants of the cheR deletion strain that could spread on soft-agar plates, with compensatory mutations mapping to either cheB [ 16 ] or tsr [ 17 ] genes. However, the compensatory mechanisms underlying this phenomenon remained unclear [ 17 ], and a subsequent study concluded that the cheR cheB mutants or the cheR revertants may rather spread in soft agar in a chemotaxis-independent fashion due to their intermediate tumbling bias, which enables slow and non-directional movement through the agar pores [ 14 ]. Such pseudotactic mutants, which carry mutations in genes encoding flagellar hook or motor proteins, have also been isolated in chemotaxis-deficient strains of other bacteria [ 18 ].

Since chemotaxis protein activities in E. coli are non-redundant and, except for CheZ, nearly universally conserved in bacterial chemotaxis pathways, it is typically assumed that all of them are necessary for efficient gradient navigation by swimming bacteria [ 5 ]. Indeed, already early experimental studies have shown that E. coli chemotaxis requires all cytoplasmic chemotaxis proteins and at least one major chemoreceptor, Tar or Tsr [ 11 ]. Strains with deletions in cheW, cheA, cheY, or all receptor genes do not phosphorylate CheY, resulting in continuous running without reorientations. Conversely, cheZ-deficient cells have an excess of phosphorylated CheY (CheY-P) and tumble most of the time. Deletions in cheR or cheB genes also result in, respectively, very low or high levels of pathway activity. In addition to their effects on the cell tumbling bias, lack of these adaptation enzymes disables temporal comparisons, making bacteria unable to efficiently navigate chemical gradients in liquid [ 12 ]. All chemotaxis mutants also have a deficiency in spreading on soft agar plates [ 13 ], which is a commonly used assay for motility and chemotaxis that relies on the spreading of motile bacteria through agar pores following self-generated gradients of consumed chemoattractant nutrients [ 14 ].

The adaptation module comprises two enzymes, the methyltransferase CheR and the methylesterase CheB, which respectively methylate or demethylate four (or five) specific glutamates on chemoreceptors. Methylated glutamates promote a high activity state of chemoreceptors. Notably, the receptors are first expressed in the intermediate activity state, with two of the four methylation sites being encoded as glutamines, which function similarly to methylated glutamates and are deamidated by CheB to glutamates. This adaptation module is unique among bacterial two-component signaling pathways but it is nearly universally present in studied chemotaxis systems, with the notable exception of gastric species of Helicobacter [ 10 ]. Enzymatic activity of the methylation enzymes depends on the receptor activity state, and the resulting negative feedback ensures that the steady-state activity of the pathway can adapt to intermediate level even in the presence of persistent stimulation. Additionally, changes in methylation occur with a delay following receptor stimulation, creating a short-term memory that swimming bacteria use for temporal comparisons of environmental conditions. Both functions of the adaptation module are assumed to be essential for the bacterial chemotaxis strategy.

(A) Schematic representation of the chemotaxis signaling pathway of Escherichia coli. The pathway includes transmembrane chemoreceptors (two major chemoreceptors, Tar and Tsr, are shown) that form sensory complexes together with a kinase CheA (A) and an adaptor CheW (W). CheY (Y) phosphorylation by CheA mediates signal transduction to flagellar motor, inducing a switch from the default counterclockwise (CCW) to clockwise (CW) rotation. The adaptation module, including CheR (R; red) and CheB (B; light blue), regulates chemoreceptor activity through methylation and demethylation of chemoreceptors on four specific glutamates (black circles). The phosphatase CheZ (Z; dark blue) is responsible for rapid CheY dephosphorylation. See text for more details. (B) Spreading of the wildtype (wt) E. coli RP437 strain and of its derivatives, ∆cheR (R), ∆cheB (B) and ∆cheZ (Z), where the corresponding individual chemotaxis gene was deleted, in TB soft agar (TBSA) after incubation for ~16 h at 30°C. The upper row shows spreading of the non-evolved wildtype and parental deletion strains (denoted by “0”). The lower row shows one line evolved for spreading in TBSA for 30 days (line 1; denoted by “1”) for each strain. Scale bars are 2 cm. (C-E) Size of the spreading rings for ∆cheR (R), ∆cheB (B) and ∆cheZ (Z) strains (C); for ∆cheA (A), ∆cheY (Y), ∆cheW (W) and receptor-less ∆mcp (M) strains (D); and for ∆(cheR cheB) (RB) strain (E) normalized to that of the wildtype E. coli RP437. Colors in (C) correspond to protein colors in (A). Spreading was measured for three independent replicates after incubation for ~16 h in TBSA. Several independent lines of evolution (indicated by numbers) are shown for each strain. Error bars indicate standard errors of the mean. P values were calculated for comparisons between spreading of evolved and respective non-evolved strains for each deletion, using two-tailed Student t-test (ns, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001).

The sensory module of E. coli chemotaxis pathway ( Fig 1A ) comprises transmembrane chemoreceptors, also known as methyl-accepting chemotaxis proteins, that control the autophosphorylation activity of the receptor-associated kinase CheA with the assistance of the scaffolding protein CheW [ 8 ]. The sensory module’s output is transmitted to the flagellar motor through the CheA-dependent phosphorylation of the response regulator CheY. The phosphorylation of CheY and its binding to flagellar motors increase when the bacterium travels in an unfavorable direction. This induces a switch in the motor rotation from the default counterclockwise (CCW) to clockwise (CW) direction, resulting in the flagellar bundle falling apart and the bacterium tumbling and reorienting. When swimming in a favorable direction, such as traveling up the gradient of attractant, the binding of attractant to receptors inhibits CheA autophosphorylation, which reduces CheY phosphorylation and, in turn, favors CCW rotation and smooth swimming. This core of the sensory and signaling module is conserved in all bacterial chemotaxis systems and is evolutionary related to the broader class of bacterial two-component pathways [ 9 ]. In addition, the chemotaxis signaling module of E. coli and closely related proteobacteria includes the phosphatase CheZ that is responsible for the rapid dephosphorylation of CheY, whereas other chemotaxis systems contain alternative phosphatases.

Most motile bacteria can follow gradients of nutrients and other stimuli in their environment through chemotaxis, which is important for growth optimization, collective behaviors, and interactions with eukaryotic hosts [ 1 , 2 ]. The core of the signaling pathway mediating chemotaxis is highly conserved among prokaryotes [ 3 , 4 ]. Escherichia coli has one of the simplest chemotaxis pathways that is composed almost exclusively of evolutionary conserved proteins and became one of the most comprehensively studied signaling systems in biology [ 5 ]. The mechanism of bacterial chemotaxis has been demonstrated to rely on temporal comparisons of the perceived alterations in environmental conditions by swimming bacteria, where the chemotaxis signaling system determines whether the bacterium should persist in its current direction of movement or reorient itself [ 6 ]. This strategy requires two modules: one for rapid environmental sensing and signal transduction and another for slower adaptation that enables short-termed temporal comparisons of environmental conditions [ 5 , 7 ].

Results

Experimental evolution can compensate for defects caused by the deletions of chemotaxis genes Experimental evolution of E. coli mutant strains was performed under selection for increased spreading on tryptone broth soft-agar (TBSA) plates for 30 cycles of up to 16 hours each (S1A Fig), and with up to four independently evolved lines. All these strains were derived from E. coli strain RP437 that is commonly used as the wildtype for studies of chemotaxis [25]. We relied on the natural mutation rate of E. coli cells, which is known to be sufficiently high to enable evolutionary tuning of motility under strong selection for chemotaxis in TBSA [22]. Consistent with numerous previous studies, none of these strains deleted for individual chemotaxis genes initially exhibited spreading in soft agar (Figs 1B-D, and S1B). However, the spreading of the evolved ∆cheR, ∆cheB, and ∆cheZ strains significantly improved compared to the original deletion strains in all evolved lines (Fig 1B). These three strains lack the auxiliary components of the signaling pathway, either the adaptation enzymes or the phosphatase (Fig 1A). The extent of the improvement varied between gene deletions and lines, with the largest improvement being observed for ∆cheR lines where spreading reached ~50% of the non-evolved wildtype (wt) (Fig 1C). Notably, a previous study has shown that the spreading of the RP437 wildtype could itself be enhanced up to 50% by the experimental evolution under similar conditions [22], meaning that the evolved ∆cheR lines in our experiments could achieve ~30% spreading of the evolved wildtype. In contrast, the absence of universally conserved core components of the chemotaxis pathway that are required for sensing and signal transduction to the motor, including CheA, CheY, CheW, or all chemotaxis receptors (MCPs), could not be compensated for by such short-term evolution (Figs 1D and S1B). Furthermore, no improvement in the spreading of a double ∆(cheR cheB) deletion strain under selection was observed (Fig 1E). This was surprising, since the ∆(cheR cheB) strain prior to evolution was spreading slightly better than the individual cheR or cheB deletion strains, likely because of its intermediate tumbling bias [14]. As the ∆cheR strain showed the largest enhancement of spreading, we evolved four additional ∆cheR lines (S2 Fig).

Evolved strains exhibit compensatory changes in motility We next investigated changes in the motility phenotypes of the evolved E. coli strains, by tracking cell swimming in liquid. Consistent with the known importance of intermediate tumbling frequency for spreading in soft agar [14,26], the evolved strains exhibited compensation for the defects in tumbling that were present in the original deletion strains. In nearly all cases, the fraction of time that cells spent tumbling became more similar to the wildtype (Fig 2A), with all evolved ∆cheR lines showing increased tumbling and all evolved ∆cheZ lines showing decreased tumbling compared to the respective non-evolved deletion strains. Cell tumbling in liquid correlated well with the increased spreading in TBSA (Fig 2B), consistent with a recent report showing that the tumbling bias of wildtype E. coli cells is approximately optimal for spreading in soft agar [26]. However, the evolutionary adjustment of tumbling was clearly not the sole determinant of the improved spreading, as strains with similar tumbling could spread very differently. This is exemplified by ∆cheB lines, where spreading of several lines has improved without changes in tumbling. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 2. Acquired changes in motility phenotypes and flagellar gene expression in the evolved lines. (A) Tumbling bias, defined as the fraction of time spent tumbling, for the wildtype, parental non-evolved strains (R0, B0 and Z0), and evolved strains (R1-R4, B1-B4, Z1-Z4), measured in three independent replicates. (B) Size of the spreading rings in TBSA, normalized to that of the wildtype strain (data from Fig 1C), plotted as a function of tumbling bias for individual strains. The dotted line indicates the tumbling bias for the wildtype. Inset: Spreading in TBSA as a function of the absolute deviation of the tumbling bias from that of the wildtype strain. The relationship between the two variables was evaluated using Spearman’s rank correlation, yielding a coefficient of –0.51. (C) Run speed between two consecutive tumbles, measured for all strains in three independent replicates. Motility phenotypes were assessed using cell tracking (see Materials and Methods). (D) Activity of transcriptional fliC promoter (PfliC) reporter, measured as fluorescence of green fluorescent protein (GFP) using flow cytometry in three independent replicates. Inset: Run speed (data from C) plotted as a function of the fliC promoter activity. Colors correspond to protein colors in Fig 1A. Significance analysis was done in comparison to the respective non-evolved deletion strains. Error bars indicate standard errors of the mean. P values were calculated using two-tailed Student t-test (ns, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001). https://doi.org/10.1371/journal.pgen.1011784.g002 Furthermore, nearly all of the evolved strains exhibited an increased swimming velocity (Fig 2C). Similar increase in velocity was previously observed during the evolution of wildtype cells for the spreading in TBSA, and it was shown to be the consequence of the elevated expression of the flagellin gene fliC and other flagellar genes [22]. Therefore, we measured activity of the transcriptional reporter of fliC promoter (PfliC) [22]. This activity was indeed significantly higher in most of the evolved strains compared to their parental strains (Fig 2D), consistent with our hypothesis that the increased swimming velocity of the evolved strains may be primarily due to the changes in flagellar gene expression. Cell swimming velocity was previously shown to initially increase with the flagellar regulon activity above that in the wildtype RP437 but subsequently saturate at approximately twofold higher expression levels [22,27], which could be observed in our data as well (Fig 2D Inset). Since spreading in soft agar requires not only motility and chemotaxis but also cell growth, we further quantified growth of the evolved strains. We observed that their growth rate generally decreased compared to the parental strains (S3 Fig). This observation could be at least partly explained by the known negative tradeoff between expression of flagellar genes and E. coli growth [22–24,27–29], and previous experimental evolution of the wildtype in TBSA indeed led to the elevated motility at the cost of slower growth [22]. Nevertheless, there was no simple correspondence between fliC promoter activity and growth of individual evolved strains, suggesting that other factors besides flagellar gene expression contribute to the reduction of their growth rate.

Chemotactic spreading is restored by cumulative effects of multiple mutations We further investigated the order in which mutations were detected over the course of evolution for several ∆cheR lines (S4A, S4B, S4C and S4D Fig), which revealed that the mutations in tsr were selected first, followed by mutations in other chemotaxis and/or flagellar genes. One of the best-spreading lines, R1, was subsequently selected to evaluate the phenotypic impacts of individual mutations and their potential epistatic interactions. R1 carries mutations in the chemotaxis genes tsr, cheB and cheZ (Fig 3A, 3C and 3D), and a mutation in the flagellar export gene fliI. When these mutations were introduced individually into the ∆cheR strain, mutations in tsr or cheB significantly increased spreading on TBSA plates (Fig 3F). Indeed, these two genes are most commonly affected in the evolved ∆cheR lines. Spreading was further increased by combinations of multiple chemotaxis and fliI mutations. Interestingly, the observed order of selection for the individual mutations during evolution of the R1 line (S4A Fig) is apparently consistent with the path of the largest stepwise increase in spreading due to addition of each subsequent mutation, tsr < tsr cheZ < tsr cheZ fliI < tsr cheB cheZ fliI. Spreading of the ∆cheR tsr cheB cheZ fliI strain largely recapitulated that of the R1 line, with the residual difference being likely due to additional mutations present in the R1 line. Indeed, inactivation of sspA – which was interrupted in R1 – was previously shown to increase flagellar gene expression [22], and the introduction of this mutation led to further increase in spreading (Fig 3F). A gradual increase in spreading was similarly observed when cheB and tsr mutations from R4 and R5 lines were introduced individually into the ∆cheR strain (S4E Fig). Similar to the R1-specific mutations, the effect of tsr mutations on spreading was stronger than that of cheB mutations. We tested the effects of R1-specific mutations on the chemotactic spreading when introduced individually in the wildtype cells (S4F Fig). Mutation in fliI led to enhanced spreading, consistent with the previous report where similar mutations were shown to increase swimming velocity [22], while all other mutations resulted in no or only modest changes in spreading in the wildtype background. This indicates that all of the mutated genes remain functional, including mutated cheB and cheZ genes that are expected to produce proteins with reduced enzymatic activities to compensate for the lack of CheR. To directly test whether the function of CheB was required for the spreading of the R1 line, we introduced the cheB deletion in the evolved strain. The R1 strain lacking cheB showed even less spreading in soft agar than the ∆cheR strain (Fig 3F), confirming that the (reduced) activity of CheB is necessary for the re-evolved spreading of ∆cheR strains. To verify the impact of mutations in cheB and in tsr genes on the functionality of their products, we characterized the pathway response in the wildtype cells carrying the R1-specific mutations in these genes (S5 Fig). This was done using the previously established assay based on Förster (fluorescence) resonance energy transfer (FRET), which monitors the phosphorylation-dependent interaction between CheY fused to a yellow fluorescent protein (CheY-YFP) and its phosphatase CheZ fused to a cyan fluorescent protein (CheZ-CFP) [41,42]. High pathway activity leads to increased phosphorylation of CheY-YFP, which results in increased complex formation with CheZ-CFP and higher energy transfer from the excited CFP donor to the YFP acceptor fluorophore, reflected by the increased ratio between the YFP and CFP fluorescence emission. Consistent with the previous work [41], inhibition of the pathway activity in the wildtype cells by stimulation with attractant (MeAsp) leads to the adaptive pathway response, characterized by the initial decrease in FRET and subsequent recovery of the pathway activity upon prolonged stimulation (S5A Fig). This recovery is mediated by the CheR-dependent receptor methylation [41]. It is followed by a characteristic overshoot of the pathway activity upon the removal of attractant, also followed by the recovery of activity, this time due to the receptor demethylation mediated by CheB. Although the R1-specific mutation in cheB (R75C) introduced in the wildtype cells did not reduce their spreading in soft agar (S4F Fig), it had a clear impact on the pathway response (S5B Fig). The strain carrying this mutation showed larger amplitude of response to the addition of saturating concentration of attractant and weaker amplitude of response to its removal, suggesting that it has higher steady-state pathway activity. Moreover, this strain showed markedly slower adaptation to the removal of attractant, mediated by CheB. These observations are consistent with our expectation that this mutation reduces the activity of CheB. Similar elevation of the adapted pathway activity was also observed upon the introduction of the R1-specific mutation in tsr (T305M) (S5C, S5D, S5E, S5F and S5G Fig). Moreover, this latter mutation affected the dose dependence of the pathway response to the Tsr ligand serine, shifting it to higher concentrations and making it apparently steeper compared to the wildtype response (S5C, S5D and S5G Fig). In contrast, it had little effect on the dose dependence of the response to MeAsp, the ligand of Tar (S5E, S5F and S5G Fig).

[END]
---
[1] Url: https://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1011784

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/