Temporal evolution of master regulator Crp identifies pyrimidines as catabolite modulator factors

The dominating Crp* mutation displays a transient phenotype

While a number of the strains isolated in the cya papillation experiment exhibited a clear maltose fermentation phenotype, curiously, the dominating CrpA144T mutation exhibited only a transient phenotype on the selective medium. Mutant papillae typically turn red on MacConkey agar when maltose is efficiently fermented (Fig. 1b), and the A144T mutant grew better than the parental strain (Fig. 1d). However, the red colour phenotype was gradually lost when A144T mutants were restreaked on fresh medium (Fig. 1e).

Additional mutations develop sequentially in crp

In the 594 sequenced crp loci, some mutations such as Q170K and S62F occur at above-average frequency but are only found in combination with other mutations – mainly A144T, but also A144E, and T140R9 (Supplementary Table 1). In line with this observation, papillae occurred at high frequency when starting with a crpA144T strain background (Fig. 1f) and deep sequencing of papillae developing from this background confirmed the appearance of mutations Q170K, S62F, and many others15. Altogether, these observations suggest that the additional crp mutations are not merely passenger mutations, hitchhiking along with Crp* mutations. Other canonical Crp* mutations such as T140K and G141D develop as single Crp mutations, but at a much lower frequency than A144T9. What makes A144T dominant over other Crp* mutations under these selective conditions, and why do second site mutations develop sequentially in the A144T background?

The dominating cmk mutation improves maltose fermentation in combination with a crpA144T mutant

We previously showed that when attempting to introduce the dominating cmk mutation (A216E) into the parental cya strain by recombineering, we were unable to isolate strains without A144T mutations spontaneously forming in crp, and that the cya crpA144T cmkA216E triple mutant fermented maltose more efficiently than the cya crpA144T double mutant9. To further study the possible link between crp and pyrimidine metabolism, we deleted crp from a cya cmkA216E strain, reintroduced crp and crpA144T on low-copy plasmids, and monitored growth on maltose MacConkey agar. This confirmed that CrpA144T appeared more active in a cmkA216E background compared to a cmk + background, both when crpA144T is expressed from the genomic locus and from a plasmid (Fig. 1g, h). What is the connection between Crp and hotspot mutations related to pyrimidine metabolism?

Cytidine and CMP promote maltose-dependent growth

We hypothesized that a metabolite in the spatiotemporal microenvironment, perhaps related to pyrimidines, was building up in the ageing colonies, and was affecting the evolutionary solution space of Crp mutants isolated. The rationale was that upon restoring growth, either by re-streaking on fresh medium or by forming a Crp* mutation in a subpopulation of the colony, the metabolite could be gradually lost together with the maltose fermentation phenotype. At the same time, additional mutations conferring independence of this metabolite would become advantageous.

Several of the cmk mutations introduced stop codons in the cmk gene9 and cmk knockout mutants have been shown to accumulate 30-fold more CMP than a wildtype strain18. We therefore speculated that CMP, or a closely related metabolite, was playing a role in the promotion of cAMP-independent growth. To test this hypothesis, we plated a cya crpA144T double mutant, a cya single mutant, and a cya + strain on maltose MacConkey agar in the presence of cytosine, cytidine, CMP, CDP, or cAMP and monitored growth on maltose MacConkey agar. We observed strong acidification of the medium by the cya crpA144T mutant only when grown in the presence of cAMP, but also weak acidification when grown with cytidine (Fig. 1h). To substantiate this observation, and because additional mutations occur at higher frequency in this background making it hard to control genetically (Fig. 1f), we complemented a cya crp strain with the low-copy plasmid version of wildtype crp or crpA144T and observed a clear fermentation phenotype of the A144T mutant when grown in the presence of cytidine or CMP (Fig. 1h). The different phenotypes of the crpA144T mutants, observed when expressed from the genome or the low-copy vector, is likely due to higher expression of crp from the plasmid, which was confirmed on the protein level by western blotting and on the RNA level by RNA sequencing (Supplementary Fig. 2). This is probably also why crpA144T expressed from a plasmid becomes toxic in presence of cAMP (Fig. 1h).

Nucleotides accumulate in ageing bacteria

To explore the physiological relevance and potential change in nucleotide levels under the experimental conditions explored here and in the different mutants isolated, we plated cya, cya crp, cya crp cmkA216E, and cya crp cmk mutants on maltose MacConkey agar, extracted metabolites from the cells from one-day and five-day-old cultures, and analysed the extracts by LC-MS (Fig. 2a). Whereas the large majority of metabolites identified varied very little in concentration between sampling days and genetic backgrounds (Supplementary Data 1), CMP increased two-to-three fold from day one to day five in the cya and cya crp strains and was significantly higher in the mutant cmk strains (Fig. 2b). This confirms previous observations that CMP levels increase in cmk knockouts and indicates that the dominating cmkA216E mutation is causing reduced Cmk activity.

Fig. 2: Cytidine and CMP accumulate in ageing bacteria.

a Schematic overview of biomass sampling from MacConkey agar plates or Simple MacConkey (SM) liquid cultures for metabolomics with liquid chromatography-mass spectrometry (LC-MS). b Absolute cytidine monophosphate (CMP) levels from sampling day one (white) and five (grey). Significance was based on two-sided paired t-tests between the groups indicated, and designated as significant if p < 0.05. Data represent four replicates. c Fold change of absolute cytidine levels in cya, cya crp, and WT (differentiated by color) for days one to six. Data represent four replicates. d Distribution of fold changes of all metabolites detected by LC-MS from day one to six for cya. The metabolites with the highest fold change are highlighted. From light to darker grey: gsn: guanosine, dimp: deoxyinosine monophosphate, adn: adenosine, glx: glyoxylate, gmp: guanosine monophosphate, cytd: cytidine (red). Data represents the average of four replicates. Source data are provided as Supplementary Data 1.

With this approach, we observed highly variable levels of the cytidine nucleoside and speculated that this could be due to secretion. To analyse the sum of nucleosides inside the cells and secreted into the medium, we repeated the experiment in liquid culture in a medium highly similar in composition to MacConkey agar, and analysed cell extracts together with the medium. This time we decided to analyse the cytidine content in wildtype, cya, and cya crp cells after one, three, and six days of incubation. Cytidine levels varied less across replicates with this approach and showed a drastic 26-33 fold increase in the cya and cya crp mutants (Fig. 2c) from day one to day six. Cytidine also increased in the wildtype strain, but only five-fold at day six compared to day one. Interestingly, this metabolomics analysis of ageing bacteria showed that, compared to other metabolites, nucleosides/nucleotides in general had the highest concentration increase over the time course of the experiment (Fig. 2d) and cAMP was the metabolite that increased the most in the wildtype strain (Supplementary Fig. 3), highlighting the well-known significance of this signaling molecule in bacterial physiology. Combined, these experiments revealed a marked increase in cytosine nucleoside/nucleotide levels five-to-six days after incubation in MacConkey-like media, in particular in cya and cmk mutants, supporting that these metabolites could play a role in the selection of crp mutants. It is interesting to note that five days is exactly the time when papillae started appearing in the original experiment9.

Formation of mutant papillae is accelerated with increased levels of cytosine nucleosides / nucleotides

Our inability to generate cmk mutants without crpA144T spontaneously forming after mutation of the genome indicates a strong selective pressure on crp when cytidine nucleotide levels are increased. To be able to better follow the evolution of crp under these conditions, we reintroduced crp on the low-copy plasmid into the cya crp, cya crp cmk, and cya crp cmkA216E strains and monitored growth and evolution on maltose MacConkey agar. After seven days in the incubator, no papillae had formed with empty vector control, whereas a few papillae formed in the presence of the crp plasmid (Fig. 3a). In contrast, pink secondary colonies formed from a large number of colonies in the mutant cmk strains with the crp plasmid. Similarly, the cya crp strain harboring the pSEVA-crp plasmid formed more papillae in the presence of exogenously added cytidine (Fig. 3b) and in both experiments the majority of these papillae were subsequently found to have developed the crpA144T mutation (Supplementary Table 2).

Fig. 3: Influence of cytidine and cmk mutations on the phenotypes of E. coli on MacConkey agar supplied with different carbon sources.

a Papillation assay of cya crp cmk + , cya crp cmk, and cya crp cmkA216E strains on maltose with crp supplied on a low-copy plasmid or the empty plasmid as control. Pictures were taken after 7 days of incubation. b Papillation assay of the cya crp strain on maltose with or without crp supplied on a low-copy plasmid and with or without 10 mM cytidine. Pictures were taken after 3 days of incubation. c Phenotypes of the E. coli wildtype strain when supplemented with different carbon sources with or without 10 mM cytidine. Fermentation phenotypes are indicated by *** (red color and media acidification), ** (red color but no media acidification), * (limited red color), or – (white color). d Representative pictures of strains, where cytidine has an inhibiting effect on fermentation. Referenced phenotypes on all carbon sources can be found in Supplementary Fig. 4. Ø; no supplementation.

Cytidine affects catabolism of other carbon sources

Our findings strongly suggest a link between cytidine, Crp, and maltose utilization in the cya mutant, but does cytidine affect carbon catabolism more broadly and in a wildtype background? To explore this, we plated the wildtype E. coli K12 MG1655 on MacConkey medium supplemented with 11 different carbon sources in the absence or presence of cytidine. Acidification by fermentation of glycerol, sorbitol, galactose, and melibiose was severely inhibited by cytidine, small negative cytidine-effects were observed on maltose, ribose, rhamnose, and fucose, whereas no significant effects were observed with glucose, lactose, and xylose (Fig. 3c, d, Supplementary Fig. 4). This suggests that cytidine plays a more general role in carbon catabolism.

Cytidine and uridine activate CrpA144T in vitro

A drawback of these in vivo screens is that it is not possible to rule out that different nucleosides and their phosphorylated counterparts are transported differently across the inner and outer membrane of E. coli, or that phosphorylation or dephosphorylation of the added compounds take place. For example, the positive effect of CMP may be due to extracellular dephosphorylation and subsequent uptake of cytidine. To directly test interactions between Crp and different metabolites, we performed in vitro label-free biolayer interferometry (BLI). Wildtype Crp and A144T were expressed and purified by affinity chromatography, mixed with different ligands, and interactions with biotinylated synthetic DNA encoding PmalT, a Crp responsive promoter from E. coli24, were analysed using the Octet RED96 system (Fig. 4a). In line with our in vivo screen, this BLI analysis showed that cytidine and cAMP activated the DNA binding activity of the A144T mutant (Fig. 4a, b), as did uridine, but not the 2’-deoxy nucleoside thymidine, CMP, or UMP (Fig. 4b). In contrast, only cAMP activated wildtype Crp (Fig. 4c). This suggests that CrpA144T interacts directly with pyrimidine nucleosides.

Fig. 4: In vitro activity of Crp and CrpA144T using biolayer interferometry.

a Upper panel: Illustration of biolayer interferometry where a biosensor coated with streptavidin binds to biotinylated DNA encoding the Crp-responsive PmalT promoter, and produces a signal when associating and dissociating with the Crp protein depending on the ligand present. Lower panel: Representative output showing the progression of output (nm) aligned to baseline. Data correspond to part of the data in panel 2b, in the absence of ligand (Ø), with the positive control cyclic adenosine monophosphate (cAMP) or with cytidine. b Association of CrpA144T to PmalT in the presence of pyrimidines (10 mM) or the positive control cAMP (0.5 mM). CMP, cytidine monophosphate; UMP, uridine monophosphate. Absence of ligand (Ø) serves as a negative control. c Association of wildtype Crp to PmalT in the presence of different pyrimidines, the positive control cAMP (white) or no ligand (grey, Ø). Data represent the average of two replicates.

Cytidine and uridine activate CrpA144T, but inhibit wildtype Crp in vivo

Cytidine and uridine have not previously been reported to interact with Crp. To explore this in more detail in vivo, we developed a Crp activity reporter with a high dynamic range based on a plasmid carrying a fusion between PmalT and the fluorescent reporter GFP (Fig. 5a). As intended, the construct responded highly sensitively to different concentrations of cAMP added exogenously in a cya background (Fig. 5a). In agreement with both our in vitro data, and our qualitative in vivo assays on maltose MacConkey, we observed an increase in activity of the CrpA144T mutant in the presence of cytidine (Fig. 5b) and uridine (Supplementary Fig. 5). Also in line with the mild inhibitory effect on maltose MacConkey, in the wildtype K12 MG1655 strain, the reporter was clearly inhibited by exogenously adding these pyrimidines, and we observed the same effects of cytidine supplementation when replacing PmalT with Plac in the reporter plasmid (Fig. 5b, Supplementary Fig. 5). Although a fixed concentration of cytidine was applied in these experiments, we also found that the cytidine-response varied with the concentration (Supplementary Fig. 6) and we verified using transcriptomics that the observed effects of cytidine supplementation are not due to varying Crp levels (Supplementary Fig. 7). This suggests that cytidine and uridine act as signaling molecules by interfering with Crp. The observed inhibitory effect of cytidine appears stronger in vivo than in vitro, which may reflect missing components in the simplified in vitro system or that pyrimidines affect more than the binding of Crp to DNA, for example Crp-RNA polymerase interactions.

Fig. 5: In vivo pyrimidine effects on Crp and CrpA144T determined with a sensitive Crp reporter.

a Illustration and validation of an in vivo Crp activity reporter based on a plasmid with the PmalT promoter controlling expression of gfp. The reporter was transformed into a cya strain, and fluorescence was detected in the presence of different concentrations of cyclic adenosine monophosphate (cAMP). A cya+ strain served as a positive control (white). Data represent the average of two biological replicates. b The effect of exogenously added cytidine (white) on the Crp activity reporter in a wildtype (cya+crp+) or mutant (cya crpA144T) strain background. No supplementation (grey, Ø) serves as a negative control. Data represent the average of three biological replicates. c Effect of overexpression?- P of umpH (grey), compared to empty vector control (white), in wildtype E. coli K12 MG1655 assayed with the GFP reporter. Data represent the average of three biological replicates. Significance was based on two-sided unpaired t-tests between the groups indicated, and designated as significant if p < 0.05. RFU, relative fluorescence units; OD, optical density.

Uridine acts similarly to cytidine by inhibiting Crp and activating CrpA144T. However, in contrast to CMP, which is not, UMP can be synthesized de novo (Supplementary Fig. 1) and biosynthesis is feedback regulated, which may explain why cmk mutations that likely lead to a build-up of both CMP and cytidine are more frequently observed under our experimental conditions. Accumulation of uracil and uridine, catalysed by the enzyme UmpH (Fig. 1c, Supplementary Fig. 1), was previously observed in mutant E. coli with defective feedback regulation of pyrimidine metabolism21. In nutrient-rich conditions uridine may be the more relevant Crp ligand as it is present at mM concentration levels25. To explore if Crp activity is affected by manipulating endogenous pyrimidine levels, we overexpressed?-P umpH from a plasmid in the presence of the PmalT-GFP reporter. Indeed, umpH expression significantly reduced expression from the Crp sensitive reporter (Fig. 5c). In a previous study, global gene regulation was explored when an overflow of uridine and uracil occurred in a pyrimidine feedback dysregulated strain21. In line with our observations, transcriptomics showed that expression of genes known to be positively regulated by Crp (yfcT, malE, malK, malM, bglB, and csgF) were repressed in this strain, whereas genes known to be downregulated by Crp (gadA, gadE, gadX, and gadW) were upregulated21. Together, these observations suggest that endogenous pyrimidine levels affect Crp signaling.

The mutation Q170K confers independence from cytidine

Additional mutations such as Q170K develop sequentially in the crp hotspot and we speculated that this was due to the transient nature of e.g. the A144T mutant phenotype – i.e. the selective advantage of the A144T mutation might decrease if pyrimidine levels drop upon resuming growth from the ageing bacterial colony. To investigate this hypothesis, we assayed the activity of both the A144T mutant and the A144T Q170K double mutant expressed in the presence or absence of 10 mM cytidine using both the GFP reporter and maltose MacConkey agar. This showed that Q170K has an activating effect on CrpA144T, much like cytidine, and that the double mutant is no longer affected by cytidine (Fig. 6a, b). It is thus plausible that A144T is strongly selected for when pyrimidines accumulate, and that when growth is restored, pyrimidine levels drop, thereby making additional mutations such as Q170K advantageous.

Fig. 6: In vivo activity and fitness effects of different Crp mutants.

a Effect of the additional Crp mutation Q170K in the presence (white) or absence (grey) of exogenously added cytidine assayed with the GFP reporter. No supplementation (Ø) serves as a negative control. Data represent the average of three biological replicates. b Phenotypes of Crp A144T Q170K compared to A144T on maltose MacConkey agar. c Relative activities of Crp* mutations T140K and G141D compared to A144T, expressed from a low-copy SEVA plasmid, in the presence (white) or absence (grey) of cytidine assayed with the GFP reporter. Data represent the average of three biological replicates. d Phenotypes of the Crp* mutations T140K, G141D, and A144T on maltose MacConkey agar. e Relative abundance of strains carrying the Crp* mutations T140K and G141D in competition with A144T when grown on agar plates with maltose supplied with cytidine (white) or water (grey, Ø). RFP fluorescence measurements of cells taken from different sections of the agar plate were used to determine the cytidine diffusion gradient. The relative abundance was calculated based on the fluorescence levels for one strain expressing rfp. Data represent the average of two biological replicates. f Illustration and representative picture of competition assay on agar plates. The two competing strains, expressing either gfp or rfp, were mixed in equal proportions and a 10 μl drop of 0.5 mM cytidine or water was applied to the center of the agar plate for diffusion. Significance was based on two-sided unpaired t-tests between the groups indicated, and designated as significant if p < 0.05. RFU, relative fluorescence units; OD, optical density.

Canonical Crp* mutations that occur at low frequency are not activated by cytidine

To fully understand the temporal evolution of Crp, it is important not only to consider the mutations that dominate, but also those that are underrepresented: The A144T (GCA- > ACA) and A144E (GCA- > GAA) mutations occur at a higher frequency than other mutations such as T140K (ACG- > AAG) and G141D (GGC- > GAC)9, but they have all been characterized previously as cAMP-independent Crp* mutants15. Further, the mutated residues are neighbours in the same structural domain, and even pairwise represent the same types of mutations (C- > A or G- > A). The latter is important because the mutations C- > A, most likely caused by oxidation of guanosine on the complementary strand, and G- > A, most likely caused by cytosine deamination on the complementary strand, are highly dominant under these conditions9,26 and mutation bias caused by the available mutational space could limit the observed solution space. This prompted us to assay the activity of these variants in response to different pyrimidine levels. We found that the T140K and G141D mutants were less active than A144T and were not activated by cytidine (Fig. 6c, d). Furthermore, a fitness assay showed a competitive advantage of A144T relative to T140K and G141D in the presence of cytidine when grown on agar plates with maltose (Fig. 6e, f, Supplementary Fig. 8).

The cytidine regulator CytR plays a role in carbon catabolism

Apart from the apparent direct effect of cytidine (and uridine) on Crp, could other mechanisms explain the link between carbon and pyrimidine metabolism? Cytidine is known to bind to the cytidine regulator CytR and interact with Crp in a small regulon mostly involved in nucleotide metabolism27. To explore a potential involvement of CytR, we generated a cya crp cytR deletion strain and monitored growth on MacConkey agar supplied with either maltose or galactose in the presence of pSEVA-crp, pSEVA-crpA144T, or empty vector control. Compared to the cytR + strain, maltose fermentation was less stimulated by cytidine in the cytR strain when expressing crpA144T (Fig. 7). However, cytidine still negatively affected galactose fermentation in the cytR strain, both with crpA144T and when wildtype crp was activated by exogenously added cAMP. This suggests both CytR-dependent and CytR-independent effects of cytidine in carbon catabolism. Interestingly, the cytR mutant grew poorly on maltose MacConkey both in the absence and presence of wildtype crp, even when cAMP was supplied exogenously. In contrast, severe growth defects were only observed on galactose MacConkey when both crp and cytR was deleted, or with the toxic combination of CrpA144T and cAMP. This suggests a complex interplay between Crp, CytR, and nucleotide levels in carbon catabolism.

Fig. 7: Phenotypes of cytR strains on MacConkey agar.

Fermentation phenotypes on MacConkey agar of MG1655 cya crp cytR (+/−) carrying the low-copy pSEVA plasmid expressing either wildtype crp (WT), the A144T mutant, or empty vector control (ctrl). The plates were supplied with maltose (mal) or galactose (gal) as a carbon source and supplemented with cytidine, cAMP or both, or without supplement (Ø). Experiments were repeated three times.

Expression of the heat shock sigma factor RpoH is perturbed in the cya mutant

Beyond the role in regulating genes involved in nucleotide metabolism, a clue to a more global regulatory role of CytR comes from its regulation of the heat shock sigma factor encoding rpoH28. To explore the potential role of rpoH, we constructed a PrpoH promoter gfp fusion and monitored the fluorescence from this reporter in the cya crp and the cya crp cytR strain in combination with the pSEVA-crp or pSEVA-crpA144T plasmids, and in the presence and absence of exogenously added cytidine and cAMP. Fluorescence from this reporter was completely absent in the cya strain with the wildtype crp plasmid, but was high both in the absence of cytR and with the combination of the CrpA144T mutant with cytidine in the cytR + background (Fig. 8a). This shows that PrpoH promoter activity is very low in the cya genetic background, confirms the previously observed regulation of rpoH by CytR and cytidine28, and shows a large impact of the CrpA144T mutant on PrpoH.

Fig. 8: The interplay of Crp and co-regulators on the studied promoters.

a Reporter activity from the rpoH promoter in MG1655 cya crp cytR (+/-) pSEVA-crp (Crp WT or A144T) in the presence (light grey) or absence (dark grey, Ø) of cytidine and cAMP (annotated by +/−). Significance was based on two-sided unpaired t-tests between the groups indicated and designated as significant if p < 0.05. b Annotations of the malT promoter with a naturally occurring mutation in red. Top: RpoD and RpoH consensus sequences as reported in litterature33, 50, 51. Middle: RpoD annotation as reported in RegulonDB52. Bottom: Annotations for hypothetical RpoH recognition sites of the same promoter (two alternatives, alt 1 vs alt 2). c Reporter activity in MG1655 (WT or cya) from the native malT promoter (WT) and from the malT promoter with the naturally occurring promoter mutation (mut*, G->T), when supplemented with (dark to light grey) nothing (Ø), cAMP, cytidine, or glucose. Data represent the average of three biological replicates. RFU relative fluorescence units, OD optical density.

To further study the role of rpoH, we attempted to delete the gene in the different genetic backgrounds studied here. RpoH has previously been reported to be essential for growth at 37 °C, but not at temperatures below 30 °C29. Curiously, we were unable to isolate rpoH mutants in the wildtype K12 MG1655 background, but were successful in different cya mutants. This again gives clues to a tight interplay between Crp and RpoH in bacterial physiology, but further phenotypic characterizations of the cya rpoH mutant strains were inconclusive due to the poor fitness in general of the rpoH strains.

A naturally occurring cis-acting mutation in the malT promoter makes it insensitive to cytidine

Due to the complex trans effects when studying perturbations to global gene regulators such as Crp and RpoH, we finally returned to the original papillae evolution experiment to look for further clues to the regulation of genes involved in maltose utilization. In the 96 sequenced papillae, one mutation in the promoter region of malT was interesting: a G- > T mutation in the −10 box (GACCTT) increases the similarity to the consensus RpoD sequence (TATAAT) (Fig. 8b). When we introduced this mutation into our PmalT-gfp reporter construct, fluorescence was heavily increased compared to the wildtype promoter construct both in the wildtype and the cya background, and the reporter was still responsive to cAMP, but no longer to cytidine (Fig. 8c). This cis mutation again points to an important interplay between sigma factors, Crp, and cytidine in maltose utilization.

Read more here: Source link