The EAL domain protein YciR acts as a trigger enzyme in a c-di-GMP signalling cascade in E. coli biofilm control

Abstract

C-di-GMP-which is produced by diguanylate cyclases (DGC) and degraded by specific phosphodiesterases (PDEs)-is a ubiquitous second messenger in bacterial biofilm formation. In Escherichia coli, several DGCs (YegE, YdaM) and PDEs (YhjH, YciR) and the MerR-like transcription factor MlrA regulate the transcription of csgD, which encodes a biofilm regulator essential for producing amyloid curli fibres of the biofilm matrix. Here, we demonstrate that this system operates as a signalling cascade, in which c-di-GMP controlled by the DGC/PDE pair YegE/YhjH (module I) regulates the activity of the YdaM/YciR pair (module II). Via multiple direct interactions, the two module II proteins form a signalling complex with MlrA. YciR acts as a connector between modules I and II and functions as a trigger enzyme: its direct inhibition of the DGC YdaM is relieved when it binds and degrades c-di-GMP generated by module I. As a consequence, YdaM then generates c-di-GMP and-by direct and specific interaction-activates MlrA to stimulate csgD transcription. Trigger enzymes may represent a general principle in local c-di-GMP signalling.

Figure 1

Integration of two c-di-GMP control modules in the regulation of the biofilm regulator CsgD. (A) Components involved in the regulation of the biofilm regulator CsgD and curli fibres in E. coli. DGCs are indicated by ovals, PDEs by hexagons, DGCs and PDEs relevant for this study are labelled by red and blue letters, respectively. This figure is a revised version of a figure previously published in the supplement of Mika et al (2012). (B) Curli gene expression in mutants with single or multiple knockout mutations in the YegE/YhjH and YdaM/YciR c-di-GMP control modules. (C) Curli gene expression in strains with mutations in YegE, YdaM and YciR that also carry low copy number plasmids (derived from the vector pCAB18) expressing the IPTG-inducible DGC YaiC or PDE YhjH. Derivatives of E. coli K-12 W3110 carrying a single copy csgB::lacZ reporter fusion as well as the indicated mutant alleles and plasmids were grown in LB at 28°C for 24 h and optical densities (OD₅₇₈) as well as specific β-galactosidase activities were determined.

Figure 2

In vitro interactions between the diguanylate cyclase YdaM, the phosphodiesterase YciR and the transcription factor MlrA. By affinity chromatography of extracts of total soluble cellular proteins, S-tagged YdaM (left panel) and S-tagged MlrA (right panel) were bound to S-protein agarose and secondary extracts containing His-tagged YciR, YdaM or YhjH as indicated or no His-tagged protein (‘pQE60 empty’) were added. His-tagged proteins co-eluted with the S-tagged bait proteins were detected by immunoblotting with anti-His antibodies. ‘Control’ samples did not include the S-tagged proteins. Unlabelled lanes contain coloured size markers.

Figure 3

Detection of in vivo interactions between YdaM, YciR and MlrA and the identification of interacting domains. (A) Using the BacterioMatch®-II two-hybrid system, the indicated proteins or protein domains were co-expressed as hybrids to lambda cI-NTD (on pBT) and RNAP alpha-NTD (on pTRG). Interaction in the combinations indicated is detected by growth on selective plates (note that in the last panel, conditions were more stringent; see Supplementary data for details). The previously known strong interaction between σ^S (RpoS) and its proteolytic targeting factor RssB (Becker et al, 1999) serves as a positive control. (B) Interactions between different domains are schematically summarised. Strong and weak interactions are indicated by the thickness of the lines, blue lines indicate conditional interactions, which are seen only with the isolated domains, but not in the presence of additional and therefore inhibitory domains (for YdaM-NTD, this inhibition is indicated as a dotted line). YdaM is drawn as a dimer, since GGDEF domains have to dimerise to be active as a DGC. Arrows in (A) point to key results, with arrow colours referring to the colouring of the strong interactions schematically summarised in (B).

Figure 4

Enzymatic activities and binding of c-di-GMP and GTP by YciR. In (A), purified YciR, YciR^AAL and YhjH (all at 0.5 μM) were assayed for PDE activity with [³²P]-c-di-GMP. Where indicated, samples also contained unlabelled GTP or [α-³²P]-GTP alone (last two lanes). (B) UV crosslinking with [³²P]-c-di-GMP and [α-³²P]-GTP (according to Christen et al, 2005) was tested for purified YciR and the indicated YciR mutant versions (0.5 or 1 μM), using YdaM (1.6 μM) and YhjH (2.5 μM) as controls. Mutations in the GGDEF and EAL motifs of YciR are as indicated, ‘G/R’ refers to a mutant YciR in which alterations in the GGDEF motif (AAAAF) and R248A were combined.

Figure 5

YciR inhibits the YdaM/MlrA-generated output and is antagonised by YegE-generated c-di-GMP acting via the EAL domain of YciR. W3110 derivatives carrying the single copy csgB::lacZ reporter fusion as well as either wild-type yciR (A) or the point-mutated chromosomal yciR^AAL allele (B) and yegE or yhjH knockout mutations as indicated were grown in LB at 28°C. OD₅₇₈ as well as specific β-galactosidase activities were determined. The wild-type yciR strain carries the yciR⁺ allele ‘back-crossed’ into the chromosome by the same procedure that was used to integrate yciR^AAL. In (C) and (D), DGC assays were performed with [α-³²P]-GTP, YdaM and substoichiometric concentrations of YciR (C), or YciR and YciR^AAL in further increasing, equal and superstoichiometric concentrations (D).

Figure 6

The function of the DGC activity of YdaM in the YdaM/MlrA-generated output. (A) Binding of c-di-GMP to the ‘I-site’ in YdaM. UV crosslinking of [³²P]-c-di-GMP to wild-type YdaM as well as to the A-site and I-site-mutated YdaM variants was assayed as in Figure 4. (B) DGC assays were performed with purified YdaM (wt), YdaM^GGAAF (A^-), YdaM^I-site (I^-) and the N-terminally truncated YdaM^ΔNTD using [α-³²P]-GTP as a substrate. Proteins were used at a concentration of 1 μM, except for YdaM^ΔNTD (3 μM). (C) Inhibition of DGC activity by excess non-radiolabelled c-di-GMP (up to 100 μM) was tested for PleD* and YdaM proteins (1 μM). (D, E) Single copy csgB::lacZ expression was tested in W3110 derivatives carrying either wild-type ydaM or the chromosomal ydaM^A-site and ydaM^I-site alleles (D), as well as yegE or yhjH knockout mutations as indicated (E). (F) The influence of the ydaM^A-site allele on csgB::lacZ expression was tested in W3110 derivatives carrying either yciR⁺ or yciR knockout alleles.

Figure 7

c-di-GMP binding to YdaM, MlrA and YciR. UV crosslinking with [³²P]-c-di-GMP (A) or [α-³²P]-GTP (B, C) tested for YdaM (1.6 μM), MlrA (2.6 μM) and YciR (0.5 μM) was assayed in the indicated combinations. Where indicated, MlrA was preincubated with a csgD promoter-carrying DNA fragment for 60 min. In (C) degradation fragments of YciR, which correspond to the EAL (YciR*) and GGDEF (YciR**) domain, were also labelled by [α-³²P]-GTP (note that labelling of YciR* requires the presence of YdaM, that is, DGC activity, which—besides it size—indicates that YciR* corresponds to the EAL domain alone). In addition, YdaM dimers were weekly detectable.

Figure 8

Model of the csgD-controlling c-di-GMP signalling cascade and its inherent feedback cycles. Module I (consisting of the DGC YegE and the PDE YhjH) controls c-di-GMP that is sensed and degraded by the PDE YciR, which together with the DGC YdaM constitutes module II. YciR is a also trigger enzyme whose second activity—the direct inhibition of YdaM and MlrA—is relieved when it is active as a PDE. YdaM is equally bifunctional. While its DGC activity contributes to the c-di-GMP pool generated by module I in a positive feedback loop, it also stimulates the transcription factor MlrA by direct interaction. DGCs are indicated by ovals and PDEs by hexagons. DGCs and high c-di-GMP-driven processes are shown in red, and PDEs and processes occurring at low c-di-GMP levels are shown in blue.