Crystal structure of master biofilm regulator CsgD regulatory domain reveals an atypical receiver domain

Abstract

The master regulator CsgD switches planktonic growth to biofilm formation by activating synthesis of curli fimbriae and cellulose in Enterobacteriaceae. CsgD was classified to be the LuxR response regulatory family, while its cognate sensor histidine kinase has not been identified yet. CsgD consists of a C‐terminal DNA binding domain and an N‐terminal regulatory domain that provokes the upstream signal transduction to further modulate its function. We provide the crystal structure of Salmonella Typhimurium CsgD regulatory domain, which reveals an atypical β5α5 response regulatory receiver domain folding with the α2 helix representing as a disorder loop compared to the LuxR/FixJ canonical response regulator, and the structure indicated a noteworthy α5 helix similar to the non‐canonical master regulator VpsT receiver domain α6. CsgD regulatory domain assembles with two dimerization interfaces mainly through α1 and α5, which has shown similarity to the c‐di‐GMP independent and stabilized dimerization interface of VpsT from Vibrio cholerae respectively. The potential phosphorylation site D59 is directly involved in the interaction of interfaces I and mutagenesis studies indicated that both dimerization interfaces could be crucial for CsgD activity. The structure reveals important molecular details for the dimerization assembly of CsgD and will shed new insight into its regulation mechanism.

Short abstract

PDB Code(s): 5XP0

Introduction

To cope with free‐floating and surface‐adhesion growth, the planktonic and biofilm phenotypes constitute the integral bacterial life. Biofilms are bacterial exopolymer‐enclosed communities.1, 2 They are mainly made up from extracellular polysaccharides and protein polymers such as curli fimbriae. This extracellular matrix mediates bacterial adhesion to abiotic or biotic surfaces. It shields bacteria from environmental stress including exposure to antibiotics, starvation and the host defense system. Therefore, biofilm formation is tightly involved in bacterial persistence, tolerance and pathogenesis. As such, biofilms form an obstacle for the eradication of infectious bacteria and lead to many chronic infectious diseases.3 Additionally, the biofilm matrix is a potential source of immune activation. For example, the biofilm formed by Salmonella enterica serovar Typhimurium (S. Typhimurium) containing amyloid curli protein and bacterial DNA was recently reported to trigger autoantibody production contributing to the progression of systemic lupus erythematosus.4

Switching between planktonic and biofilm mode of growth is critical in bacterial development and must be tightly regulated.5 The biofilm master regulator CsgD is such a key transcriptional response regulator controlling the formation of curli fimbriae and cellulose production.6, 7 CsgD modulates the csg (curli specific gene) operon encoding the curli assembly components major curli subunit CsgA, minor curli subunit CsgB;8 chaperone CsgC9 and its secretion components secretion factor CsgE,10 CsgF11 and secretion channel CsgG,12, 13 which are highly conserved among E. coli, S. Typhimurium and many other Enterobacteriaceae.14, 15 Besides, the CsgD regulatory network is cross‐talking with many other transcriptional regulators and at least 18 consensus DNA binding sites were identified in S. Typhimurium.16 For example, CsgD activates the diguanylate cyclase adrA gene, which encodes the enzyme for the synthesis of the secondary messenger cyclic dimeric GMP (c‐di‐GMP), further enhancing cellulose production.17

CsgD expression displays a bistable pattern characterized by a dramatically different expression level that results in two subpopulations of biofilm cells: in aggregated cells, expression is high while planktonic cell microcolonies display low expression.18 This CsgD‐based discrimination is accompanied by differential gene expression provoking S. Typhimurium planktonic cells to be more competent for virulence while aggregated cells mediate persistence.19 Expression of CsgD is tightly controlled by the response regulator protein OmpR and c‐di‐GMP.18, 20 In addition, expression of CsgD is highly sensitive to environmental stimuli such as starvation, iron, temperature, pH, oxygen tension and osmolality.21, 22

CsgD was classified to be the LuxR response regulatory family with conserved Helix‐Turn‐Helix (HTH) DNA‐binding motif, which typically denoted in the two‐components regulatory system and can be activated through phosphorylation by its cognate sensor histidine kinase. The conserved phosphorylation site aspartate (D59) was found as from regulatory receiver domain, the cognate sensor histidine kinase has not been identified yet. CsgD consists of an N‐terminal regulatory domain (receiver domain as in response regulator) and a C‐terminal HTH DNA binding motif. The regulatory domain of CsgD executes its transcriptional function upon upstream signal transduction. The conserved aspartate (D59) in the regulatory domain can be phosphorylated, potentially by a cognate sensor histidine kinase, which decreases its DNA binding properties, hence dramatically reducing the biofilm formation.23 However, another member of the LuxR and CsgD family of prokaryotic regulators, Vibrio cholerae VpsT, is directly modulated by c‐di‐GMP binding which results in dimerization, rather than by phosphorylation through an upstream kinase.24

Given the importance of the regulatory domain of CsgD in signal transduction and activity regulation, we carried out the crystallography and related mutagenesis investigation of CsgD regulatory domain. We provide the crystal structure of Salmonella Typhimurium CsgD regulatory domain at 2.0 Å, the structure reveals important molecular details for the assembly of CsgD and will shed new insight into the regulation mechanism of biofilm formation.

Results

Crystal structure of CsgD regulatory domain reveals an atypical receiver domain folding

CsgD dependent transcriptional regulation is modulated by its N‐terminal regulatory domain. To gain insight in the mode of CsgD regulation, we determined its crystal structure to 2.0 Å resolution. Detailed data collection and refinement statistics are shown in Table 1. There are two CsgD regulatory domain molecules found in the asymmetry unit. The monomer of CsgD regulatory domain constitutes 5 β strands and 5 α helices, with the β strands stabilizing as a parallel β sheet surround by the α helices [Fig. 1(A)]. Interestingly, compared to the canonical 5 times βα repeat motif folding of LuxR/FixJ family response regulator receiver domain, the CsgD regulator domain lacks the α2 helix which represent as disorder loop and involves a α5 helix directly following the α4 helix [Fig. 1(A)]. CsgD belongs to the LuxR response regulatory family and its regulatory domain shows low sequence identity to other response regulator such as well characterized LuxR and PhoP (around 10%) and a moderate sequence identify of 30% to the master regulator VpsT from Vibrio cholerae, which involved in the matrix production and motility. VpsT was also identified to be LuxR/CsgD response regulatory family and demonstrated with a non‐canonical α6β5 receiver domain fold, which has an extra α6 following the (βα)₅ fold.24 The alignment of CsgD and VpsT indicated that the CsgD showed similarity to the VpsT fold with RMS of 0.83 between 128 Cα atoms [Fig. 1(B)]. The α2 helix from VpsT was shown to be well defined as helix, while it presented as a disorder loop in CsgD with relative flexibility. Furthermore, the VpsT extended α6 helix could be aligned with CsgD α5 helix [Fig. 1(B)]. The sequence alignment as judged from the structure has shown that there are 5 residues missing in between the α2 helix from CsgD as compared to the VpsT, which may result in loop formation and its flexibility [Fig. 1(C)].

Table 1.

X‐Ray Data Collection and Refinement Statistics

Crystal	CsgD regulatory domain
Data collection
Spacegroup	P41212
a, b, c (Å)	57.8, 57.8, 208.1
α, β, γ (°)	90, 90, 90
Resolution (Å)	40.1–2.0 (2.07–2.0)
Rmerge	0.084 (0.75)
Rmeas	0.03(0.266)
Multiplicity	8.5 (8.6)
CC(1/2)	0.999 (0.864)
CC*	1 (0.963)
I/σ(I)	17.44 (2.93)
Completeness (%)	98.81 (98.63)
Wilson B‐factor (Å2)	31.77
Refinement
Total Reflections	207427 (20427)
Unique Reflections	24539 (2376)
R work/R free	0.1946/0.2306
Number of atoms:
Protein	2111
Water	138
Ligands	3
Average B‐factor (Å2)	50.34
Protein ADP (Å2)	50.78
Ligands (Å2)	331.13
Water ADP (Å2)	44.03
Ramachandran plot:
Favored/Allowed (%)	97/3
Root‐Mean‐Square‐Deviation:
Bond lengths (Å)	0.012
Bond Angle (°)	1.13
PDB code	5XP0

Statistics for the highest resolution shell are shown in parentheses.

Figure 1.

CsgD regulatory domain monomer structure. A. Crystal structure of CsgD regulatory domain constitutes 5 β strands and 5 α helices, Compared to the canonical (βα)₅ response regulator receiver domain folding, the CsgD regulator domain lacks the α2 helix which now represent as disorder loop. The structure involves a α5 helix directly following the α4 helix. B. Structure alignment of CsgD (Cyan) and VpsT (Red) regulatory domain. The α2 helix from VpsT was disorder in CsgD and CsgD α5 can align with the extended VpsT α6 could. C. Sequence alignment of CsgD and VpsT as judged from the secondary structural information. The secondary structure was schematically illustrated as cylinder (Helix) and arrow (Strand). The α helix of CsgD (Blue) and VpsT (Red) were labeled under the sequence. VpsT α2 helix was shown as red cylinder. The potential conserved phosphorylation D was highlighted with red star.

CsgD regulatory domain assembles with two dimerization interfaces

What especially noteworthy of the CsgD regulatory domain crystal packing, dimerization interfaces I and II were found near the N‐ and C‐terminus, which assembles into a oligomeric structure [Fig. 2(A)]. The N‐terminal dimerization interface I was mainly formed by the loop between β1 and α1 and the loop between β3 and α2, whereas the C‐terminal dimerization interface II mainly consisted of α5 and the loop between α4 and α5 [Fig. 2(A)]. The dimerization interface I and II buried surface areas of ∼760 Ų with ΔG −9.4 kcal/mol and ∼514 Ų with ΔG −8.9 kcal/mol, respectively, as determined by PDB‐PISA.25

Figure 2.

CsgD regulatory domain assembles with two dimerization interfaces. A. CsgD receiver domain crystal structure indicated two dimerization interface I and II packing, the dimerization interface I was mainly formed by the N‐terminal α1–β2 (blue) with ΔG = −9.4 kcal/mol and the dimerization interface II was mainly stabilized with the C‐terminal α5 (orange) with ΔG = −8.9 kcal/mol. The buried surface area in dimerization interface I and II are ∼760 and ∼514 Ų, respectively, as determined with PDBPISA. B. Residues involved in the dimerization interface I interactions. The residues Lys18, Leu21, Gln22, Asp59, Glu62, and Asn88 from two monomers (blue and green) are labeled in orange and purple, respectively, hydrogen bonds are shown with dash lines. The dimerization interface I with the A panel cyan monomer is shown in surface colored related to charge. C. Residues involved in the dimerization interface II. A hydrophobic groove is formed in one molecule (surface colored related to charge) between C‐terminal α5 and the other N‐terminal part, which can anchor‐hold the α5 from the second molecule (Green cartoon), Ile140, Ser137, and Gln133 are the involved key residues, 90° rotation view is shown. The surface was shown in the default atom color setting: carbon as cyan, oxygen as red, and nitrogen as blue.

The dimerization interface I is stabilized by residues Lys18, Leu21, and Gln22 from the loop between β1 and α1; Asp59, Glu62 from the loop between β3 and α2 and N88 from the loop between β4 and α3 [Fig. 2(B)]. There are eight hydrogen bonds and two salt bridges formed between them (Table SI). The C terminal dimerization interface II was mainly involving a patch of residues including Ile140, Ser137 and Gln133 from the C‐terminal helix α5. A hydrophobic groove was formed in one molecule between the α5 helix and the other αβα sandwich N‐terminal structure, which can anchor‐hold the α5 from the second molecule [Fig. 2(C)]. The CsgD sequences are highly conserved in E. coli and S. Typhimurium with a sequence identity of 88% in the regulatory domain and all the residues involved in the two distinct dimerization interfaces are identical. The CsgD regulatory domain dimerization interface I and II shares a high level of structural similarity with the VpsT c‐di‐GMP independent and stabilized dimerization interface, as illustrated by the RMSD value of 2 Å between 256 Cα atoms (Fig. S1).

Both dimerization interfaces I and II are crucial for CsgD function

To explore the physiological relevance of the two dimerization interfaces, we constructed multiple point mutants covering residues involved in the two dimerization interfaces and performed the Congo red assay monitoring curli and cellulose production.26 Congo red mainly stains the amyloid curli fibers and endows the curli‐producing colony with an “rdar” (red, dry, and rough) morphotype, which can serve as a surrogate for the amount of curli produced.27

CsgD is the transcriptional factor positively regulating expression of the csgBAC operon encoding the Curli subunit CsgA and CsgB, and the di‐guanylate cyclase AdrA leading to cellulose production. As expected, deletion of csgD dramatically reduces the number of rdar morphotype colonies and resulted in white and smooth morphotypes on Congo red agar plate [Fig. 3(A)]. The ΔcsgD complemented with pBADCsgD recovers curli production whereas the ΔcsgD transfected with an empty pBAD vector resulted in a similar morphotype as ΔcsgD [Fig. 3(A)]. We further constructed a series of point mutation strains as K18A, L21A/Q22A, D59A, E62A, N88A, and I140A, altering residues involved in the dimerization interface I and II respectively. The L21A/Q22A, D59A, and I140A mutations were shown to significantly display decreased curli production, and the putative phosphorylation site D59A mutation yielded a strong decrease in curli production similar as the ΔcsgD mutant. The K18A and N88A mutations had moderate impacts on curli production while E62A mutation presented only a mild effect on the curli production [Fig. 3(A)]. Since the csgD expression controls both the curli and cellulose production, the rdar morphotype is also altered between ΔcsgD and WT group.

Figure 3.

Both CsgD dimerization interfaces are crucial for curli production examined by Congo red assay. A. Congo red agar assay experiment of WT, CsgD knockout, pBAD empty vector, pBADCsgD, K18A, L21AQ22A, D59A, E62A, Q88A, and I140A strains in Salmonella Typhimurium IR715 background. The red color degree serves as a surrogate for the amount of curli production. B. Congo red assay of planktonic cell using the same strains as panel A. The Congo red absorption of all strains were quantified and normalized by cell density (OD490/OD600). The tubes picture on the right indicated the pBAD empty vector and pBADCsgD strains. Data are cell density value means ± SD of 5–7 replicates.

We carried out the Congo red assay to further quantify curli production in liquid culture. The solution concentration of Congo red is inversely correlated with the amount of curli produced by the bacterial cells, which would sequester the Congo red from the solution. Therefore, curli production can be quantified by normalizing the OD490 (Congo red adsorption wavelength) against the OD600 values. Consistent with the plate assays, the L21A/Q22A, D59A, and I140 mutations significantly reduced the curli production, while the K18A, N88A mutants showed only a mediocre decrease while the E62A mutation exhibited the weakest effect [Fig. 3(B)].

Since both dimerization interfaces of the CsgD receiver domain affected the production of curli, we reasoned that the genes controlled by the CsgD would also be modulated by the mutations of the key residues involved in the dimerization interface. We tested the expression of two genes whose transcripts were reported to be correlated with CsgD, that is, the major curlin subunit csgA and cellulose related diguanylate cyclase adrA.17 In the CsgD knock out and the empty vector transformed strains, the csgA expression was almost silenced, and in the mutants involved in the two different dimerization interfaces, that is, L21A/Q22A, D59A, and I140A, csgA expression was significantly reduced whereas csgA expression in K18A, E62A, and N88A mutant strains was almost unaffected [Fig. 4(A)]. CsgD knockout reduced adrA expression by 50%–60%, which is consistent with a previous report.16 The L21A/Q22A mutation resulted in reduction of adrA expression close to the CsgD knock out level, while the rest of the mutants including K140A, D59A, N88A, and K18A resulted in modest reduction of adrA expression with a ratio of 20%–30% [Fig. 4(B)]. Intriguingly, the adrA expression is only affected at the early stages of growth since extended culture time invariably led to the recovery of adrA expression in different mutants, indicating the presence of a feedback and/or compensatory pathway for adrA expression regulation. In view of the finding that D59A and I140A mutations in two different interfaces both significantly influenced the function of CsgD, we further validated the curli production in wildtype, D59A, I140A, or empty vector transformed strains at a cellular level using scanning electron microscopy (SEM). The SEM results indicated that the wildtype strain indeed produces extracellular matrix components curli and cellulose, which traps the individual cells to form a biofilm [Fig. 4(C)]. By contrast, the D59A and I140A strains displayed a phenotype similar to the knockout strain and barely presented any curli fimbriae [Fig. 4(C)].

Figure 4.

Both CsgD dimerization interfaces affected csgA and adrA gene expression validated with RT‐PCR and SEM. A. The curlin major subunit csgA gene expression level tested by RT‐PCR in WT, CsgD knockout, pBAD empty vector, pBADCsgD, K18A, L21AQ22A, D59A, E62A, Q88A, and I140A strains in 1 and 3 h, respectively. Data are means ± SD of 5–7 replicates. B. The diguanylate cyclase adrA gene expression level examined by RT‐PCR in all the strains same as panel A in 1 h. Data are means ± SD of 5–7 replicates. C. Scanning Electron Microscopy experiments of WT, pBAD empty vector, D59A, and I140A. The curli fibers can be seemed around the individual cells in WT indicated by red arrows, however the D59A and I140A mutation represent similar phenotype as pBAD empty vector with rarely curli fiber.

Discussion

We here provide structural insights of the master biofilm regulator CsgD regulatory domain that revealed an atypical response regulatory receiver domain folding. Compared to the canonical (βα)5 response regulatory receiver domain folding, the CsgD regulator domain lacks the α2 helix which now represents as disorder loop; the structure also indicated a α5 helix directly following the α4 helix which shared a structural similarity with the α6 from the non‐canonical VpsT receiver domain. The CsgD regulatory domain assembles in the crystal employing two dimerization interfaces (I and II) mainly through α1 and α5 respectively, which has shown similarity to the c‐di‐GMP stabilized and independent dimerization of master regulator VpsT from Vibrio cholerae. Hence, the assembly of CsgD and VpsT regulators belong to a distinct class of LuxR family of response regulators. The potential phosphorylation site D59 is directly involved in the interaction of interfaces I and the dimerization interface II was stabilized by c‐di‐GMP in VpsT. In VpsT, this dimerization interface is stabilized by interaction between two c‐di‐GMP molecules and the (W[F/L/M]PR) motif, which is different from the corresponding region containing a conserved YF[T/S]Q motif in CsgD, and there is no direct evidence that c‐di‐GMP is the ligand of CsgD yet. We performed the interaction study between CsgD and c‐di‐GMP and no binding event was observed as indicated by isothermal titration calorimetery (unpublished data). Both of the dimerization interfaces are necessary for CsgD function and mutants of residues in interface I (L21A/Q22A, D59A) and in interface II (I140A) displayed a relatively reduction in CsgD activity, i.e. promotion of curli production. The fact that we observed similar phenotype for L21A/Q22A and D59A may result from the fact that Q22 forms hydrogen bonds with the phosphorylation site D59. Mutation of residues involved in stabilization of both dimerization interfaces affects the gene expression of curlin major subunit CsgA and diguanylate cyclase adrA whose promoter activity are modulated by CsgD. D59 is highly conserved in transcriptional regulators and is suggested to be the phosphorylation site. Previously, phosphorylation of D59 was shown to decrease the protein stability and DNA binding property.23 Here we have shown that D59 involved directly in the interaction of dimerization interface I and its phosphorylation may affect the downstream gene regulation through interfere with the dimer assembly. The bacterial secondary messenger c‐di‐GMP is involved in the control of CsgD activity,28, 29 but it seems not to be directly sensed like in VpsT in Vibrio cholerae. Additionally, the EAL‐like protein STM1344 controls csgD expression, EAL protein are commonly known as c‐di‐GMP hydrolyzing enzymes, but although STM1344 does not even bind c‐di‐GMP is still retained an indirect role in c‐di‐GMP turnover.30 The role of c‐di‐GMP involved in CsgD regulation will need further investigation. CsgD expression represents a bistable pattern in the biofilm subpopulation representing planktonic and biofilm cells. This hints that the binding of the CsgD to its promoter has to be tightly tuned. The molecules details of the CsgD regulatory domain structure will pave the way for the understanding of the CsgD regulation mechanism.

Materials and Methods

Bacterial strains, plasmids, and culture conditions

All strains and plasmids used in this study are summarized in Table SII. The primers used for cloning and site directed mutagenesis are documented in Table SIII. Unless specifically indicated, all strains were grown in Lysogeny Broth (LB: 10 g/L tryptone, 5 g/L yeast extract, and 10g/L NaCl).

Construction of mutant strains

For construction of an arabinose‐inducible CsgD expression plasmid, a DNA fragment containing the csgD coding sequence was prepared by PCR using genomic DNA of a virulent, nalidixic acid‐resistant derivative of strain ATCC 14028, S. Typhimurium IR715, as a template along with a pair of primers csgD‐F, and csgD‐R. After digestion with NcoI and EcoRI, the PCR‐amplified fragment was inserted at the corresponding site of pBAD‐HisB to generate the pBADCsgD plasmid. The construction of chromosomal CsgD deletion was performed using the one step chromosomal gene inactivation procedure.31 In brief, a pair of primers (CsgD‐KO‐F and CsgD‐KO‐R) was used to amplify a Kanamycin cassette from pKD4 with flanking DNA to the gene targeted for deletion. Approximately, 300 ng PCR‐product was electroporated into S. Typhimurium IR715 containing pKD46, using lambda red recombination method to replace ORF of csgD with Kanamycin cassette. Point mutations were introduced in the S. Typhimurium IR715 knock out background with the pBAD system. The point mutation was generated through the PCR based site‐directed mutagenesis approach described before.32 The csgD knockout strains and pointed mutation strains were verified with PCR and further validated by sequencing.

CsgD regulatory domain protein expression and purification

The S. Typhimurium CsgD regulatory domain (1–144) DNA fragments were inserted into pET28a with N terminus 6 × His tag, and then transformed into E.coli BL21 (DE3) star host cells. The cells were grown in LB medium supplemented with kanamycin (100 μg/mL) at 20°. Protein expression was induced with 1mM IPTG followed by growth for 6h and harvesting by centrifugation at 7000g. The cell pellet was resuspended in 50 mM Tris–HCl, pH 8.0, 300 mM NaCl and 10 mM MgCl₂ in the presence of protease inhibitors and further lysed by sonication. The cell debris was centrifuged at 25 000 rpm for 30 min and the clarified lysate was loaded onto a Ni‐NTA (Qiagen) column after filtration with a 0.22 μm syringe filter cap. The purified protein was concentrated and subjected to size exclusion chromatography (SEC) on a Superdex 75 pg column (GE Healthcare) pre‐equilibrated with 25 mM Tris, pH 8.0, 150 mM NaCl, 10 mM MgCl₂, 5% glycerol or PBS buffer. The fractions containing the pure dimeric CsgD regulatory domain protein (>98% purity as judged by mass spectrometry, column calibration curve and SDS PAGE) were pooled and concentrated for further usage. The online multiple sequence alignment software ClusterW was used for the sequence alignment.

Crystallization and data collection

Crystallization trials with purified and monodisperse preparation of CsgD regulatory domain were set up using the sitting drop vapor diffusion method by mixing 1 µL of 6–8 mg/mL protein sample with 1µl of reservoir solution at 20°. The CsgD regulatory domain crystallization yielded diffraction quality crystals in two conditions: (1) 1.6M magnesium sulfate heptahydrate, 0.1M MES, pH 6.5; (2) 1.5M ammonium sulfate, 0.1M BIS‐TRIS propane pH 7.0. All the crystals were harvested from the crystallization condition drop and flash frozen with a cryostream. Data were collected from single crystal diffraction at a wavelength of 0.98 Å at 100 K in the Shanghai Synchrotron Radiation Facility (SSRF) BL18U1 and BL19U1. Symmetry in both type of crystals corresponded to spacegroup P4₁2₁2 with α = β= γ = 90°, a = 57.8 Å, b = 57.8 Å, c = 207.7 Å. A complete 2 Å diffraction dataset was generated and further used for structure determination.

Structure determination

X‐ray diffraction data were processed with XDS;33 the structure of CsgD regulatory domain was determined by molecular replacement Phaser program using the Vibrio cholerae VpsT receiver domain (PDB: 3KLN) as search model.24 Structure refinement was performed via alternating rounds of manual model building in Coot and Phenix.Refine.34, 35 The structure alignment comparison (Root Mean Square Deviation RMSD calculation) was made using PyMOL, and structure interface and assembly analysis was done with the PDBe PISA sever.25 The final CsgD regulatory domain structure contains two molecules of CsgD regulatory domain per asymmetric unit cell including 6–142 residues and 2 Mg²⁺ ions, 99 water molecules.

Congo red assay

CsgD knock out and point mutant strains were grown in an overnight culture and spotted on salt‐free Congo Red (CR) plates (10 g/L tryptone, 5 g/L yeast extract, 10 mM L‐Arabinose, 40 mg/mL CR, 20 mg/mL Coomassie Blue). Plates were incubated for 48 days at 26°, and the appearance of red colonies indicated curli fiber production and its binding to CR. Quantification of CR‐binding was performed by measuring the amount of CR binding in planktonic cells following a previously described procedure with modification.36 The overnight CsgD knock out and mutant strains preculture were grown in 50 mL of salt‐free LB (10 g/L tryptone, 10 mM L‐Arabinose, 40 mg/mL CR) at 26° for 1–3 h, 1 mL of culture was collected and OD600 were measured and then centrifuged at 13 000g for 2 min, and the supernatant was spectrophotometrically measured at 490 nm (Abs490 unbound). Prior to incubation, salt‐free LB with CR (10 g/L tryptone, 10 mM L‐Arabinose, 40 mg/mL CR) was measured at 490 nm (Abs490 initial). The Abs490 bound was therefore calculated from (Abs490 initial to Abs490 unbound), and the amount of CR bound to cells was obtained from a standard curve constructed using 0–40 mg of CR dissolved in T‐broth. Hence the relative CR values were generated by (Abs490 initial to Abs490 unbound)/OD600 and further normalized compared to the pBADCsgD value.

RNA isolation and qRT‐PCR

Five hundred microliters of CsgD knock out and mutant strains from overnight preculture were grown in 50 mL of salt‐free LB (10 g/L tryptone, 5 g/L yeast extract) at 26°. 10 mM L‐Arabinose was added when OD600 reached 1.0 and further growth for 1–h. Total RNA was isolated from planktonic cells using a Qiagen RNeasy Mini kit, RNA integrity was checked by the Abs260/Abs280 ratio of 2.0–2.1. Expression of csgA and adrA was determined by a two‐step RT‐PCR using the Power SYBR Green PCR Master Mix (Bio‐Rad) and Real‐Time PCR System (Bio‐Rad). First‐strand cDNA synthesis from total RNA was performed with the High‐Capacity cDNA Reverse Transcription Kit (Bio‐Rad) using 1 µg of total RNA as the template following the manufacturer's instructions. Primers were annealed at 60°, and the rrsG was used as reference gene to normalize all data. All qRT‐PCR primers used for the qRT‐PCR (Table SIII) was verified using normal PCR. Fold changes in various csgA and adrA transcripts in CsgD knock out and mutant strains in relative to pBADCsgD were calculated using the 2^−ΔΔCt formula.

Scanning electron microscopy (SEM)

The strains were grown on the glass slides in the 10 g/L tryptone, 5 g/L yeast extract, and 10 mM L‐Arabinose media for 48 h. The supernatant was discarded and glutaraldehyde was used to fix the cells on the slide overnight. The slides were washed and further dehydrated with graded ethanol series. The dehydrated slides were dried and coated with gold. The gold‐coated slides were examined with an FEI Quanta FEG 250 SEM under the low vacuum mode at 5.0 kV.

Conflict of Interests

The authors declare there is no conflict of interest.

Author Contribution

YW and FZ designed and conducted the experiments, YW and ZO performed the crystallography and physiological studies; YW and WH performed the SEM experiments. BD, YS and WL provided strains and new reagents. YW and FZ wrote the paper with contribution from all the authors.

Accession Number

PDB access codes: 5XP0

Supporting information

Supplementary materials includes graph of the structure alignment of both dimerization interface between CsgD and VpsT (Fig. S1), details of the dimerization interactions (Table SI) and constructs and primers used in this study (Table SII, SIII).

Supplementary Figure S1 Structural alignment of CsgD dimerization interface with VpsT. A. Alignment of CsgD dimerization interfaces I with VpsT c‐di‐GMP independent dimerization interface. B Alignment of CsgD dimerization interface II with VpsT c‐di‐GMP stabilized dimerization interface, c‐di‐GMP was shown in sticks.

Supplementary Table S1. Detailed interaction involved in dimerization interface I stability

Supplementary Table S2. Bacterial strains used in this study

Supplementary Table S3. Primers used in this study