Functional Role of N- and C-Terminal Amino Acids in the Structural Subunits of Colonization Factor CS6 Expressed by Enterotoxigenic Escherichia coli

ABSTRACT

CS6 is a common colonization factor expressed by enterotoxigenic Escherichia coli. It is a two-subunit protein consisting of CssA and CssB in an equal stoichiometry, assembled via the chaperone-usher pathway into an afimbrial, oligomeric assembly on the bacterial cell surface. A recent structural study has predicted the involvement of the N- and C-terminal regions of the CS6 subunits in its assembly. Here, we identified the functionally important residues in the N- and C-terminal regions of the CssA and CssB subunits during CS6 assembly by alanine scanning mutagenesis. Bacteria expressing mutant proteins were tested for binding with Caco-2 cells, and the results were analyzed with respect to the surface expression of mutant CS6. In this assay, many mutant proteins were not expressed on the surface while some showed reduced expression. It appeared that some, but not all, of the residues in both the N and C termini of CssA and CssB played an important role in the intermolecular interactions between these two structural subunits, as well as chaperone protein CssC. Our results demonstrated that T20, K25, F27, S36, Y143, and V147 were important for the stability of CssA, probably through interaction of CssC. We also found that I22, V29, and I33 of CssA and G154, Y156, L160, V162, F164, and Y165 of CssB were responsible for CssA-CssB intermolecular interactions. In addition, some of the hydrophobic residues in the C terminus of CssA and the N terminus of CssB were involved in the stabilization of higher-order complex formation. Overall, the results presented here might help in understanding the pathway used to assemble CS6 and predict its structure.

IMPORTANCE Unlike most other colonization factors, CS6 is nonfimbrial, and in a sense, its subunit composition and assembly are also unique. Here we report that both the N- and C-terminal amino acid residues of CssA and CssB play a critical role in the intermolecular interactions between them and assembly proteins. We found mainly that alternate hydrophobic residues present in these motifs are essential for the interaction between the structural subunits, as well as the chaperone and usher assembly proteins. Our results indicate the involvement of the side chains of identified amino acids in CS6 assembly. This study adds a step toward understanding the interactions between structural subunits of CS6 and assembly proteins during CS6 biogenesis.

INTRODUCTION

Enterotoxigenic Escherichia coli (ETEC) infection is the most common cause of infantile diarrhea, which accounts for approximately 210 million episodes and 380,000 deaths annually (1). Plasmid-encoded colonization factors (CFs) and heat-labile and heat-stable (LT and ST, respectively) enterotoxins are the major determinants of ETEC virulence. Colonization is the prerequisite step in initiating the virulence mediated by the interaction of CFs with their cognate receptors on the intestinal epithelium. So far, 25 different CFs have been identified and subdivided on the basis of antigenicity, molecular weight, and structural morphology (2, 3).

Of the 25 known CFs, CS6 is the predominant one; it has been present on approximately 30% of the ETEC isolates tested worldwide (4). The CS6 operon consists of four genes, cssA, cssB, cssC, and cssD. CssA and CssB are the structural subunits, whereas CssC and CssD are chaperone and usher proteins, respectively (5). CS6 is a unique CF in terms of its assembly and subunit composition. Unlike most other CFs, CS6 is afimbrial and composed of two heterologous structural subunits, CssA and CssB, that form a tight protomer in equal stoichiometry (2, 5, 6). The structural subunits are synthesized from a precursor polypeptide and transferred into the periplasm, where chaperone proteins protect them from misfolding and proteolytic degradation (7, 8). However, the stability of CssA was shown to be more dependent on interaction with CssC than on interaction with CssB (8). According to the proposed model of CS6 assembly by Wajima et al. (9), CssA forms a complex with CssB and CssC. The CssA-CssB-CssC complex is then recognized by CssD, and CssA-CssB is transported to the cell surface through the outer membrane.

CssA and CssB do not have sequence similarity to any other subunit of chaperone-usher (CU)-assembled fimbriae. However, the N termini of both subunits showed the presence of a donor strand (G_d) motif characterized by alternating hydrophobic and hydrophilic residues (10). The C termini of CssA and CssB showed the presence of a chaperone protein recognition motif, as defined for P pilus structural subunits. This motif is characterized by a series of alternating hydrophobic residues flanked by glycine located 14 residues upstream from the C terminus and a penultimate tyrosine (11). A recent structural study demonstrated that CssA and CssB are assembled by the principle of donor stand complementation involving the N- and C-terminal motifs (10).

The CssC chaperone protein belongs to the subfamily of FGL periplasmic chaperone proteins involved in the assembly of ultrathin fimbriae. The FGL chaperone protein-assembled adhesive fibers consist of linear polymers of one or two types of protein subunits of an incomplete Ig-like fold. In the polymer, subunits are tightly connected in a head-to-tail fashion by the donor strand complementation mechanism. The FGL chaperone-assisted fimbrial structures are thinner than adhesive pili and have a tendency to aggregate that ensures tight contact between the bacterial pathogen and the host by a zipper-like mechanism (12).

In this study, we used alanine scanning to identify amino acid residues in the N and C termini of CssA and CssB that play an important role in intermolecular interactions between CS6 structural subunits and assembly proteins.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

E. coli XL1-Blue was used for plasmid construction, and E. coli BL21(DE3) was used for expression of recombinant proteins. Clinical ETEC isolate 4266, belonging to serogroup O167 and isolated from a diarrheal patient, which expresses only CS6 as a CF antigen (6), was used in this study. This wild-type (WT) ETEC strain was also used as the parental strain to construct CS6 plasmids pCssABCD (pCS6), pcssACD, and pcssBCD as described earlier (9). All of the strains were maintained at −80°C as a 15% glycerol stock. For the expression of CS6, the bacterial strains were grown in CFA medium (1% Casamino Acids, 0.15% yeast extract, 0.05% MgSO₄, 0.0005% MnCl₂, pH 7.4) at 37°C or in Luria-Bertani medium (BD Difco) with appropriate antibiotics.

Site-directed mutagenesis.

pCS6 (cssABCD cloned into pSTV28) was used as the template (5 to 50 ng) for replacement of respective amino acids with alanine in the N- and C-terminal regions of CssA (GenBank accession no. U04846) and CssB (GenBank accession no. U04844) with the QuikChange Site-directed mutagenesis kit (Stratagene). The genetic code for alanine was GCT/GCA/GCC/GCG. In this study, amino acids are numbered from the beginning of the nascent protein (including the signal peptide). The DNA sequence of the mutated pCS6 template was confirmed with the ABI PRISM 3200 genetic analyzer.

Cell culture.

Human intestinal Caco-2 cells were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 20% fetal bovine serum (Sigma-Aldrich), 1% (vol/vol) nonessential amino acids, and 1% (vol/vol) penicillin-streptomycin solution at 37°C in a 5% CO₂ incubator.

Adhesion assay.

A mid-log-phase bacterial suspension of 10⁷ CFU ml⁻¹ was centrifuged, and the pellet was suspended in DMEM without any antibiotic. This suspension (10⁷ CFU) was added to epithelial cells grown in 12-well plates to 80 to 90% confluence. After 3 h of incubation at 37°C in 5% CO₂, cells were washed three times with phosphate-buffered saline (PBS) and detached with 0.1% Triton X-100. Adherent bacteria were counted after serial dilution by plating on MacConkey agar (BD Difco) plates (13). We used BL21 expressing CS6 as the WT and BL21 as a blank. The binding data are presented as percent binding. Bacterial strains that showed binding similar to that of BL21 were considered to have 0% binding, and strains that showed binding similar to that of the WT were considered to have 100% binding.

RNA expression analysis.

For RNA isolation, strain BL21 bacteria were grown to mid-log phase at 37°C. mRNA was prepared by using bacterial cultures with TRIzol (Invitrogen) in accordance with the manufacturer's protocol. Ten micrograms of RNA was treated with RNase-free DNase I (NEB) and stored at −80°C until further use. Reverse transcription (RT) was performed with 1 μg of total RNA and an RT system (Promega) in accordance with the manufacturer's instructions to obtain cDNA. Equal amounts of RT-PCR products of the cssA and cssB genes of CS6 were run on a 1.5% agarose gel. The parC gene was used as an internal control.

Protein estimation.

Protein was estimated by a modification of the Folin-Ciocalteu method (14) with bovine serum albumin (Sigma-Aldrich) as the standard.

Determination of expression of CS6 subunits on the bacterial cell surface by enzyme-linked immunosorbent assay (ELISA) and Western blotting.

To quantitate the surface expression (SE) of CS6 subunits on the bacterial cell surface, we used 10⁸ CFU of bacteria to coat wells and incubated them overnight at 4°C (15). Unbound bacteria were decanted, and the wells were washed three times with PBS; this was followed by blocking in 5% nonfat skim milk in PBS. After washing, the bound fraction was determined with anti-CssA or anti-CssB polyclonal antibody (1:300) as the primary antibody, followed by a horseradish peroxidase (HRP)-conjugated secondary antibody (1:1,000). The amount of protein was determined with a standard curve obtained with purified CS6. For the standard curve, different concentrations of CS6 were used to coat each well in triplicate and the bound CS6 protein was detected by anti-CssB antibody (13).

For determination of CS6 SE by SDS-PAGE, equal numbers of bacteria were subjected to heat-saline extraction (13). The heat-saline extract (HSE) was separated by 15% SDS-PAGE. Gels were transferred to nitrocellulose for Western blot analysis. Blots were blocked in 5% nonfat skim milk in Tris-buffer saline plus 0.1% Tween 20 and incubated in anti-CssA or anti-CssB primary antibody solution overnight at 4°C and an HRP-conjugated isotype-appropriate secondary antibody for 30 min at room temperature. Blots were developed by HRP-dependent chemiluminescence and visualized by exposure to X-ray film.

Preparation of periplasmic extract (PE).

A bacterial cell pellet was obtained from an overnight culture by centrifugation. Three milliliters of extraction buffer (25% sucrose, 30 mM Tris-HCl, 5 mM EDTA, 1.5 mg ml⁻¹ lysozyme, pH 7.4) was added per gram of cells and incubated for 10 min at room temperature with gentle stirring. Three volumes of ice-cold water was then added, and the mixture was stirred gently for another 10 min on ice. The suspension was centrifuged at 12,000 rpm for 15 min to obtain the periplasmic fraction.

Determination of expression of CS6 subunits in the periplasmic fraction by Western blotting.

PEs from equal numbers of bacteria were separated by 15% SDS-PAGE. Separated proteins were electroblotted to nitrocellulose membrane (Millipore) with an electroblotter (Bio-Rad) and detected by specific antibodies. DnaK was used as a loading control with anti-DnaK antibody (Abcam) for Western blotting.

ELISA for detection of chaperone protein-subunit or subunit-subunit interactions.

To determine protein-protein interactions, rabbit polyclonal antibodies against CssA, CssB, and CssC were diluted 1:100 in carbonate buffer (0.025 M NaHCO₃, 0.025 M Na₂CO₃, pH 9.5). Then, 100 μl of the diluted capture antibody (against CssA) was added to a flat-bottom, high-protein-binding, 96-well microplate (Nunc) and incubated overnight at 4°C (16). The next day, the wells were blocked with 300 μl of blocking buffer (5% nonfat dry milk in PBS–0.1% Triton X-100 [PBS-T]) for 2 h at room temperature. After the wells were washed three times with PBS-T, PEs (0 to 200 μg of total protein) of various mutant proteins were added to a final volume of 100 μl per well and the plates were incubated for 2 h at room temperature. The wells were again washed thrice with PBS-T, 100 μl of either anti-CssC or anti-CssB rabbit polyclonal antibody diluted 1:100 in PBS-T was added to each well, and the plates were incubated for 1 h at room temperature. The wells were washed thrice with PBS-T, and the HRP-conjugated anti-rabbit IgG secondary antibody (Sigma-Aldrich) was added at a 1:1,000 dilution to each well for 30 min of incubation at room temperature. For detection, the color was developed with 3,3′,5,5′-tetramethylbenzidine (BD OptEIA, San Diego, CA) as the substrate. Absorbance at 450 nm was recorded by an ELISA reader (Bio-Rad).

Antibodies used in this study.

The anti-CssA and anti-CssC antibodies against the most antigenic peptides were commercially raised in rabbits by the custom antibody generation service of Thermo Scientific, whereas the anti-CssB antibody against the whole protein was raised in rabbits in our laboratory in accordance with a standard procedure. DnaK was used as a loading control in the case of PE. The anti-DnaK antibody was obtained commercially from Abcam (ab 69617).

Immunoprecipitation assay.

PEs (100 μg of total protein) from various mutant proteins were incubated with specific antibodies (anti-CssA) overnight at 4°C and with preswollen protein A Sepharose beads for 4 h under rotary agitation. Beads were washed thrice with 1 ml of wash buffer (20 mM Tris-HCl, 150 mM NaCl, 0.1% Tween 20, pH 8.0), and bound proteins were eluted by boiling for 5 min with SDS-sample buffer before separation by 15% SDS-PAGE, followed by immunoblotting with specific (anti-CssB or anti-CssC) antibodies.

Protein purification and formation of CssA-CssB oligomers.

Recombinant CS6 was purified from BL21(DE3) bacteria as described previously for the WT strain of ETEC (6). Briefly, the pellet from bacteria grown overnight was suspended in 0.85% NaCl and then incubated at 60°C for 25 min with mild shaking at 30 rpm and centrifuged at 8,000 × g for 20 min; this was the HSE. The HSE was then subjected to ammonium sulfate precipitation, and the 60% ammonium sulfate saturation pellet was purified by anion-exchange chromatography with a Q-Sepharose FF column (GE Healthcare) with a linear gradient of 1 M NaCl–20 mM Tris-HCl (pH 6.8). This was followed by gel filtration chromatography with a Superdex G200 column (1 by 40 cm; GE Health Sciences). Homogeneity was monitored by 15% SDS-PAGE.

The peak fractions obtained by gel filtration of WT and mutant protein samples were mixed with sample loading buffer (without SDS and β-mercaptoethanol) and then analyzed for their oligomerization pattern by 10% native PAGE at room temperature.

CD.

The secondary structure of the protein samples was determined by circular dichroism (CD) (Table 1). Before being studied, the samples (400 μg ml⁻¹) were dialyzed in phosphate buffer, pH 7.4. For measurement in the far-UV (180- to 260-nm) range, a cylindrical quartz cuvette with a path length of 1 mm was used. Each spectrum represents the average of five successive scans performed at a speed of 20 nm min⁻¹ and a bandwidth of 1 nm. Appropriate baseline subtraction and noise reduction analysis were performed. Ellipticities were measured, and molar ellipticities were calculated with the equation ε = (θ · 10)/(1 · c), where ε is ellipticity in millidegrees cm⁻¹ M⁻¹ (M is molecular mass), c is the protein concentration (mg ml⁻¹), and 1 mm is the path length. The relative percentages of different types of secondary structures were calculated by the CDNN software.

TABLE 1.

Secondary structural elements of WT and mutant CS6 determined by CD spectroscopy^a

Protein	Helix	Sheet	β Turn	Random coil
WT CS6	16.5	66.2	21.8	44.3
Mutant CssA proteins
R19A mutant	16.5	66.3	21.8	44.5
E21A mutant	16.5	66.2	21.8	44.3
G141A mutant	15.8	66.3	21.5	45.2
I142A mutant	16.8	64.5	21.3	45.2
V149A mutant	15.8	66.4	21.6	45.2
Y151A mutant	16.3	65.9	22.0	44.6
Y152A mutant	17.2	65.8	21.8	44.0
Mutant CssB proteins
G22A mutant	16.2	65.8	21.9	45.8
Y26A mutant	16.8	58.9	21.3	50.8
L29A mutant	16.5	66.3	21.8	44.4
V31A mutant	16.1	66.4	21.5	45.3
V33A mutant	16.9	65.8	21.8	44.3
I35A mutant	16.3	65.9	22.0	44.6
F39A mutant	17.4	66.3	21.6	47.4

^aShown are the relative percentages of different types of secondary structures in protein samples determined by CD in the far UV range (180 to 260 nm). The values shown were calculated by the CDNN software.

Computational analysis.

The WT and mutant forms of the CssA N terminus (CssA_N)-CssB C terminus (CssB_C) and CssB_N-CssA_C models were constructed on the basis of two recently available donor strand complemented crystal structures, i.e., CssAdsB (Protein Data Bank [PDB] code 4B9I) and CssBdsA (PDB code 4B9G). For this, we used the phyre2 homology-based protein modeling web server (http://www.sbg.bio.ic.ac.uk/phyre2) (17). The models were validated by the ProSA web server (https://prosa.services.came.sbg.ac.at/prosa.php) prior to any further analysis (18). Finally, PyMOL was used to visualize the three-dimensional (3D) models generated.

Statistical analysis.

All of the data were analyzed by one-way analysis of variance software. A P value of ≤0.05 was considered statistically significant. Three independent experiments were done in triplicate sets, and the data represent the mean ± the standard error for these independent events.

RESULTS

Experimental design to examine the CssA and CssB mutant proteins.

In order to investigate the functionally important residues in the CssA and CssB subunits of CS6 during its assembly, we focused on two regions, the N and C termini, as signified by Roy et al. (10). CS6 intersubunit interactions occur by insertion of the CssA donor strand (N terminus) into acceptor clefts (C terminus) of the CssB subunit and vice versa. According to Roy et al., the donor strand in CssA (G_dA) is 15 residues long, whereas the donor strand in CssB (G_dB) is 12 residues long (10). G_dA inserts five residues into the pockets of the acceptor cleft in the CssB subunit, and in a similar way, G_dB inserts five residues into the acceptor cleft of CssA. I33 in G_dA and I35 in G_dB occupy pocket 5 (P5) of the respective acceptor cleft and are essential for initiation of donor strand exchange (DSE). Similarly, T31 in G_dA and V33 in G_dB reside in P4 and P3 is occupied by V29 in G_dA and V31 in G_dB. The donor residues in these three pockets are inserted deeply. However, P1 and P2 are shallow; thus, the donor residues lie on the interactive surface. P1 in both structures is filled by the side chain of K25 in G_dA and K27 in G_dB. P2 has been occupied by large donor residues, F27 in G_dA and L29 in G_dB. In addition, −P1 and −P2 are occupied by the residues involved in nonclassical donor strand interactions. I22 and A23 in G_dA occupy −P2 and −P1, respectively. Similarly, −P2 and −P1 in CssA are occupied by W24 and Y26, respectively, in G_dB.

Here, we targeted 20 amino acid residues from the N termini of both of the subunits for alanine scanning mutagenesis. Hence, we constructed individual alanine substitutions in donor strand residues that occupy the pockets, i.e., I22, A23, K25, F27, V29, T31, and I33 in G_dA and W24, Y26, K27, L29, V31, V33, and I35 in G_dB. We also targeted other amino acid residues that are not inserted into the acceptor cleft, i.e., R19, T20, E21, T24, N26, P28, S30, and T32 in CssA and G22, N23, Q25, S28, D30, N32, and N34 in CssB. Similarly, we replaced 20 amino acid residues with alanine at the C termini of both of the subunits. We targeted all of the residues present in the F strands of CssA and CssB. In addition, we also mutated D135, A136, E137, I138, P139, P140, Y152, V153, and N154 in CssA_C and N148, S149, G150, Y151, S152, and S153 in CssB_C. Next, we tested the abilities of these mutant proteins to assemble functional CS6 by adhesion assay with intestinal Caco-2 cells. Alanine substitutions that result in reduced adherence were tested for the presence of CssA and CssB on the bacterial cell surface and their level of expression in these mutant proteins. An alanine substitution in either subunit that results in a defect in the SE of structural subunits may be due to lack of interaction either between the subunits or with assembly proteins (chaperone and usher). Therefore, each mutant protein with a defect in expression was examined to locate the step in the assembly pathway that was hindered.

Effects of substitution of alanine for the N-terminal amino acid residues of CssA.

We have used alanine scanning to define the role of N-terminal amino acid residues of CssA responsible for interactions with subunit CssB and chaperone protein CssC during CS6 structural assembly. Twenty independent alanine substitutions were made, and CssA N-terminal mutant (CssA_mutN) proteins were generated. These mutant proteins were coexpressed with WT CssB to produce mutant CS6 (CS6_A-N) expressed by E. coli BL21. The binding abilities of E. coli strains expressing these 20 different CS6_A-N proteins were examined by using Caco-2 intestinal epithelial cells (Fig. 1). The data showed that E. coli expressing WT CS6 adhered with a value of 65 × 10⁵ CFU ml⁻¹ (taken as 100%) and E. coli BL21 adhered with a value of 8 × 10⁵ CFU ml⁻¹(taken as 0%). Eleven of the E. coli strains expressing CS6_A-N proteins did not show any difference in binding to Caco-2 cells with respect to E. coli expressing WT CS6. Two of the 20 strains expressing CS6_A-N proteins showed reduced binding to Caco-2 cells. The strains expressing R19A and E21A mutant proteins showed 20 and 40% less binding ability, respectively, than E. coli expressing WT CS6. Seven of the strains expressing CS6_A-N proteins (those expressing the T20A, I22A, K25A, F27A, V29A, I33A, and S36A mutants) failed to show any binding (binding was similar to that of E. coli BL21).

FIG 1.

Cell binding assay of N-terminal CssA mutants. A total of 10⁷ CFU of BL21 bacteria expressing WT or mutated CS6 were incubated with Caco-2 intestinal cells for 3 h, and adherent bacteria were determined by the plate count method. Each bar represents the mean ± the standard error for three independent experiments. *, P < 0.05; **, P > 0.05.

To investigate the reason for this differential binding, we checked the SE of the CS6_A-N mutant proteins. Seven CS6_A-N mutant proteins with no binding ability were not expressed on the bacterial cell surface. Another two CS6_A-N mutant proteins showing reduced binding could be correlated with less SE of CS6_A-N (Fig. 2a). Similar results were also found when SE was analyzed densitometrically by measuring the intensity of the bands obtained from those CS6_A-N mutant proteins that showed either reduced or no binding to Caco-2 cells (Fig. 2b).

FIG 2.

(a) The levels of CS6 on the cell surfaces of 10⁸ CFU WT and CssA_mutN ETEC strains were determined. CS6 was detected by anti-CS6 antibody, followed by incubation with an HRP-conjugated secondary antibody. The amount of CS6 was determined from a standard curve. Each bar represents the mean ± the standard error for three independent experiments. *, P < 0.05. (b) HSEs from equal numbers of WT and CssA_mutN ETEC bacteria were separated by 15% SDS-PAGE and then Western blotted. Western blots were visualized with anti-CssA antibody. (c) Equal amounts of PEs were separated by 15% SDS-PAGE and then Western blotted. Western blots were visualized with anti-CssA and anti-CssB antibodies with DnaK as a loading control.

Next, we examined the PE for the presence of CssA_mutN and CssB in those mutants that were unable to express CS6_A-N on the bacterial cell surface. We detected CssA_mutN in the PEs of three mutants, i.e., those expressing I22A, V29A, and I33A substitutions. In the cases of the T20A, K25A, F27A, and S36A CS6_A-N mutant proteins, we could not detect CssA_mutN. However, all of the strains expressing these proteins expressed RNA at a level similar to the WT. WT CssB was present in the PEs of all of the mutants. It appears that these four mutant proteins were incapable of maintaining CssA_mutN in the periplasm because of a defect in their interaction with CssC (Fig. 2c). The other probable reason might be that alanine substitution led to misfolding of these mutant proteins and made them degradable.

Next we looked for the interaction between CssA_mutN and CssB in CS6_A-N mutant proteins where both subunits were present in the periplasm. We analyzed the I22A, V29A, and I33A CS6_A-N mutant proteins for CssA-CssB interaction. Immunoprecipitation and ELISA failed to demonstrate interactions between the I22A, V29A, and I33A CssA_mutN proteins and CssB (Fig. 3a and b). This suggests that I22, V29, and I33 might be involved in the interaction with CssB.

FIG 3.

(a) Coimmunoprecipitation of CssA and CssB. The immunoprecipitated (IP) complex and input of each sample were probed with anti-CssA or anti-CssB antibody. (b) Detection of CssA-CssB protein complexes by ELISA. PEs from WT and mutant proteins were analyzed for CssA-CssB interaction with anti-CssA antibody as the capture antibody and anti-CssB antibody as the detection antibody. (c) Coimmunoprecipitation of CssA and CssC. (d) Detection of CssA-CssC protein complexes by ELISA.

Further, we looked for the interaction of these three CssA_mutN proteins with chaperone protein CssC by coimmunoprecipitation in PE. In the cases of the I22A and V29A mutant proteins, CssA_mutN was precipitated along with CssC. This demonstrated that these amino acid residues are not involved in the interaction with CssC. However, for the I33 substitution, coimmunoprecipitation produced a negative result (Fig. 3c). This indicated the involvement of I33 in the interaction with CssC. The ELISA results were similar; they also show that CssA with the I22A or V29A substitution was able to interact with the chaperone protein, whereas CssA with the I33A substitution was not (Fig. 3d).

Effects of alanine substitutions of the C-terminal amino acid residues of CssA.

CssA C-terminal mutant (CssA_mutC) proteins were coexpressed with WT CssB to produce mutant CS6 (CS6_A-C) in E. coli BL21. Thirteen of 20 E. coli strains expressing CS6_A-C did not show any difference in binding to Caco-2 cells with respect to that of E. coli expressing WT CS6. The five CS6_A-C mutant proteins carrying G141A, I142A, V149A, Y151A, and Y152A CssA_mutC showed 74, 80, 80, 67, and 71% reductions in binding ability, respectively. Two of the CS6_A-C mutant proteins (the Y143A and V147A mutants) did not show any binding (Fig. 4).

FIG 4.

Cell binding assay of C-terminal CssA mutants. A total of 10⁷ CFU of BL21 bacteria expressing WT or mutated CS6 were incubated with Caco-2 intestinal cells for 3 h, and adherent bacteria were determined by the plate count method. Each bar represents the mean ± the standard error for three independent experiments. *, P < 0.05.

When the SE of these CS6_A-C mutants was examined, the G141A, V149A, Y151A, and Y152A CS6_A-C mutants but not the I142A mutant showed reduced binding because of less CS6_A-C SE (Fig. 5a and b). However, the Y143A and V147A CS6_A-C mutants could not bind because of the absence of CS6_A-C on the bacterial cell surface (Fig. 5a and b). These CssA_mutC proteins could not be detected in the PE (Fig. 5c), maybe for reasons similar to those we predicted for the CS6_A-N mutant proteins, i.e., lack of interaction with the chaperone protein.

FIG 5.

(a) The levels of CS6 on the cell surfaces of WT and CssA_mutC ETEC strains were determined by using 10⁸ CFU of bacteria. CS6 was detected by anti-CS6 antibody. The amount of CS6 was determined from a standard curve. Each bar represents the mean ± the standard error for three independent experiments. *, P < 0.05. (b) HSEs from equal numbers of WT and CssA_mutC ETEC bacteria were separated by 15% SDS-PAGE and then Western blotted. Western blots were visualized with anti-CssA antibody. (c) Equal amounts of PEs were separated by 15% SDS-PAGE and then Western blotted. Western blots were visualized with anti-CssA and anti-CssB antibodies with DnaK as a loading control.

Effects of alanine substitution for the N-terminal amino acid residues of CssB.

As in CssA, we made alanine substitutions in CssB to create CS6_B-N (CssA-CssB_mutN) proteins. The seven E. coli strains expressing CS6_B-N mutant proteins carrying G22A, Y26A, L29A, V31A, V33A, I35A, and F39A CssB_mutN had 61, 74, 62, 74, 70, 68, and 66% less binding ability, respectively, than E. coli expressing WT CS6. In addition, two CS6_B-N mutant proteins (the W24A and E36A mutants) lost the ability to bind intestinal epithelial Caco-2 cells (Fig. 6).

FIG 6.

Cell binding assay of N-terminal CssB mutants. A total of 10⁷ CFU of BL21 bacteria expressing WT or mutated CS6 were incubated with Caco-2 intestinal cells for 3 h, and adherent bacteria were determined by the plate count method. Each bar represents the mean ± the standard error for three independent experiments. *, P < 0.05.

When we examined the SE of the CS6_B-N mutant proteins with reduced binding ability, we found that the seven CS6_B-N mutants (the G22A, Y26A, L29A, V31A, V33A, I35A, and F39A mutants) showed reduced SE (Fig. 7a and b). W24A and E36A mutants showed no CS6_B-N expression on the bacterial cell surface or in PE (Fig. 7b and c). In these two cases, the substitutions in CssB affected the existence of CssB_mutN, as well as CssA, in the periplasm. However, in both cases, subunits were expressed at RNA levels similar to those of the WT. It might be possible that CS6_B-N was more susceptible to proteasomal degradation because of an unstable conformation.

FIG 7.

(a) The levels of CS6 on the cell surfaces of WT and CssB_mutN ETEC strains were determined by using 10⁸ CFU of bacteria. CS6 was detected by anti-CS6 antibody. The amount of CS6 was determined from a standard curve. Each bar represents the mean ± the standard error for three independent experiments. *, P < 0.05. (b) HSEs from equal numbers of WT and CssB_mutN ETEC bacteria were separated by 15% SDS-PAGE and then Western blotted. Western blots were visualized with anti-CssB antibody. (c) Equal amounts of PEs were separated by 15% SDS-PAGE and then Western blotted. Western blots were visualized with anti-CssA and anti-CssB antibodies with DnaK as a loading control.

Effects of alanine substitution for the C-terminal amino acid residues of CssB.

Finally, we substituted alanine for 20 C-terminal amino acid residues of CssB and generated CssB_mutC proteins. These were coexpressed with WT CssA to produce mutant CS6 (CS6_B-C) in E. coli BL21. Fourteen of 20 E. coli strains expressing CS6_B-N did not show any difference in binding to Caco-2 cells (Fig. 8). However, six CS6_B-N mutants (the G154A, Y156A, L160A, V162, F164, and Y165A mutants) showed no binding because there was no expression of CS6_B-C in the HSE, suggesting that these mutants were unable to assemble CS6 on the bacterial cell surface (Fig. 9b).

FIG 8.

Cell binding assay of C-terminal CssB mutants. A total of 10⁷ CFU of BL21 bacteria expressing WT or mutated CS6 were incubated with Caco-2 intestinal cells for 3 h, and adherent bacteria were determined by the plate count method. Each bar represents the mean ± the standard error for three independent experiments. *, P < 0.05.

FIG 9.

(a) The levels of CS6 on the cell surfaces of WT and CssB_mutC ETEC strains were determined by using 10⁸ CFU of bacteria. CS6 was detected by anti-CS6 antibody. The amount of CS6 was determined from a standard curve. Each bar represents the mean ± the standard error for three independent experiments. *, P < 0.05. (b) HSEs from equal numbers of WT and CssB_mutC ETEC bacteria were separated by 15% SDS-PAGE and then Western blotted. Western blots were visualized with anti-CssB antibody. (c) Equal amounts of PEs were separated by 15% SDS-PAGE and then Western blotted. Western blots were visualized with anti-CssA and anti-CssB antibodies with DnaK as a loading control.

In the PE, both CssA and CssB_mutC were detectable (Fig. 9c). However, immunoprecipitation and ELISA failed to demonstrate interaction between CssA and CssB_mutC in these six mutants (Fig. 10a and b). This suggests that G154, Y156, L160, V162, F164, and Y165 might be involved in the interaction with CssA.

FIG 10.

(a) Coimmunoprecipitation of CssA and CssB_mutC. The immunoprecipitated (IP) complex and input of each sample were probed with anti-CssA or anti-CssB antibody. (b) Detection of CssA-CssB_mutC protein complexes by ELISA. PEs from WT and mutant proteins were analyzed for CssA-CssB_mutC interaction with anti-CssA antibody as the capture antibody and anti-CssB antibody as the detection antibody.

Effects of alanine substitutions on CS6 oligomerization.

Since CS6 forms concentration-dependent oligomers in solution (10, 19), we analyzed the oligomer formation of those mutant CS6 proteins that were expressed on the bacterial cell surface at reduced levels. Three CssA_mutC proteins, i.e., the G141A, V149A, and Y151A mutants, showed disruption in the oligomeric pattern, whereas the other two mutant proteins, i.e., the I142A and Y152A mutants, did not show any oligomeric pattern. However, two CssA_mutN proteins with reduced SE (the R19A and E21A mutants) showed similar oligomeric patterns comparable to that of WT CS6 (Fig. 11a).

FIG 11.

Ten percent native PAGE showing CS6 (500 ng μl⁻¹) oligomerization and SDS-PAGE showing heat-denatured CS6 visualized by silver staining. (a) N- and C-terminal mutants of CssA. (b) N-terminal mutants of CssB.

Similarly, seven of the CssB_mutN proteins showed disrupted oligomer formation. The Y26A, V31A, V33A, and F39A mutant proteins were unable to form oligomers, whereas the G22A and I35A mutant proteins had altered oligomeric patterns. However, a single mutant CS6_B-N protein, which had L29A CssB_mutN, did not show any change in oligomer formation (Fig. 11b).

The gel filtration elution profile showed a characteristic shift in the elution time of the peak fraction between WT CS6 and CssA/CssB mutant proteins with altered oligomeric patterns (Table 2). Mutant proteins that showed no oligomer formation in native PAGE eluted much later than the WT because of their low-molecular-weight conformation.

TABLE 2.

Elution times of the peak fraction of WT CS6 and mutant proteins by gel filtration with Superdex G200

Protein	Elution time (min)
WT CS6	44.2
Mutant CssA proteins
R19A mutant	44.2
E21A mutant	44.2
G141A mutant	51.85
I142A mutant	73.4
V149A mutant	52
Y151A mutant	52.3
Y152A mutant	71
Mutant CssB proteins
G22A mutant	50.1
Y26A mutant	73.4
L29A mutant	44.2
V31A mutant	73.4
V33A mutant	73.4
I35A mutant	52.3
F39A mutant	70.95

^a Shown are the times of elution of the peak fractions of the protein samples during gel filtration chromatography with a Superdex G200 column (1 by 40 cm).

The above observations indicate that alanine substitutions at these positions affect the stability of the CssA-CssB oligomeric complex. From these results, we could predict that C-terminal residues of CssA and N-terminal residues of CssB might be involved in higher-order complex formation.

DISCUSSION

A recent structural study demonstrated that CssA and CssB are assembled into functional CS6 by the principle of donor stand complementation involving the N- and C-terminal motifs. CS6 is assembled through an interaction between the N terminus of CssA and the C terminus of CssB (CssA_N-CssB_C) together with a CssA_C-CssB_N interaction (10). By alanine scanning, we identified the amino acid residues in CssA_N, CssA_C, CssB_N, and CssB_C that play important roles in the intermolecular interactions at different stages of CS6 assembly.

We found some, but not all, of the amino acids that are involved in these interactions (Fig. 12). Among the important amino acid residues, we found that mainly alternate hydrophobic residues present in these motifs are essential for the interaction between the structural subunits and the chaperone and usher assembly proteins. Our results clearly indicate the importance of side chains of the amino acids identified in CS6 assembly.

FIG 12.

The core structure of CssB is lime green. The residues involved in CssA_N-CssB_C interaction are orange and brick red in G_dA_N and CssB_C, respectively. (a) The G_dA_N residues that were replaced with alanine are shown as sticks, where bright orange indicates critical residues and light yellow indicates residues with no distinct change. (b) The CssB_C residues that were replaced with alanine are shown as sticks, where brick red indicates critical residues and raspberry indicates residues with no distinct change. Based on the available structure of CssBdsA (PDB code 4B9G). The core structure of CssA is cyan. The residues involved in CssB_N-CssA_C interaction are magenta and blue in G_dB_N and CssA_C, respectively. (c) The G_dB_N residues that were replaced with alanine are shown as sticks, where magenta indicates critical residues and pink indicates residues with no distinct change. (d) The CssA_C residues that were replaced with alanine are shown as sticks, where blue indicates critical residues and light blue indicates residues with no distinct change. Based on the available structure of CssAdsB (PDB code 4B9I).

In the case of CssA_N-CssB_C (CssBdsA in reference 10) donor strand complementation, the donor strand G_dA inserts amino acids K25, F27, V29, T31, and I33 into the acceptor cleft of the CssB subunit (10). P1 and P2 in CssB are occupied by K25 and F27, respectively. Mutation of these two residues in this study resulted in complete loss of the CssA subunit. Apart from these two donor strand residues, we found two other residues of CssA_N (i.e., T20 and S36) that showed no SE and loss of CssA_mut in the periplasm, indicating instability. It is possible that the alanine substitutions leads to misfolding of these mutant proteins and thus become more susceptible to degradation. Alternatively, these amino acids might act as the key residues for chaperone protein-subunit interaction. Disruption of this interaction lead to protein instability and degradation (7). In PapA and CooA, some of the residues in the N terminus have been shown to be important for interaction with the chaperone protein (11, 20).

Apart from these four CssA_N residues, i.e., T20, K25, F27, and S36, mutation of three other residues, i.e., I22, V29, and I33, also resulted in loss of SE, although these were present in the periplasm. We found that I22 and V29 interacted with CssC but CssA_I33A did not. It appears that association with the chaperone protein might not be that important to attain a stable conformation of the CssA_I33A mutant protein. We did not detect any interaction of the I22, V29, or I33 CssA_N mutant protein with CssB. The V29 residue in G_dA is placed deeply inside P3 of the CssB subunit. The alanine substitution at this position may be a poor fit between the donor strand and the acceptor cleft and therefore lead to disruption of the hydrophobic interactions. This explains the instability of the CssA-CssB heterocomplex with the CssA_V29A mutant protein. Residue I33 is located inside a hydrophobic core of P5 surrounded by mainly nonpolar residues Y156, Y151, Y108, and D57 of CssB. Therefore, substitution of alanine for I33 resulted in loss of the CssA-CssB complex because of a loss of hydrophobic interactions. It is also known that P5 is the site of DSE initiation and must be occupied by a large donor residue such as Ile (21, 22). Therefore, in the case of the CssA_I33A mutant protein, DSE is perturbed, which leads to subsequent disruption of CS6 assembly. Interestingly, substitution of alanine for T31, which is placed in P4 of CssB, did not affect CS6 assembly. The side chain of T31 forms an H bond with the main chain of CssB (10). The mutation disrupted this H bonding; however, it did not affect the formation of CS6. This indicates that CS6 stability depends primarily on intersubunit hydrophobic interactions.

The CssBdsA interactive surface also includes nonclassical donor strand interactions that contribute significantly to the total binding area. Such interactions seem to be important for DSE in intersubunit interactions (10). I22 in −P2 of CssB is involved in a nonclassical interaction and might be critical for DSE. I22 is surrounded by residues Y165, F164, and V75 of CssB. Thus, side chain alterations lead to loss of essential hydrophobic interactions and thus prevent CssA-CssB heteropolymer formation by the CssA_I22A mutant protein. Therefore, it is predicted that amino acid residues I22, V29, and I33 are specifically involved in the interaction with CssB mainly through hydrophobic interactions.

In a similar way, when F strand (10) residues G154, Y156, L160, V162, F164, and Y165 of CssB_C were replaced with alanine, intersubunit interaction was lost. We predict that the association between CssA and CssB initiates within the periplasm through the interaction between residues of CssA_N and CssB_C. We could not experimentally define the roles of T20, K25, F27, and S36 of CssA_N in the interaction with CssB_C because of the instability of CssA_mutN in the periplasm. However, structural analysis of CssBdsA suggested that F strand residues of CssB are present in close proximity (within 5 Å) to G_dA residues, i.e., CssB_Y165-CssA_T20, CssB_F164,Y165-CssA_I22, CssB_V162-CssA_K25, CssB_L160,V162-CssA_F27, CssB_L160-CssA_V29, and CssB_G156-CssA_I33. These CssA_N and CssB_C pairs interact primarily through hydrophobic interactions between their side chains. Therefore, alanine substitutions at these positions disturb the crucial hydrophobic interactions that provide stability to the CssA_N-CssB_C complex.

We showed that alteration of G_dB residue Y26, L29, V31, V33, or I35 results in reduced SE of CS6 and an oligomerization pattern different from that of the WT strain. In the case of CssB_N-CssA_C (CssAdsB) donor strand complementation, the donor strand G_dB inserts amino acids K27, L29, V31, V33, and I35 into the acceptor cleft of the CssA subunit (10). The CssB_L29A mutant protein showed an unaltered oligomeric pattern. This might be due to the fact that L29 is not inserted deeply into P2 of CssA. V31 is placed deeply inside P3. An alanine substitution at this position may be a poor fit for stable interactions among neighboring residues, and thus, CssB_V31A is unstable. Hydrophobic residue V33 fits perfectly into highly hydrophobic P4 of CssA. I142 and Y143 of CssA are positioned close to V33. Thus, the side chain substitution leads to a loss of crucial hydrophobic interactions. Similarly, I35 is situated deeply in P5. Residues L59, V52, P140, and G141 of CssA are present within 5 Å of I35, thus creating a hydrophobic environment. Thus, the side chain mutation presumably disrupts the stable hydrophobic interactions among these residues. I35 is also critical for the initiation of DSE. Large aromatic residues W24 and Y26 occupied −P2 and −P1, respectively, and were involved in nonclassical donor strand interactions. Interestingly, the surface topology of the CssB_W24A and CssB_Y26A mutant proteins showed a cavity in the interacting regions of CssA and CssB (Fig. 13b and c). Alanine, with insufficient size, is a poor fit for these pockets and results in an unstable CssA-CssB complex in the case of mutant protein CssB_Y26A, as indicated by an altered oligomeric pattern. However, in the CssB_W24A mutant protein, both CssB and CssA were degraded. It is possible that alteration of W24 leads to an unstable conformation of CS6 to the extent that both CssA and CssB become more susceptible to degradation in the periplasm. In addition to CssB_W24A, CssB_E36A showed a similar effect.

FIG 13.

3D homology model of CssAdsB. The core structure of CssA is cyan. The residues involved in donor strand complementation are orange and magenta in G_dB_N and CssA_C, respectively. (a) Molecular surface representation of WT CssAdsB. (b, c) Molecular surface representations of W24A mutant CssAdsB and Y26A mutant CssAdsB showing cavity formations (circled).

The two other CssB_N mutant proteins, i.e., CssB_G22A and CssB_F39A, also result in reduced SE and an altered oligomerization pattern. The side chain substitution of these two hydrophobic amino acids presumably disrupts the stable hydrophobic interactions.

In a similar way, replacement of F strand residues G141, I142, V149, Y151, and Y152 of CssA_C also showed reduced SE of CS6 and an altered oligomerization pattern. These F strand residues of CssA_C were present close to G_dB residues, i.e., CssA_Y152:CssB_W24, CssA_V147:CssB_L29, CssA_V147:CssB_V31, CssA_Y143:CssB_V33, and CssA_G141:CssB_I35. These hydrophobic interactions also provide stability to CS6. In addition, it is possible that the interaction between the side chains of the amino acid residues of CssA_C and CssB_N is involved in the interaction with usher protein CssD. As a result, defective interaction between CssA and CssB leads to altered oligomeric assembly and reduced SE. However, we observed that in the case of the CssA_I142A mutant protein, oligomeric assembly was altered but there was no change in its expression. This result might be due to improper interaction between the mutant protein and CssD.

We found that Y143 and V147 mutant CssA_C proteins were undetectable in the periplasm. It was known that the stability of nascent CssA depends upon its interaction with chaperone protein CssC in the periplasm (7). We presume that, in this case, nascent CssA was unstable; hence, it was degraded and could not be detected experimentally. A similar phenomenon was also observed in the case of CooA of CS1 (20), where lack of interaction with the periplasmic chaperone protein resulted in the proteolytic degradation of CooA. Here, these two residues, i.e., Y143 and V147, might be involved in the interaction with CssC, mainly through hydrophobic interaction between their side chains. Y143 in CssA showed a function similar that of conserved F152 in PapA and Y159 in CooA; i.e., it was involved in the interaction with the chaperone protein (11, 20). In the case of P pilus structural subunits, the chaperone protein (PapD) recognition motif at the C terminus was characterized by alternating hydrophobic residues flanked by a glycine located 14 residues upstream of the C terminus and a penultimate phenylalanine, i.e., F152 in PapA (11). In CssA, Y143 is located at that position (Fig. 14). The roles of Y143 and V147 of CssA_C in the interaction with CssB_N have been difficult to judge because of the instability of CssA_mutC in the periplasm. In this case also, it is possible that these amino acid residues initially take part in the CssA_C-CssC interaction. When CssC dissociates from the CssA-CssB complex, these residues interact to form stable CS6.

FIG 14.

Alignment of amino acid sequences of conserved C-terminal motifs present in CU-assembled fimbriae. The alignment shows the presence of a conserved glycine at position 14 and an aromatic/hydrophobic amino acid at position 12 from the C terminus (term). The red residues are involved in the interaction with the chaperone protein.

In addition to the hydrophobic interaction between CssA_Y152 and CssB_W24, these two residues may also take part in a π stacking interaction (23). This is the only pair in CS6 exhibiting this interaction. Since W24 is a critical residue involved in such a unique interaction, it is a highly probable drug target to inhibit the formation of functional CS6.

On the basis of the above interpretations and available facts regarding CU-assembled fimbriae, there might be two probable ways CS6 is assembled. CssA and CssB interact with CssC to form CssA-CssC and CssB-CssC complexes. CssA-CssC and CssB-CssC then travel to usher protein CssD and assemble at the usher protein assembly platform one after another to form functional CS6. This classical CU assembly pathway has been reported previously for type I pili and P pili of uropathogenic E. coli (24, 25, 26, 27, 28). The other model of CS6 assembly, proposed by Wajima et al. (9), also supports the formation of CssA-CssC and CssB-CssC complexes first. We predict that in the CssB-CssC complex, the β strand of CssC is replaced by the N-terminal extension of CssA, forming a CssA_N-CssB_C-CssC complex. CssC then targets the CssA_N-CssB_C protomer to usher protein CssD (9), and CssC dissociates from the complex, allowing other important residues to stabilize the CssA_N-CssB_C interaction through the side chains identified. Thereafter, the CssA_C-CssB_N interaction is then driven in the presence of CssD through DSE. The final outcome is a spherical, hetero-oligomeric complex of (CssA-CssB)_n on the bacterial outer surface (19).

The present study adds a step toward understanding the interactions between structural subunits of CS6 and its assembly proteins during CS6 biogenesis. This interaction model might help in the future development of a novel therapeutic intervention against the predominant CF of ETEC.