Fusobacterium spp. target human CEACAM1 via the trimeric autotransporter adhesin CbpF

Abstract

Neisseria meningitidis, Haemophilus influenzae, and Moraxella catarrhalis are pathogenic bacteria adapted to reside on human respiratory mucosal epithelia. One common feature of these species is their ability to target members of the carcinoembryonic antigen-related cell adhesion molecule (CEACAM) family, especially CEACAM1, which is achieved via structurally distinct ligands expressed by each species. Beside respiratory epithelial cells, cells at the dentogingival junction express high levels of CEACAM1. It is possible that bacterial species resident within the oral cavity also utilise CEACAM1 for colonisation and invasion of gingival tissues. From a screen of 59 isolates from the human oral cavity representing 49 bacterial species, we identified strains from Fusobacterium bound to CEACAM1. Of the Fusobacterium species tested, the CEACAM1-binding property was exhibited by F. nucleatum (Fn) and F. vincentii (Fv) but not F. polymorphum (Fp) or F. animalis (Fa) strains tested. These studies identified that CEACAM adhesion was mediated using a trimeric autotransporter adhesin (TAA) for which no function has thus far been defined. We therefore propose the name CEACAM binding protein of Fusobacterium (CbpF). CbpF was identified to be present in the majority of unspeciated Fusobacterium isolates confirming a subset of Fusobacterium spp. are able to target human CEACAM1.

Keywords: CEA; CEACAM1; Fusobacterium; Fusobacterium nucleatum; TAA; Type V secretion; adhesion; binding; host–pathogen interaction; trimeric autotransporter adhesin.

Figure 1.

Densitometric plot of CEACAM1-Fc binding to oral bacterial species. Bacterial isolates were standardised by spectrophotometry, applied to nitrocellulose, and overlaid with CEACAM1-Fc (1 μg ml⁻¹). Mx: CEACAM-binding UspA2V expressing M. catarrhalis strain MX1 (positive control), Nm: Opa negative variant N. meningitidis strain C751 (negative control), Fnv: Fusobacterium vincentii. Means and standard errors of six independent experiments are shown. *p < 0.001.

Figure 2.

CEACAM1 N-domain binding to F. vincentii. Densitometric plot of Fnv overlaid with CEACAM1 (CC1-Fc), CEACAM1 N-domain (N-Fc), CEACAM1 I91A construct (I91A-Fc), or CD33 construct (CD33-Fc) used as a negative control. All constructs were overlaid at 2 μg ml⁻¹. Mean values and standard errors of three independent experiments are shown. *p < 0.05.

Figure 3.

Identification of the CEACAM-binding ligand of F. nucleatum and F. vincentii following Western blotting and co-immunoprecipitation. Western blot of Fv (a) of Fnn (b) under non-reducing conditions overlaid with CEACAM1-Fc (lane 1) or CD33-Fc (lane 2; both 1 μg ml⁻¹) and detected via an appropriate secondary antibody conjugated to AP. CEACAM1 binding to a band of ~150 kDa was observed to be present in lane 1 but no binding was observed in the CD33-Fc control (lane 2). (c) Coomassie stained gel showing co-precipitated proteins from octyl glucoside extracts of Fn and Fv using CD33-Fc or CEACAM-Fc as indicated. A band of ~150 kDa was co-precipitated with CEACAM1-Fc but not CD33-Fc for each subspecies as indicated (*). (d) Corresponding Coomassie stained Western blot of Fv co-precipitated using either CD33 or CEACAM1-Fc constructs. The band indicated (*) represents the CEACAM1-binding ligand of Fv and was subsequently subjected to N-terminal sequencing. Representative data from one of three repeated experiments are shown.

Figure 4.

Identities and domain organisation of CbpF-like autotransporters from Fusobacterial spp. Domain annotation models of the family members indicating number of YadA-like head unit repeats and stalk coiled-coil regions relative to another CEACAM-binding autotransporter UspA1. Models of both UspA1 and YadA are included for comparative purposes. The approximate location of the KW15 epitope is indicated in FN1499 (*).

Figure 5.

Immunological reactivity of anti-KW15 antiserum with co-immunoprecipitated proteins and proteins solubilised using formic acid. (a) Western blot showing binding of anti-KW15 antiserum to the protein co-precipitated using CEACAM1 (CC1) compared to the control co-precipitation which used protein A-sepharose (PAS) alone. Bands were observed at ~150 kDa (*) for both Fn and Fv as indicated. (b) Fn lysate treated with or without formic acid (FA) prior to electrophoresis and Western blotting. The blots were overlaid with anti-KW15 antiserum. Anti-KW15 bound to both a 150 kDa oligomeric (O) band and a 50 kDa monomeric protein band (M), whilst CEACAM1 only bound to the oligomeric form of CbpF. Data are representative of two independent experiments.

Figure 6.

Comparison of CbpF presence in Fusobacterium species. (a) Western blots of lysates of Fusobacterium species nucleatum (Fn lane 1), vincentii (Fv lane 2), polymorphum (Fp lane 3), and a second vincentii strain (formerly F. nucleatum subspecies fusiforme (Fv2 lane 4)) were overlaid with anti-KW15, control pre-bleed serum, CEACAM1-Fc, and CD33-Fc as indicated. Note the CEACAM1-Fc and anti-KW15 binding bands in Fn, Fv, and Fv2 (*) but not in Fp. (b) PCR of cbpF from F. spp. Note the ~1,500 bp product for Fn, Fv, and Fv2 but no product as expected for Fp or Fa. Data are representative of two independent experiments.

Figure 7.

Analyses of CEACAM1 binding and cbpF gene presence in clinical isolates of F. nucleatum. (a) Western blot of representative clinical Fn isolates overlaid with CEACAM1-Fc. Strain designations are indicated above each lane. Note the absence of binding for 2B6 compared to the other isolates which expressed a ~150 kDa CEACAM1-binding protein. (b) PCR for cbpF from representative clinical isolates. Strain designations are indicated above each lane. Note the lack of product for isolate 2B6 corresponding with a lack of CEACAM1 binding observed in A. (c) Analysis of CbpF migration from the two distinct phylogenetic groups. Western blot overlay indicating the relative migration of CbpF from group I (lane 1 and 4) and group II (lane 2). Lane 1, 2B17, lane 2 2B16, lane 3 2B14 (non-binding control), and lane 4 2B13. Blot was overlaid with CEACAM1-Fc at 1 μg ml⁻¹ and detected via an alkaline phosphatase-conjugated secondary antibody. Data are representative of two independent experiments.

Figure 8.

Recombinant CbpF–CEACAM1 interaction. Recombinant CbpF protein was immobilised on ELISA plates and overlaid with CEACAM1-Fc (Black diamonds) or the corresponding I91A mutant (grey squares) Binding was detected using an appropriate alkaline phosphatase-conjugated secondary antibody and chromogenic substrate. Data shown are means of three independent experiments ± SD.

Figure 9.

Recombinant CbpF interactions with different CEACAMs. Of purified CbpFI (ATCC25586) and CbpFII (2B3) (3 pmol) immobilised onto ELISA plates. Subsequently, 1 pmol of each CEACAM-F_C variant were used followed by an anti-human secondary antibody to detect binding. Each CEACAM-F_C conjugate contained all the extracellular IgC- and IgV-like domains except for the ΔN mutant that lacked the N-terminal IgV-like domain. A secondary antibody negative control (Ab) was included. Mouse CEACAM1b (mCC1-Fc) and CEACAM8 (CC8-Fc) were also used as these have not previously been shown to bind any human pathogen adhesins. Three independent replicates are shown ± SD. A one-way ANOVA ($F\left(15\right)=370.8$; $p<0.0001$) followed by a post-hoc Tukey HSD test identified significant differences between CEA and CEACAM1 (***$p<0.0001$) for CbpF I and II. Both CC1-Fc and CEA bound to CbpF I and II significantly more than all other CEACAM constructs used ($p<0.0001$).

Figure 10.

Inhibition of N-Fc binding to a clinical isolate of Fn. Lysate 2B3 was separated by SDS-PAGE using 5% trench well gels. Following Western blotting, strips were but and overlaid with CEACAM1 N-Fc alone (0.2 μg ml⁻¹; lane 1), in the presence of YTH71.3 (anti-CEACAM N-domain antibody raised in rat, 10 μg ml⁻¹; lane 2), in the presence of control antibody Kat4c (anti-CEACAM antibody raised in mouse not recognising N-Fc, 10 μg ml⁻¹; lane 3). Lane 4 strip was overlaid with CEACAM1 N-Fc (0.2 μg ml⁻¹; lane 1) in the presence of a rat isotype control antibody. Lane 5 strip was overlaid with CD33-Fc (0.2 μg ml⁻¹) as a negative control. Receptor binding was detected using anti-human-Fc conjugated to alkaline phosphatase and developed using NBT/BCIP. N-Fc binding to both clinical isolates was completely inhibited in the presence of YTH71.3 but not the control antibody Kat4c or the rat isotype control. Image is representative of two independent experiments.

Figure 11.

Recombinant CbpF interactions with CEACAM1 mutant library. (a) Equimolar amounts of each CbpF protein (3 pmol) were overlaid with 15 fmol of CEACAM1-3-Fc (CC1-3-Fc) mutants in ELISA assays. In addition, CEACAM1-4-Fc (CC1-Fc) and CEACAM1-A1BA2-Fc (ΔNCC1-Fc) were used as positive and negative controls, respectively. In addition, an anti-human IgG-Fc-AP only control was used (Ab). The ELISA was developed for 5 hrs at 37°C using the SigmaFast® kit. (a) two-way ANOVA followed by a post-hoc TukeyHSD test was used to identify significant difference between conditions (N-terminal mutant: $F\left(19\right)=42.86$, $p<0.0001$. Three independent replicates are shown ± SD. (b) Structure of CEACAM1 N-domain displaying relative positions of the amino acids (shown in orange) which were mutated and used for this study. The CFG face is shown in blue, whilst the remainder of the structure is shown in green.