Siglec-7 Mediates Immunomodulation by Colorectal Cancer-Associated Fusobacterium nucleatum ssp. animalis

Fusobacterium nucleatum is involved in the development of colorectal cancer (CRC) through innate immune cell modulation. However, the receptors of the interaction between F. nucleatum ssp. and immune cells remain largely undetermined. Here, we showed that F. nucleatum ssp. animalis interacts with Siglecs (sialic acid–binding immunoglobulin-like lectins) expressed on innate immune cells with highest binding to Siglec-7. Binding to Siglec-7 was also observed using F. nucleatum-derived outer membrane vesicles (OMVs) and lipopolysaccharide (LPS). F. nucleatum and its derived OMVs or LPS induced a pro-inflammatory profile in human monocyte-derived dendritic cells (moDCs) and a tumour associated profile in human monocyte-derived macrophages (moMϕs). Siglec-7 silencing in moDCs or CRISPR-cas9 Siglec-7-depletion of U-937 macrophage cells altered F. nucleatum induced cytokine but not marker expression. The molecular interaction between Siglec-7 and the LPS O-antigen purified from F. nucleatum ssp. animalis was further characterised by saturation transfer difference (STD) NMR spectroscopy, revealing novel ligands for Siglec-7. Together, these data support a new role for Siglec-7 in mediating immune modulation by F. nucleatum strains and their OMVs through recognition of LPS on the bacterial cell surface. This opens a new dimension in our understanding of how F. nucleatum promotes CRC progression through the generation of a pro-inflammatory environment and provides a molecular lead for the development of novel cancer therapeutic approaches targeting F. nucleatum-Siglec-7 interaction.


INTRODUCTION
Colorectal cancer (CRC) is one of the most frequently diagnosed malignancies worldwide, accounting for 10% of all cancers and for approximately 20% of all cancer-related deaths in developed countries (1). While CRC incidence commonly appears in ages over 50 years old, recent years have shown an increased incidence in younger adults which may be associated to lifestyle factors (2,3).
Tumours that arise at epithelial barrier surfaces of the body harbour extensive microbiota, and the importance of these microbes in CRC is now widely acknowledged. The enrichment of Fusobacterium spp. in CRC tissues, as revealed by whole genome sequencing, showed that the most abundant species is Fusobacterium nucleatum (4)(5)(6). Patients with F. nucleatum associated carcinoma have a shorter survival period (7). In addition, F. nucleatum appears to contribute to the chemoresistance of CRC (8)(9)(10). Among F. nucleatum subspecies, F. nucleatum ssp. animalis is most predominant in CRC specimens (11).
Immune cells express a large variety of glycan-binding receptors or lectins, which sense and respond to changes in the glycan signature of their environment leading to the activation or inhibition of immune processes (25). Siglecs (sialic acid-binding immunoglobulin-like lectins) are a large family of lectins found on innate immune cells and tumourinfiltrating T cells, which inhibit immune activation after sensing sialic acid-containing glycans (26,27). Individual family members exhibit preferences for sialosides of various linkages to underlying glycan motifs, but many of the physiological ligands, glycoproteins or glycolipids, they interact with remain largely unknown (28,29). The expression of Siglecs on immune cells is cell type dependent (30). Siglecs are transmembrane proteins, which possess an extracellular portion characterized by a V-set immunoglobulin-like domain, containing the carbohydrate recognition domain (CRD), and one or more C2-set immunoglobulin-like domains. The majority of Siglecs possess immunoreceptor tyrosine-based inhibitory motifs (ITIMs) in their intracellular domain (26). Siglec ligands can be presented on the cell on which the Siglec is expressed (cis ligands), or on glycans in the extracellular matrix of other cells (trans ligands) (28). Although the CRDs of most Siglecs have some specificities towards certain sialylated structures, several Siglecs have a broad and overlapping ligand specificity (28). These glyco-immune checkpoints have been proposed as new targets for cancer immunotherapy (31)(32)(33). The working hypothesis on the role played by Siglecs in cancer is that immune cells expressing Siglecs are inhibited upon binding to their ligands on cancer cells. Indeed, enzymatic removal of sialic acids from cancer cell surfaces was shown to enhance immune cell-mediated clearance of those cells through loss of Siglec-7 and Siglec-9 binding in cis (34) although the range of physiological ligands of Siglec-7 and Siglec-9 remain to be identified (35). Recently, a genome-wide CRISPR screens revealed the glycoprotein CD43 expressed on leukemia cells as a highly specific ligand for Siglec-7 and blocking the interaction relieved Siglec-7-mediated inhibition of immune killing activity (33). In addition, the tumour immune-suppressive effect of Siglec-7 was recently demonstrated in vivo (36), further supporting the proposed role of Siglec-7 as an immune checkpoint receptor.
Several clinically relevant pathogens have evolved mechanisms of molecular mimicry by displaying sialylated structures on their surface to overcome the cis interactions of Siglecs on the surface of immune cells. For example, Campylobacter jejuni strains can interact with Siglec-7 and sialoadhesin (Siglec-1) via their lipooligosaccharides (37) and to Siglec-10 via a sialic acid-like molecule, pseudaminic acid present in the flagella (38), while Siglec-7 showed sialic acid-independent binding to b-protein expressed on Group B Streptococcus surface (39). Additionally, human Siglecs have evolved to recognise non-Neu5Ac ligands present on external stimuli such as microbes (40). In this work we hypothesised that Siglecs may be involved in the recognition of F. nucleatum strains by immune cells, contributing to the tumorigenesis of these strains in CRC.

Bacteria Growth, Preparation and Quantification
F. nucleatum ssp. animalis ATCC 51191 isolated from clinical samples was obtained from ATCC in partnership with LGC standards ltd. F. nucleatum was cultured in tryptic soy broth media (Becton Dickinson) supplemented with 5 mg/ml hemin (Sigma) and 1 mg/ml menadione (Sigma). For binding and human cell co-culture experiments, bacteria were centrifuged at 15,000 x g for 5 min, and the cells were fixed with 4% paraformaldehyde (PFA) (Electron Microscopy Sciences/CN Technical Services ltd) for 45 min at room temperature (RT), in the dark, followed by two washes in Dulbecco's Phosphatebuffered Saline (DPBS) (Lonza).
Bacteria were quantified by spectroscopy with OD 600nm of 1 corresponding to 10 9 cells/ml or by imaging flow cytometry (Amnis ImageStream x Mk II) (as described below).

LPS Extraction
F. nucleatum ATCC 51191 bacterial cells were harvested by centrifugation, lyophilised, and extracted by the hot phenol/water method, as previously described (41). Each phase was dialysed against distilled water to remove the phenol, freeze-dried, and analysed by 12% sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE). After the water/phenol extraction, F. nucleatum LPS extracted from was detected in the water phase by silver nitrate staining (42). The phases containing LPS were further purified by enzymatic digestion (DNAse, RNAse and proteinase K) as previously described (41), followed by centrifugation at 6,000 rpm for 30 min at 4°C and ultracentrifugation at 30,000 rpm for 4 h at 4°C. To separate the O-antigen (OPS) and lipid A domains, LPS were mild acid hydrolysed by acetic acid 1% (100°C, 2-3 h). The OPS domain of the strains containing ulosonic residues was further partial depolymerised. The solution was centrifuged and the supernatants were freeze-dried and further purified by gel filtration chromatography.

F. nucleatum OMV Purification and Characterisation
F. nucleatum derived OMVs were collected from the cell culture supernatant, as described previously by Liu et al. (43) with some modifications. Briefly, F. nucleatum ssp. cells were cultured until reaching OD 600nm of 0.7-1.2. Cells were centrifuged at 8,500 x g for 15 min at 4°C. The supernatant was collected, and vacuum filtered using 0.22 mm membrane. The filtered supernatant was concentrated by spin-filtration using 100,000 molecular weight cut-off filter unit (Sartorius). OMVs were recovered from the filter using sterile DPBS and further purified by density gradient ultra-centrifugation. For the gradient, Optiprep media (60% w/v, Sigma) was diluted in 0.85% w/v NaCl and 10 mM tricine-NaOH pH 7.4 solution to make 35%, 30%, 25% and 20% density solutions. The OMVs were mixed with 40% Optiprep solution and placed at the bottom of a 13.2 ml Ultra-clear tube (Beckman Coulter) and Optiprep (2 ml) was added subsequently by density-decreasing order. The preparation was ultracentrifuged at 135,000 x g for 16 h at 4°C with minimum acceleration and deceleration using a SW41 Ti rotor (Beckman Coulter). From the top to the bottom, 1 ml fractions were collected and analysed by SDS-PAGE in 4-15% Mini-PROTEAN ® TGX ™ Precast Protein Gel (BIO-RAD). The OMV-containing fractions were diluted with sterile DPBS and ultracentrifuged at 200,500 x g for 2 h at 4°C using a Type 45 Ti rotor (Beckman Coulter). OMVs were resuspended in sterile DPBS and then filtered using a 0.22 mm membrane.
Purified OMVs were quantified and measured for their particle size using a NanoSight LM12 (Malvern Panalytical). Briefly, the samples were diluted 100 times in 1 ml DPBS and loaded onto the instrument's chamber by a syringe and the sample were slowly released. The considered particle size of each OMV sample were the mean of triplicates. Instrument settings used: camera shutter 1035, camera gain 680, capture duration 60 sec.

Semi-Quantitative Analysis of LPS in OMVs by Gas Chromatography-Mass Spectrometry (GC-MS)
The content of LPS in F. nucleatum-derived OMVs was evaluated by analysing the fatty acids content. Based on the chemical structure of the lipid A component of F. nucleatum LPS in the bacteria (44), C14:0 (or myristic acid) was considered as the reporter group for LPS, while C16:0 (palmitic acid) and C18:0 (stearic acid) were considered the reporters for the phospholipids.
F. nucleatum-derived OMVs (1 mg) were treated with HCl/ MeOH (1 ml, 1.25 M, 80°C, 16 h) and lipids, derivatised as methylesters, were extracted with hexane (41). This analysis estimated the amount of each fatty acid (C14:0, C16:0 and C18:0) by correcting the areas of the corresponding peaks with a response factor, made by using an array of standard solutions and by setting C16:0 as internal standard. Areas were correlated by a linear regression. The methanol layer after extraction with hexane, was used to countercheck the data from lipid analysis, by verifying the presence of 3-deoxy-2-keto-D-manno-octulosonic acid and L-glycero-D-manno-heptose, both markers of the LPS molecules. Identification of the fatty acids or the monosaccharide constituents, was performed by comparing the retention time and the fragmentation pattern of each peak to a relevant standard.
All chemical derivatives were analysed by using a Gas Chromatography-Mass Spectrometry (GC-MS) Agilent Technologies 7820A (Santa Clara, CA, USA) equipped with a mass selective detector 5977B and a HP-5ms capillary column Agilent, Italy (30 m x 0.25 mm i.d., 0.25 mm as film thickness, flow rate 1 ml/min, He as carrier gas). Electron impact mass spectra were recorded with ionisation energy of 70 eV and an ionising current of 0.2 mA. The temperature program used was: 150°C for 5 min, 150 up to 300°C at 10°C/min, 300°C for 12 min.
Siglec-7-Fc purification was carried out using gravity-flow column (BIO-RAD) packed with protein A-Sepharose (Sigma) washed with DPBS. The harvested CHO supernatant was added to the column and the column washed with DPBS. To elute Siglec-7-Fc, a solution of 0.1 M glycine, pH 3 was added to the column and fractions (0.5 ml) were collected in 1 M Tris, pH 8 (for neutralisation). The protein concentration in fractions was quantified by Nanodrop (Thermo Fisher Scientific).
For the ELISA-based binding assays, bacteria (10 7 cells) or bacteria-derivatives (10 mg/ml LPS or 10 8 OMV particles) in DPBS solution were coated in a 96-well plate, o/n at 4°C. Following a washing step with 0.05% tween in PBS (washing buffer) the plate was incubated with 1% BSA for 1 h at RT. Followed by 3 times washing the plates were incubated with precomplexed Siglec-Fc and a-Fc-HRP for 2 h at RT. Briefly, Siglec-Fc protein (4 mg/ml) was incubated with 50,000X a-human-Fc-HRP (Abcam) for 1 h at RT. Following 3 washes with washing buffer, the plate was incubated with 3,3′,5,5′tetramethylbenzidine (TMB) (Biolegend) until colour development. Colour development was stopped by the addition of 2 N H 2 SO 4 and the absorbance was measured at 450 nm with reference at 570 nm. Data were analysed in GraphPad Prism 6.

STD NMR Analysis
Spectra were acquired on a Bruker 600 MHz AVANCE NEO equipped with a cryo probe and analysed using the TOPSPIN 4.1.0 software. The partial depolymerised OPS derived from F. nucleatum ssp. LPS were prepared in deuterated PBS buffer (20 mM PBS, NaCl 150mM, pH= 7.4), using protein-ligand ratios varying from 1: 20 to 1: 80 with 15 µM of Siglec-7-Fc protein. STD NMR experiments were acquired at 298 K with 32 k data points and zero-filled up to 64 k data points prior to processing. The Siglec-7-Fc resonances were saturated applying 40 Gauss pulses with a length of 50 ms, setting the on-resonance pulse at aromatic region (7.5/6-5 ppm) and the off-resonance pulse frequency at 100 ppm. Under these experimental conditions, very low residual signals were observed in some of the STD NMR spectra for the ligands in the free state which were taken into account during data processing. To suppress the water signal, an excitation sculpting with gradient pulses (esgp) was applied and to reduce the NMR signals of Siglec-7-Fc, a spin-lock filter (20 ms) was used.

Culture of Human Primary Immune Cells and U-937 Monocytic Cell Line
Human peripheral blood was obtained from haemochromatosis patients undergoing a therapeutic venesection at the Norfolk and Norwich University Hospital (Norwich, UK). Blood collection in this study was approved by the Faculty of Medicine and Health Sciences Research Ethics Committee REC reference number 2013/2014 -14HT (University of East Anglia).

Imaging Flow Cytometry
For counting bacteria, 10,000 events were collected and processed using the IDEAS (Amnis) software. Bacteria density ''objects/ml'' were selected in the bright field channel (M04) of the Aspect Ratio_M04 versus Area_M04 dot plot.
For internalisation assays, human cells (monocyte-derived or U-937-PMA) at 5 × 10 6 cells/ml were incubated with 5 × 10 7 FITC-stained F. nucleatum ssp. for 4 h. Cells were washed with FACS buffer and analysed by ImageStreamx Mk II (Amnis). Using the INSPIRE software, a total of 5,000 FITC stained cells were collected. The percentage of internalised bacteria were determined using the internalisation wizard with erode mask function at 7 number of pixels.

RESULTS
F. nucleatum ssp. animalis Binding to Siglecs Revealed Specific Binding to Siglec-7 The binding of F. nucleatum ATCC 51191 was first tested against a range of human recombinant CD33-related Siglec-Fc proteins including Siglec-3, Siglec-5, Siglec-7, Siglec-9 and -10 by flow cytometry. A clear shift in fluorescence was observed in the presence of Siglec-7 with approx. 90% of F. nucleatum population bound to Siglec-7, while 60% and 30% of F. nucleatum population bound to Siglec-5 and Siglec-3, respectively, and only 17% of the population bound to Siglec-9 and -10 ( Figure 1A). To determine if the binding to Siglec-7 was glycan-mediated, inhibition binding assays were carried out in the presence of ganglioside GD3, a known Siglec-7 ligand (48). A significant decrease in Siglec-7 binding to F. nucleatum ATCC 51191 was observed in the presence of GD3, showing an approx. 92% reduction of the bacterial cell population bound to Siglec-7 ( Figure 1B). This result suggests that Siglec-7 Vset domain is implicated in the binding between Siglec-7 and F. nucleatum ssp. To investigate whether the binding of F. nucleatum ATCC 51191 to Siglec-7 was mediated by sialic acid exposed on the bacterial cell surface, the bacterial cells were treated with neuraminidase A, a sialidase with broad specificity to (a2-3,6,8,9) sialylated linkages, cleaving linear and branched non-reducing terminal sialic acid residues from glycoconjugates. The sialidase pre-treatment only led to a small reduction in the binding of F. nucleatum to Siglec-7 as shown by flow cytometry (Figure S1), in agreement with the absence of sialic acid in F. nucleatum ATCC 51191 LPS (24). Next, we conducted binding assays between the monocytic cell line U-937 (wild type (WT) or Siglec-7 -/-) and F. nucleatum ATCC 51191 by flow cytometry. Our results showed a reduction of F. nucleatum associated with U-937-Siglec-7 -/cells as compared to WT cells ( Figure 1C), supporting an interaction between F. nucleatum ssp. and Siglec-7 when expressed on the cell surface although other receptors may be involved in the interaction between F. nucleatum ATCC 51191 and U-937 cells.

F. nucleatum Derived LPS or OMVs Bind to Siglec-7
To determine the role of LPS in the binding of F. nucleatum ATCC 51191 to Siglec-7, LPS was extracted from F. nucleatum ATCC 51191 by the hot phenol/water method (41) and further purified by enzymatic digestion (41). The SDS-PAGE of the extracted F. nucleatum-derived LPS showed the typical LPS ladder-like pattern and lower average molecular weight distribution when compared to E. coli O127:B8 ( Figure S2).
Using an ELISA-type assay, we showed that the whole F. nucleatum cells and the derived LPS bound to Siglec-7-Fc (Figures 2A, B) while no binding of F. nucleatum-derived LPS was observed against Siglec-9-Fc used as a control ( Figure 2B). To map the relevant positions of F. nucleatum LPS involved in the interaction with Siglec-7 and gain a first evaluation of the ligand epitopes, the partially depolymerised O-antigen chain (OPS) isolated from F. nucleatum ATCC 51191 was analysed by STD NMR (49) ( Figure 2D). Interestingly, STD enhancements, together with changes in the relative intensity of STD signals with respect to the reference spectrum, were detected, clearly indicating that F. nucleatum ATCC 51191 OPS structure was recognised by and interacted with Siglec-7-Fc. Despite the significant overlapping of ligand resonances which impaired a detailed analysis of the protons involved in the recognition and binding process, the fingerprint of STD NMR spectrum allowed to identify the ligand regions in close contact with Siglec-7-Fc. Therefore, STD NMR analysis confirmed binding of Siglec-7 to F. nucleatum ATCC 51191 OPS, even though it lacks nonulosonic acid residues ( Figure 2D). Next, we purified outer-membrane vesicles (OMVs) produced by F. nucleatum ATCC 51191 by density gradient ultracentrifugation, resulting in pure and spherical particles with a diameter range from 30 to 250 nm ( Figure S3A). We showed by GC-MS that LPS is a main constituent (approximately in 60-70% mol/mol) of F. nucleatum-derived OMVs ( Figure S3B). We then tested the ability of F. nucleatum-derived OMVs to bind to Siglec-7 ( Figure 2C). F. nucleatum-derived OMVs from ATCC 51191 bound to Siglec-7 at levels comparable to LPS under the conditions tested ( Figure 2C).
Together these data identified F. nucleatum LPS present on whole cells or OMVs as a new ligand to Siglec-7-Fc.

F. nucleatum Modulates Immune Response in a Cell Subset Specific Manner
To investigate the impact of F. nucleatum ssp. on the host immune response, myeloid cells, moDCs and moMfs, were generated from human blood, and stimulated with F. nucleatum ATCC 51191 or with F. nucleatum ATCC 51191-derived LPS and OMVs.
F. nucleatum bacterial cells were shown to associate with the cell surface of moDCs or moMfs as determined by imaging flow cytometry ( Figure S4). Stimulation of moDCs with F. nucleatum at MOI 5 resulted in a marked increase in cytokine production of TNFa, IL-8 (p < 0.0001) ( Figure 3A) and an induction of CD86  and PD-L1 as compared to the unstimulated control ( Figure 3B). A different profile was observed with moMfs, where stimulation with F. nucleatum led to a significant induction of IL-10 and IL-8 production (p < 0.0001), low levels (n.s.) of TNFa production ( Figure 3A) and to the induction of PD-L1 and downregulation of CD86 cell surface markers as compared to the unstimulated control ( Figure 3B). The results were dose-dependent, with a marked increase in cytokine production when cells were stimulated at MOI 50 as compared to MOI 5 ( Figure S5A). This acquired moMf phenotype was also observed using the macrophage like cell line U-937 after F. nucleatum ATCC 51191 stimulation, leading to high IL-10 and low TNFa levels ( Figure S5B). Consistent with these results, we showed, using imaging flow cytometry, that both moDCs and moMfs were able to internalise F. nucleatum ( Figure 3C), with moDCs showing approx. 10% less internalisation as compared to moMfs ( Figure 3C). Next, we stimulated moDCs or moMfs with F. nucleatum ATCC 51191-derived LPS or OMVs. In moDCs, treatment with OMVs or with LPS at 10 mg/ml but not 1 mg/ml induced TNFa production ( Figures 4A, B). In moMfs, stimulation OMVs or LPS (at both 10 or 1 mg/ml) showed significant induction of IL-10 at levels comparable to the whole bacteria ( Figures 4A, B). When moMfs or moDCs were treated with LPS or OMVs, there was an upregulation of the CD80 cell surface marker expression as compared to the unstimulated control, as showed with the whole bacteria. In moDCs, LPS stimulation led to an induction of CD86 expression ( Figure 4C), as also observed with the whole bacteria, while stimulation with OMVs showed a reduction of CD86 expression compared to the unstimulated control shows the superimposition of the reference 1 H NMR spectrum (in black) and STD NMR spectrum (in green), the 1 H-13 C HSQC spectrum (blue/red) and the chemical structure of F. nucleatum OPS repeating units. Statistical analyses were performed by t-test (for panels 2Aand 2C) or one-way ANOVA followed by Tukey's test. P < 0.05 was considered as statistically significant. **p < 0.01, ***p < 0.001, ****p < 0.0001, n.s., not statistically difference. ( Figure 4D). Stimulation of moMfs with LPS (at 10 or 1 mg/ml) showed a reduction of CD86 expression ( Figure 4C), as also observed with the whole bacteria. Overall, our results suggest that moDCs stimulated with F. nucleatum ATCC 51191 and derived components (OMVs and LPS) show a pro-inflammatory profile while F. nucleatumtreated moMfs acquire a M2-phenotype which is associated with tumour progression (50).

Siglec-7 Is Involved in F. nucleatum-Mediated Immune Response
To obtain direct evidence of the contribution of Siglec-7 in F. nucleatum ATCC 51191 interaction with myeloid human cells, we used CRISPR-Cas9 editing to generate Siglec-7 deficient U-937 cells and assayed the effect of F. nucleatum stimulation on the immune response of differentiated U-937 WT or Siglec-7-deficient (Siglec-7 -/-) cell lines ( Figure 5). The expression of Siglec-7 in these cell lines was confirmed by flow cytometry ( Figure S6). A significant increase (p < 0.01) in TNFa and IL-10 cytokine production was observed in F. nucleatum ssp.-stimulated U-937-Siglec-7 -/as compared to WT cells ( Figure 5A).
Next, we carried out silencing of Siglec-7 in primary moDCs and moMfs by siRNA. Using flow cytometry, we confirmed that Siglec-7 was expressed on the cell surface of moDCs and moMfs ( Figure S7A), and that expression could be reduced by up to 40% in moDCs as compared to the scramble control ( Figure S7B) while no significant reduction in expression could be achieved in moMfs. We then analysed the cytokine profile and expression of cell surface markers following stimulation of silenced or scramble control moDCs with F. nucleatum ATCC 51191 at MOI 5 ( Figure 5B). We showed that stimulation of Siglec-7 silenced moDCs with F. nucleatum produced statistically significant (p < 0.05) lower TNFa levels as compared to scramble moDCs ( Figure 5B). No differences in cell surface marker expression were observed between F. nucleatumstimulated Siglec-7 silenced or scramble moDCs ( Figure 5C). Together these data suggest that Siglec-7 is involved in the induction of a pro-inflammatory response in moDCs by F. nucleatum ATCC 51191. (C) Internalisation of F. nucleatum ATCC 51191 into moDCs or moMfs. Images were taken with a 40X objective. For the cytokine quantification, data shown are the mean of triplicates ± SD derived from one representative experiment reproduced in three independent experiments. Statistical analyses were performed by one-way ANOVA followed by Tukey's test. P < 0.05 was considered as statistically significant. ****p < 0.0001, n.s., not statistically difference.  Data shown are the mean of triplicates ± SD derived from one representative experiment reproduced in three independent experiments. Statistical analyses were performed by one-way ANOVA followed by Tukey's test. P < 0.05 was considered as statistically significant. **p < 0.01, ****p < 0.0001, n.s., not statistically difference. while the FadA adhesin, recognises and binds host surface components, such as vascular endothelial cadherin (54) and epithelial cadherin (16). F. nucleatum FadA expression was also shown to be upregulated in CRC tissue (16). Interaction of FadA with E-cadherin triggers the expression of oncogenes, such as c-MYC and inflammatory genes, through the b-catenin cascade and the upregulation of annenix A1 (16,17). However, the receptors involved in the interaction between F. nucleatum ssp. and immune cells remain largely unknown.
Here, we showed that F. nucleatum ssp. animalis ATCC 51191 interacts with Siglec-7 expressed by immune cells and that binding is LPS-mediated. Human NK cells, macrophages, dendritic cells constitutively express Siglec-7, and the colonic lamina propria monocytes and macrophages represent the major Siglec-7 positive populations (55). Extracellularly, Siglec-7 has a sialic acid-binding V-set domain which we demonstrated was implicated in the binding to F. nucleatum. Siglec-7 has been shown to bind to the sialylated ganglioside GD3 (56), and Nlinked disialyl Lewis a in the normal colonic epithelium (55). At the molecular level, Siglec-7 has been reported to bind to terminal sialic acid moieties with diverse underlying glycan structures. We recently uncovered the LPS structure of F. For the cytokine and internalisation analyses, data shown are the mean of triplicates ± SD and duplicated ± SD, respectively, derived from one representative experiment reproduced in three independent experiments. Statistical analyses were performed by two-way ANOVA followed by Tukey's test. P < 0.05 was considered as statistically significant. ***p < 0.001, ****p < 0.0001, n.s., not statistically difference.
GlcpNAc3NAlaA-(1!3)-a-D-FucpNAc4NR-(1!], (R= Acetylated 60%), and a bis-phosphorylated hexa-acylated lipid A moiety (24). It therefore likely that the LPS glycans other than sialic acid moieties may contribute to the binding of F. nucleatum ATCC 51191 to Siglec-7, consistent with the results of the sialidase treatment. This was further confirmed by STD NMR, showing that the OPS extracted from F. nucleatum ATCC 51191 was recognised by Siglec-7, revealing new ligand epitopes not restricted to nonulosonic acids (neuraminic acid and fusaminic acid). The discovery that F. nucleatum LPS is a ligand for immune checkpoint receptors like Siglec-7 opens up new blockade strategies and studies are in progress to gain further structural insights into the broad ligand specificity of Siglec-7 towards the bacterial glycan structures revealed in this work.
We showed that F. nucleatum ATCC 51191 induced a proinflammatory profile in moDCs and a tumour associated profile in macrophages (moMfs and U-937 cells) and that Siglec-7 contributed to these cell-specific responses using Siglec-7 RNA-silenced moDCs and Siglec-7 deficient U-937 cells. In macrophages, F. nucleatum ATCC 51191 induced the expression of IL-10, IL-8 cytokines and PD-L1 marker and a downregulation of CD86 cell surface marker expression, characteristic of macrophage type 2 (M2) polarisation (12,13). These results are in agreement with previous studies showing an infiltration of M2-macrophages in F. nucleatum ssp. positive clinical CRC specimens (57) and a M2 acquired phenotype in macrophage-like cell lines stimulated with F. nucleatum ATCC 10953 (13), and F. nucleatum ATCC 25586 (58). A recent study showed that Siglec-7 and -9 induce the polarisation of monocytes into a tumour-associated macrophage (TAM) phenotype and the induction of tumour-associated cell surface markers such as PD-L1 (59). The moDC response to F. nucleatum ATCC 51191 suggests that Siglec-7 contributes to the pro-inflammatory response. These cellspecific phenotypes could be recapitulated using F. nucleatum derived OMVs or LPS, implicating LPS as a ligand of the interaction with Siglec-7. The interaction of F. nucleatum derived LPS has been reported with TLR-4 leading to polarisation of macrophages, a process that is associated with tumour cell proliferation and metastasis (13). Our findings that LPS and OMVs influence innate immune cell responses is supported by recent studies showing that F. nucleatum OMVs can trigger inflammation of human intestinal epithelial cells (IECs) (60), by promoting NF-kB activation in a TLR-2-dependent manner (61).
Siglec-7 can bind to a range of human cell types (such as basophils, eosinophils, NK cells and splenocytes), illustrating its role of 'self' recognition (62). The response we observed upon interaction with F. nucleatum differs from the canonical inhibition of immune activity observed between immune cells and cancer cells (25) but is consistent with in vivo work using ApcMin/+ model reporting that F. nucleatum induced expression of the genes encoding several pro-inflammatory cytokines, including TNFa, IL-6, IL-8 and IL-1b (12) which mirrors human RNA-seq data from patients bearing high F. nucleatum loads in their colorectal tumours (12). To our knowledge, only one study, using Siglec-7 silencing approach in monocytes, also showed association of Siglec-7 with pro-inflammatory cytokine production, upon interaction with yeast particles in a sialic acid-independent manner (63). These differences in cell immune response could be attributed to the mode of recognition, the nature of the interactions (cell-cell or cellmicrobe), or the heterogeneity in pattern recognition receptor (PRR) expression in the different cell subsets used in the in vitro studies. Indeed, other PRRs may act synergistically with Siglec-7 to contribute to a distinct immune response. For example, TLR-4, a toll-like receptor with an intracellular activation motif has been shown to establish a direct interaction with Siglecs including Siglec-7 (64). Therefore, since TLR-4 is expressed in U-937 (65) and moDCs (66), our findings could be the result of a synergetic effect between Siglec-7 and TLR-4. This interaction could also contribute to the capacity of F. nucleatum ssp. to promote chemoresistance of CRC by inhibition of cancer cell apoptosis (8)(9)(10). A recent in vivo study using humanized immunocompetent mice, showed that Siglec-7 and -9 could be potential targets to enhance anti-tumour immunity (36). In the future, it will be interesting to study the effect of F. nucleatum-Siglec-7 interaction in vivo, using humanised immunocompetent murine model, as Siglec-E, the closest murine homolog of Siglec-7, does not recognise F. nucleatum ssp. (data not shown), consistent with the lack of direct homology between murine and human Siglecs (67).
It was recently reported that Siglec−7 is expressed in macrophages in CRC tissue from patients and that high levels of Siglec−7 expression in tumour tissues are associated with shorter overall survival in patients treated with immunotherapy for metastatic CRC (68). Mirroring this, an independent human study reported that patients with high relative abundance of F. nucleatum in tumour tissues compared to matched control tissues have a higher incidence of regional lymph node metastases (69). These studies support the translation of our findings to humans. The interaction of F. nucleatum ssp. with Siglec-7, leading to a proinflammatory microenvironment, provides a mechanism underpinning these associations in patients and initial evidence that blocking this interaction may be a potential strategy to alleviate the progression of F. nucleatum associated CRC.
In summary, our results reporting LPS-mediated interaction of F. nucleatum and derived OMVs with Siglec-7, add a new dimension in our understanding of the role of Siglecs in cancer progression. Given the role of F. nucleatum in influencing CRC tumorigenesis and response to cancer treatment, there is significant interest in developing strategies that target F. nucleatum, preferably in the tumour tissue. However, antimicrobial strategies are limited due to concerns about antibiotic resistance and the mutualistic role of F. nucleatum in the oral cavity and other mucosal sites of humans (70). Targeted glycan interventions to displace Siglec-7-F. nucleatum interactions could prove an effective way of improving current approaches for the treatment of cancer by targeting F. nucleatum in the tumour environment and without compromising the rest of the gut microbiome or inducing antimicrobial resistance.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding author.