Plasmids and constructs used in this study are listed in Table 1, and sequences can be found in the supplements (Table 1). Constructs made in this study were cloned using the Gibson assembly (Gibson et al., 2009).

Bacteria were cultivated in Lysogeny broth (LB, Miller formulation). For whole-cell assays with fluorescence detection, Escherichia coli Top10 glmS:sfGFP (AS75) was used and grown in the presence of arabinose (Saragliadis and Linke, 2019). For protein purification, genes encoding the proteins of interest were expressed in E. coli BL21 (DE3) Gold. Generally, bacteria were grown at 37°C to the desired OD₆₀₀. During overexpression, the temperature was shifted to 23°C after induction.

pASK-IBA3_YadA_O:8/O:9 was transformed into E. coli BL21 (DE3) Gold and grown on ampicillin plates. A single colony was inoculated into 20 ml of LB medium supplemented with 100 μg/ml ampicillin and grown at 37°C overnight (o/n). The following day, a 2 L subculture was prepared and grown in a home-built fermentation system (a system where air is bubbling through bottles of growth medium that stand in a temperature-controlled water bath) until an OD₆₀₀ of 0.5–0.7 was reached. The temperature was shifted to 23°C, and expression was induced with 0.2 μg/ml anhydrotetracycline (AHTC). Protein expression was allowed for 16 h. The culture was harvested by centrifugation at 4,000 × g. Afterward, the pellet was resuspended in Tris-buffered saline (TBS) buffer (20 mM Tris pH 7.5, 300 mM NaCl, 20 mM imidazole) with 8 μg/ml lysozyme and a pinch of DNAse. The suspension was subjected to cell lysis using a French press after addition of a HALT protease inhibitor mix (1:500, Thermo Fisher Scientific; 1861278). The lysate was centrifuged for 1 h at 69,600 × g, and the supernatant was then filtered through a 0.2 μm filter and subjected to Ni-NTA affinity chromatography (Cytiva, 17531901). As YadA with C-terminal His₆-tag elutes at high imidazole concentrations (160–500 mM), the protein was pure enough for binding experiments after Ni-affinity chromatography. The protein was subjected to dialysis against TBS buffer (20 mM Tris pH 7.5, 150 mM NaCl).

E. coli AS75 with pASK-Iba4C_YadA_O:8 or pASK-Iba4C_YadA_O:9 was grown o/n in LB medium supplemented with 0.02% (w/v) arabinose and 100 μg/ml ampicillin. The next day, the cultures were diluted 1:100 in 20 ml LB medium supplemented as before and grown at 37°C to an OD₆₀₀ of 0.5. YadA expression was then induced by the addition of AHTC to a final concentration of 0.2 μg/ml and grown for another 3 h at 37°C. YadA expression was checked for by visual inspection for auto-aggregation (Trunk et al., 2018). In the meantime, clear flat-bottom 96-well plates were coated with 100 μl of a 10 μg/ml Vn solution, from either plasma (Gibco, PHE0011), recombinantly expressed in HEK cell cultures (Merck/Millipore, SRP3186), or E. coli (Thermo Fisher Scientific, A14700), by incubation for 1 h at room temperature (RT). The Vn solution was discarded from the plates, and the wells were washed three times with TBS (20 mM Tris pH 7.5, 150 mM NaCl). Afterward, the wells were blocked using 3% bovine serum albumin (BSA) in TBS. The bacteria were then harvested, washed twice with TBS, and resuspended in TBS with 0.1% BSA to achieve an OD₆₀₀ of 0.2. One hundred microliters of the bacterial suspension was added per well and incubated for 1 h at RT. After that, the wells were washed three times using TBS. Lastly, the wells were filled with 100 μl TBS buffer, and fluorescence was measured using an excitation wavelength of 488 nm and recording the emission at 533 nm (BioTek Synergy H). For experiments with deglycosylated Vn, the experiment was performed the same way, but deglycosylated Vn (see section “Deglycosylation of Vitronectin”) was used for coating.

A clear 96-well plate was coated with Vn and blocked as described before (section “Vitronectin Binding Experiments With Whole Bacteria”). Then 100 μl of a 10 μg/ml YadA solution in 0.1% (w/v) BSA in TBS (20 mM Tris pH 7.5, 150 mM NaCl) was added to the wells and incubated for 1 h at RT. After the wells were washed twice with 0.1% BSA in TBS and once with TBS with Tween-20 (TBS-T), Ni-horseradish peroxidase (HRP) conjugates were used for the detection of bound, His₆-tagged YadA. One hundred microliters of a Ni-HRP conjugate solution at a final concentration of 5 μg/ml (Thermo Fisher Scientific, 15165) was incubated for 1 h at RT in 3% BSA in TBS per well. This was discarded, and the wells were washed three times with TBS-T and once with TBS. Binding was detected using 150 μl of a 1 mg/ml ABTS solution in ABTS buffer (2.43 ml of 100 mM citric acid, 2.57 ml of 200 mM Na₂HPO₄, 5.0 ml H₂O, 10 μl H₂O₂) (Thermo Fisher Scientific, 34026; VWR, ICNA0219502305). The color development was stopped by adding 100 μl of 1% (w/v) SDS after incubation at RT, and the absorption was measured at 405 nm in a BioTek Synergy H plate reader.

For deglycosylation of Vn, 20 μg of Vn from the respective sources (in water) was mixed with 2 μg of the glycopeptidase PNGase F (500 U) (Promega, V4831) and incubated at 37°C for 19 h. The non-deglycosylated control samples of Cn were incubated at 37°C for 19 h, omitting the PNGase F. For PNGase F control samples, 2 μl of PNGase F was added to water and incubated as described before. Successful deglycosylation was checked for on a SDS-PAGE gel with subsequent silver staining (Nesterenko et al., 1994).

Glass coverslips were coated with 50 μl Vn (10 μg/ml) at 4°C o/n. An o/n culture of E. coli AS75 harboring pASK-IBA4C_YadA_O:8/O:9 was inoculated into LB supplemented with 20 μg/ml chloramphenicol and 0.02% w/v arabinose. The next day, the culture was diluted 1:100 in the same broth, and the culture was grown to OD₆₀₀ of 0.5 followed by induction with 0.2 μg/ml AHTC and yadA expression for 3 h at 37°C. In the meantime, Vn-coated coverslips were incubated with TBS or 100 μM heparin-disaccharide I-S (Merck, H9267-1MG) for 1 h at RT where applicable. After that, all coverslips were blocked with 3% (w/v) BSA in TBS for 1 h at RT. One hundred microliters of 5 × 10⁸ bacteria in suspension were centrifuged at 4,000 × g for 5 min and resuspended in either TBS (20 mM Tris pH 7.5, 150 mM NaCl) or 100 μM heparin-disaccharide in TBS and incubated for 1 h at RT. After that, the bacteria were centrifuged down again and washed three times in 100 μl TBS. Finally, the bacteria were resuspended in 1 ml 3% BSA in TBS. Three hundred microliters of the bacteria was added to the coverslips and incubated for 30 min at RT. The supernatant was discarded, and the coverslips were washed three times with TBS and fixed with 500 μl of 4% (w/v) paraformaldehyde in TBS for 20 min at RT. Finally, the coverslips were mounted in 5 μl ProLong Glass Antifade Mountant (Invitrogen, P36980) and dried o/n. Microscopy was performed using a fluorescent microscope (Zeiss Axioplan 2) and a 100× oil immersion objective. For quantification, images were converted into binary files, and the area of the particles was calculated using Fiji (Supplementary Figure 3). Mean areas were plotted including the standard error of the mean.

An o/n culture of E. coli AS75 harboring pASK-IBA4C_YadA_O:8/O:9 was inoculated into LB supplemented with 100 μg/ml ampicillin and 0.02% (w/v) arabinose. The next day, the culture was diluted 1:100 in the same broth and grown to OD₆₀₀ of 0.5 followed by induction with 0.2 μg/ml AHTC and yadA expression for 3 h at 37°C. The culture was diluted to an OD₆₀₀ of 1.0, and 50 μl was centrifuged down at 4,000 × g for 5 min. The pellets were then resuspended in 50 μl TBS or TBS supplemented with 100 μM heparin-disaccharide (Merck, H9267-1MG). This was incubated at 37°C in a shaking incubator for 30 min. Five microliters of each solution was wet-mounted onto microscope slides, and the edges were sealed using a CoverGrip coverslip sealant (Biotium, 23005). Microscopy was performed using a fluorescence microscope (Zeiss Axioplan 2) and a ×100 oil immersion objective. For quantification, images were converted into binary files, and the particle sizes were calculated using Fiji (Supplementary Figure 4). The area of each individual particle was plotted in a column scatter plot.

Nitrocellulose membranes were cut and transferred into a six-well plate. Three 2 μl drops of a 700 μg/ml purified YadA_O:8 or YadA_O:9 solution were applied onto the membrane and air-dried. Then, the membrane was blocked with 5% BSA in TBS-T (20 mM Tris pH 7.5, 150 mM NaCl, 0.2% Tween-20) for 1 h at RT. Five hundred microliters of a 100 μM biotinylated heparin (Merck, B9806-10MG) solution in TBS-T was incubated on the membrane for 1 h at RT. The membrane was washed three times with TBS-T and afterward incubated with 500 μl of 1:10,000 diluted Strep-Tactin–HRP conjugate (IBA Lifesciences, 2-1502-001) in 5% BSA in TBS-T for 30 min at RT. After the membrane was washed three times with TBS-T and once with TBS (20 mM Tris pH 7.5, 150 mM NaCl), a 500 μl ECL reagent (Thermo Fisher Scientific, 320106) was added, and the membrane was immediately imaged using a Kodak Image Station 4000R.

An o/n culture of E. coli AS75 pASK-IBA4C_YadA_O:8/O:9 was grown in the presence of 0.2% (w/v) arabinose and 100 μg/ml ampicillin. This culture was diluted 1:100 the next morning and grown to an OD₆₀₀ of 0.5. YadA full-length expression was induced by addition of 0.2 μg/ml AHTC. Expression was allowed for 3 h at 37°C. Uninduced bacteria were used as a control. The bacteria were diluted to an OD₆₀₀ of 0.2, spun down, and resuspended in phosphate-buffered saline (PBS). One hundred microliters of that bacterial solution was pipetted into 96-well plates and centrifuged at 4,000 × g. After that, 100 μl of a 10 μg/ml biotinylated heparin (Merck, B9806-10MG) solution in 3% (w/v) BSA in TBS was added and incubated at RT for 0.5 h. The plate was washed three times with TBS. The plate was centrifuged as before after every wash before discarding the washing buffer. Strep-Tactin–HRP conjugates (IBA Lifesciences, 2-1502-001) at 1:1,000 in 3% (w/v) BSA in TBS were added and incubated for 30 min at RT. The plate was washed as described before. The ABTS solution was prepared, and color development was stopped as described before. Wells that did not contain any bacteria were used as background controls. Absorbance at 405 nm was measured in a plate reader (BioTek Synergy H).

One hundred microliters of 10 μg/ml YadA in TBS was coated into a 96-well plate by incubation at RT for 1 h. The plate was washed three times with 200 μl TBS (20 mM Tris pH 7.5, 150 mM NaCl) and blocked using 200 μl of 3% BSA in TBS. Afterward, 100 μl of biotinylated heparin dilution (0–6.75 μg/ml) in TBS was added to the wells and incubated for 1 h at RT. The wells were washed three times with TBS as described above and blocked with 3% BSA in TBS for 1 h at RT. Strep-Tactin–HRP (IBA Lifesciences, 2-1502-001) at 1:1,000 was added in 3% BSA in TBS and incubated for 1 h at RT. The wells were washed again as described earlier, and an ABTS solution was used for detection as described before. After color development, the reaction was stopped by adding 100 μl of a 1% SDS solution. Absorbance at 405 nm was measured in a plate reader (BioTek Synergy H).

For binding data analysis, data are shown as means ± SD and were analyzed using a one-way ANOVA including Tukey’s test. For data plotting and statistical analysis, OriginPro and “R” were used. For microscopy analysis, mean particle areas ± SEM were plotted. As the particles sizes were not normally distributed, non-parametric testing including a Kruskal–Wallis test and subsequent Wilcox testing were applied to test for significance. Significance levels are indicated in the graphs with p < 0.05 (*), p < 0.01 (^**), or p < 0.001 (^***).

We first aimed to describe the molecular details of the interaction between the YadA head domain of Y. enterocolitica serotype O:9 (YadA_O:9) and Vn. YadA_O:9 harbors an additional, N-terminal, 31-residue stretch (Figure 1A) that has been described to interact with Vn (Mühlenkamp et al., 2017). We started out replicating the experiments done by Mühlenkamp et al. (2017). For these enzyme-linked immunosorbent assay (ELISA)-like binding experiments, Vn from different sources was used. Vn purified from plasma (Vn_plasma), Vn expressed in HEK cell culture (Vn_HEK), and Vn recombinantly expressed in E. coli (Vn_Ec) were tested for their capacity to be bound by E. coli expressing either full-length YadA_O:8 or YadA_O:9 (Figure 1B). Additionally, Vn binding by purified YadA head domains was tested (Figure 1C). We reasoned that, if YadA indeed bound to a conserved sequence within Vn, the binding should happen irrespective of the origin of Vn and only with YadA_O:9. Indeed, the whole-cell assays show that only YadA_O:9-expressing bacteria bound to Vn_plasma and Vn_HEK (Figure 1B). Binding between Vn_Ec and YadA_O:9 could not be observed. Bacteria expressing YadA_O:8 did not bind to either Vn variant (Figure 1B). These findings were corroborated by assays using purified YadA head domains. While we observed some binding of purified YadA_O:8 to Vn_plasma, no binding to Vn_HEK and Vn_Ec was observed (Figure 1C). The weak binding of purified YadA_O:8 can be explained by Vn_plasma being contaminated with other proteins (Supplementary Figure 1). YadA_O:9 on the other hand showed clear binding to all Vn variants, with reduced binding to Vn_Ec. Based on these findings, we sought to investigate the difference between YadA_O:9 binding to Vn_plasma/HEK and binding to Vn_Ec.

Due to the observation that YadA_O:9 binds Vn_plasma and Vn_HEK but shows at least reduced binding to Vn_Ec, we wanted to investigate whether YadA_O:9 actually binds a stretch within Vn or whether it either recognized a folded binding site or the glycosylations of Vn. Vn is heavily glycosylated with at least three N-linked glycans at residues N86, N169, and N242 (Figure 2A). As eukaryotic proteins recombinantly expressed in E. coli are usually not glycosylated, we first tested the latter hypothesis. We used PNGase F, a glycopeptidase that selectively removes glycans directly at the N-linkage by cleaving the glycosidic bond between asparagine and the core GlcNAc. With the deglycosylated Vn, the binding assays were repeated to see whether binding could be abrogated by removal of the N-linked glycosylations. In Figure 2B, the fluorescence-based whole-cell assay using E. coli AS75 expressing either full-length YadA_O:8 or full-length YadA_O:9 is shown. No binding was observed with cells expressing YadA_O:8, which fits the hypothesis, as the postulated Vn binding stretch is not present in YadA from Y. enterocolitica serotype O:8. In the case of binding of bacteria expressing full-length YadA_O:9, a clear difference in binding to Vn_plasma was observed between the glycosylated Vn_plasma and deglycosylated Vn_plasma (Figure 2B). For Vn_HEK, no change in binding of YadA_O:9-expressing E. coli AS75 before and after glycosylation was observed. We can at this point not say as to why no change was observed for bacterial binding of Vn_HEK compared to deglycosylated Vn_HEK. Vn_Ec was bound in neither the glycosylated nor the deglycosylated state. While this supported our hypothesis that the glycan residues of Vn might be involved in the YadA_O:9–Vn interaction rather than the proteinaceous part of Vn, we also repeated the binding assay using purified YadA head domains from both serotypes of Y. enterocolitica (Figure 2C). While, as expected, YadA_O:8 did not bind to Vn-coated plates, neither to the untreated nor to the deglycosylated version, YadA_O:9 bound to both untreated Vn_plasma and untreated Vn_HEK (Figure 2C). Untreated Vn_Ec was not bound as already shown in Figure 1C. After deglycosylation with PNGase F, neither Vn_plasma nor Vn_HEK was bound by YadA_O:9 anymore, further supporting our hypothesis of YadA_O:9 interacting with the N-linked glycans.

Heparin was described to abrogate the interaction between Vn and YadA_O:9 (Mühlenkamp et al., 2017). We next wanted to investigate whether the potential YadA_O:9 glycan interaction might be the cause for this observation. It was hypothesized before that heparin blocks the YadA binding site on Vn. As in the globular state, the heparin binding site in Vn is mostly hidden inside the core of the protein; this seemed unlikely to be the reason for YadA_O:9 binding inhibition (Hayashi et al., 1985; Izumi et al., 1989; Zhuang et al., 1996; Leavesley et al., 2013). Coverslips were coated with untreated Vn_plasma, Vn_HEK, or Vn_Ec. YadA_O:9 (full length)-expressing, fluorescent bacteria were checked for binding (Figure 3, left column). To check for the influence of heparin on this interaction, we also prepared samples where we either preincubated Vn with heparin (Figure 3, middle column) or preincubated YadA_O:9-expressing bacteria with heparin (Figure 3, right column). In the fluorescence microscopy adhesion assay, we observed only minimal adhesion of bacteria to Vn_Ec (Figure 3, bottom row). When coverslips had been coated with Vn_plasma or Vn_HEK, adhesion was observed only in the absence of heparin. In cases where Vn was preincubated with heparin, bacteria expressing YadA_O:9 adhered to Vn to a comparable level as in the untreated samples (Figure 3, left and middle columns). When YadA_O:9-expressing bacteria were preincubated with heparin, reduced binding to untreated Vn_plasma and Vn_HEK was observed (Figure 3, right column). Quantifications of the area of the particles reflect the tendencies seen in the experiment, where preincubation of the bacteria expressing YadA_O:9 with heparin seems to reduce binding to Vn whereas preincubation of Vn with heparin did not change the adhesion of YadA_O:9-expressing bacteria. This observation further strengthened our hypothesis that the YadA Vn-binding loop aids in adhesion of YadA_O:9 to glycan moieties.

YadA, as an adhesin, is involved in autoaggregation, which has been described as an important mechanism for immune evasion during infection as well as for biofilm formation (Trunk et al., 2018). We have observed earlier that the interaction with other adhesin targets, such as ECM molecules, interferes with autoaggregation (manuscript in preparation). We thus wanted to investigate what effect heparin might have on autoaggregation mediated by YadA_O:9. We expressed YadA_O:8 or YadA_O:9 full length in fluorescent E. coli AS75 and allowed for autoaggregation of these samples. Uninduced samples served as a control. Half of the samples were then preincubated with heparin-disaccharide. The uninduced samples did not show any autoaggregation behavior, either in the presence or in the absence of heparin (Figure 4, rows 1 and 3). The induced YadA_O:8 samples autoaggregated to similar degrees both in the absence and in the presence of heparin (Figure 4, row 2). Fluorescent bacteria expressing YadA_O:9 showed autoaggregation in the absence of heparin but reduced autoaggregation in the presence of heparin (Figure 4, lower row). This indicates that, indeed, heparin binding of YadA_O:9 dissolves the autoaggregation tendencies caused by surface expression of YadA_O:9. The dispersion of particles sizes (μm²) is shown in Figure 4B. The scatter plots reflect the difference in aggregate (particle) sizes. A reduction in area between aggregates of E. coli expressing YadA_O:9 with and without addition of heparin-disaccharide can be seen shifting from large aggregates to smaller aggregates or fully disaggregated samples.

To test for a direct interaction between YadA_O:9 and heparin, we used E. coli AS75 cells expressing either YadA_O:8 or YadA_O:9. The bacteria were immobilized in a 96-well plate to capture biotinylated heparin. Bound biotinylated heparin was detected using Strep-Tactin–HRP. While no binding of biotinylated heparin to YadA_O:8-expressing bacteria was observed, bacteria expressing YadA_O:9 clearly showed heparin binding (Figure 5A). To support these results, we used electrochemical impedance measurements to measure bacterial binding to a heparin-coated surface. Biotinylated heparin was coated onto a biosensor using matrix-embedded NeutrAvidin (Ahmed et al., 2013). The change in impedance was then measured upon binding of E. coli AS75 expressing either YadA_O:8 or YadA_O:9 (Figure 5B). Please note that negative binding values are due to stronger adhesion of E. coli expressing YadA_O:8/O:9 to uncoated electrodes that were used as a background and subtracted. While, for uninduced E. coli AS75 and E. coli expressing YadA_O:8, no change in impedance was observed, we could clearly measure binding of E. coli expressing YadA_O:9 by a significant change of impedance (Figure 5B). We then aimed to test for binding of heparin to purified YadA head domains. A dot blot using immobilized YadA_O:8 and YadA_O:9 head domains to detect binding of biotinylated heparin was performed. While no heparin binding was observed for either the buffer control or the YadA_O:8 head domain, a signal could be observed for the binding of biotinylated heparin to the immobilized YadA_O:9 head domain (Figure 5C). To quantify the binding, we immobilized the head domains of YadA_O:8 and YadA_O:9 in a 96-well plate and tested for binding at various concentrations. We observed that at 450 μg/ml of heparin, binding between YadA_O:8 or YadA_O:9 head domains and heparin is significantly different (Figures 5D,E). Repeating the assay with a dilution series of biotinylated heparin allowed us to investigate the concentration dependency of the binding. Using a fifth-party logistics fit, we estimate the (apparent) K_D to be approximately 30 nM. Furthermore, this experiment allows for an estimation of the binding ratio between YadA_O:9 and biotinylated heparin. The binding ratio is estimated to be 1:1 (YadA_O:9 monomer to biotinylated heparin). We can at this point not claim an accurate K_D or binding ratio as heparin varies in length but averages at 15 kDa (Shriver et al., 2012).