Ucl fimbriae regulation and glycan receptor specificity contribute to gut colonisation by extra-intestinal pathogenic Escherichia coli

Extra-intestinal pathogenic Escherichia coli (ExPEC) belong to a critical priority group of antibiotic resistant pathogens. ExPEC establish gut reservoirs that seed infection of the urinary tract and bloodstream, but the mechanisms of gut colonisation remain to be properly understood. Ucl fimbriae are attachment organelles that facilitate ExPEC adherence. Here, we investigated cellular receptors for Ucl fimbriae and Ucl expression to define molecular mechanisms of Ucl-mediated ExPEC colonisation of the gut. We demonstrate differential expression of Ucl fimbriae in ExPEC sequence types associated with disseminated infection. Genome editing of strains from two common sequence types, F11 (ST127) and UTI89 (ST95), identified a single nucleotide polymorphism in the ucl promoter that changes fimbriae expression via activation by the global stress-response regulator OxyR, leading to altered gut colonisation. Structure-function analysis of the Ucl fimbriae tip-adhesin (UclD) identified high-affinity glycan receptor targets, with highest affinity for sialyllacto-N-fucopentose VI, a structure likely to be expressed on the gut epithelium. Comparison of the UclD adhesin to the homologous UcaD tip-adhesin from Proteus mirabilis revealed that although they possess a similar tertiary structure, apart from lacto-N-fucopentose VI that bound to both adhesins at low-micromolar affinity, they recognize different fucose- and glucose-containing oligosaccharides. Competitive surface plasmon resonance analysis together with co-structural investigation of UcaD in complex with monosaccharides revealed a broad-specificity glycan binding pocket shared between UcaD and UclD that could accommodate these interactions. Overall, our study describes a mechanism of adaptation that augments establishment of an ExPEC gut reservoir to seed disseminated infections, providing a pathway for the development of targeted anti-adhesion therapeutics.

SPR analysis. SPR analysis was carried out using a Biacore T200 system (Cytivia) as previously 125 described [15] with minor modifications. Briefly, UclD and UcaD were immobilized onto flow cells 126 of a CM5 sensor chip using amine coupling at 20 µg/mL in 10 mM sodium acetate pH 4.0 and at a 127 flow rate of 5 µL/min for 10 minutes. Glycans were chosen based on positive glycan array results for 128 each of the proteins with each glycan run across a dilution range starting at a maximum concentration 129 of 100 µM and minimum concentrations tested being 1.6 nM using single cycle kinetics. Results were 130 analysed using the Biacore T200 evaluation software. 131 Crystallization and crystal structure determination. 132 UclD LD : UclD LD crystals were produced using the hanging drop method, with drops containing 1 µL 133 of protein (10 mg/mL) and 1 µL of well solution (20-25% w/v PEG 3350, 0.1 M Bis-Tris propane 134 pH 6.5, 0.2 M sodium iodide). The crystals appeared within 1-5 days. The crystals were cryoprotected 135 in glycerol (80% well solution and 20% (v/v) glycerol) before flash-cooling in liquid nitrogen. X-ray 136 diffraction data were collected from a single crystal at the Australian Synchrotron MX1 beamline, 137 using a wavelength of 0.9537 Å. Data collection was performed using Blu-Ice software [17], indexed 138 and integrated using MOSFLM and scaled with AIMLESS within the CCP4 suite [18]. Molecular 139 replacement was initially attempted using several published fimbrial adhesin structures as templates, 140 but a solution could not be obtained. Because the crystallisation condition contained sodium iodide, 141 we determined the UclD LD structure by SAD phasing using the CRANK2 [19] pipeline of the CCP4 142 suite and a dataset (2.85 Å resolution) collected at wavelength of 1.3776 Å, where the anomalous 143 scattering properties of iodide are still significant. Iodine atoms were located by SHELXD [20], and 144 automatic model building was performed using Buccaneer [21] and Refmac [22]. The higher 145 resolution UclD LD structure was subsequently solved by molecular replacement using PHASER [23]. 146 The model was refined using Phenix [24], with iterative model building carried out between rounds 147 of refinement using Coot [25]. Structure validation was performed using MolProbity [26]. The 148 structure was refined to final Rwork/Rfree values of 24.8%/30.6% (Table A in S1 Text). The moderate 149 quality of the UclD LD dataset (Rmeas of 27.9%, Table A in S1 Text) is a likely reason for the high Rfree 150 value. The final UclD LD model contains residues 21-215. Electron density was not observed for 151 residues 43-48, suggesting that these regions have a disordered or flexible conformation in the 152 crystals. The coordinates and structure factors have been deposited in the PDB with ID 7MZP. 153 UcaD LD : UcaD LD crystals were produced using the hanging drop method with drops containing 1 µL 154 of protein (8-16 mg/mL) and 1 µL of well solution (0.1 M sodium citrate buffer pH 4.5-5.5, 2-3 M 155 NaCl). The crystals appeared within 3-5 days. The crystals were cryoprotected in Paratone-N, before 156 flash-cooling in liquid nitrogen. X-ray diffraction data were collected from a single crystal at the 157 Australian Synchrotron MX2 beamline, using a wavelength of 0.9537 Å. Data collection was 158 performed using Blu-Ice software [17], indexed and integrated using MOSFLM and scaled with 159 AIMLESS within the CCP4 suite [18]. The structure was solved by molecular replacement using 160 PHASER [23] and UclD LD as the template. The model was refined using Phenix [24] and structure 161 validation was performed using MolProbity [26]. The structure was refined to final Rwork/Rfree values 162 of 16.9 %/19.2 %, respectively ( The docked structure with the best superimposition between the fucose moiety of lacto-N-182 fucopentose VI and the fucose molecule observed in the UcaD LD :Fuc complex was then subjected to 183 further optimization by a 40 ns MD simulation using the AMBER force-field implemented in the 184 YASARA software suite [29]. A representative energy-minimised snapshot from the MD trajectory 185 was used for the analyses. Galβ1-4GlcNAcβ1-3Galβ1-4(Fucα1-3)Glc 8G 4.877±0.33 3.611±0.49

395
Red indicates binding as shown in Fig 5. Binding indicates a value of greater than 1-fold of average 396 background plus 3 standard deviations as described in the MIRAGE table (Table F)

Method of preparation:
The preparation of UclD and UcaD are explained in the Materials and Methods section.

Sample modifications
UclD LD and UcaD LD are His-Tagged proteins.
Assay protocol See Materials and Methods.

Glycan Library
Glycan description for defined glycans Glycans in this study are listed in Table E  Dispensing mechanism Contact printing using 946NS6 pins with a 6 pin in a 3 columns x 2 rows configuration.
Glycan deposition Approximately 1.8 nl per spot is printed according to manufactures guidelines.
Printing conditions Array were printed with dehumidification at a maximum humidity of 60% relative humidity (Standard laboratory starting humidity of 75-90%) at 22ºC.
Glycans were left to react with the slide for at least 8 hours after the print was completed.

5. Glycan Microarray with "Map"
Array layout The array consists of a single array of glycans split between 6 pins (3 columns x 2 rows) with 4500μm row and column spacing. Each pin printed a 20 columns x 16 rows with 200μm spot spacing (centre to centre) with a minimum spot size of 100μm. Each sample is printed in quadruplicate with each of the 6 print areas including at least three negative control samples (print solution only) and two positive control samples consisting of one sample of fluoroscienamine and one sample of a mixture of rabbit anti-mouse antibody labelled with Alexa 555 and Alexa 647. Positive controls provide proof of successful immobilization of the amine reagents and provides for orientation for analysis. The antibodies also can provide controls for secondary antibodies used in experiments (if applicable).
Glycan identification and quality control Data processing Data was exported as a CSV file and exported to Microsoft Excel.

Glycan Microarray Data Presentation
Data presentation Binding data is presented in Fig 5 together with SPR data. The yes/no binding including glycan identification is shown in Table E in S1 Text.

Data interpretation
We only use glycan arrays as a yes/no binding tool. Due to this we look only at binding that is unambiguously above background vs lack of binding above background. Average background + 3x standard deviation of the background of 20 sets of 4 spots of DMF:DMSO only spots is applied to determine if binding observed is significantly above background. Only spots with values equal to or greater than this value were considered as binding from data of any tested slide. These values are slide dependent.

Conclusions
UcaD LD and UclD LD both recognize glycans but even though they are highly similar the glycans recognized are different.