Structural properties of a haemophore facilitate targeted elimination of the pathogen Porphyromonas gingivalis

Porphyromonas gingivalis is a keystone bacterial pathogen of chronic periodontitis. P. gingivalis is unable to synthesise the porphyrin macrocycle and relies on exogenous porphyrin, including haem or haem biosynthesis intermediates from host sources. We show that under the iron-limited conditions prevailing in tissue environments, P. gingivalis expresses a haemophore-like protein, HusA, to mediate the uptake of essential porphyrin and support pathogen survival within epithelial cells. The structure of HusA, together with titration studies, mutagenesis and in silico docking, show that haem binds in a hydrophobic groove on the α-helical structure without the typical iron coordination seen in other haemophores. This mode of interaction allows HusA to bind to a variety of abiotic and metal-free porphyrins with higher affinities than to haem. We exploit this unusual porphyrin-binding activity of HusA to target a prototypic deuteroporphyrin-metronidazole conjugate with restricted antimicrobial specificity in a Trojan horse strategy that effectively kills intracellular P. gingivalis.

invading the gingival epithelial cell, P. gingivalis is exposed to an intracellular environment in which free iron and porphyrin are restricted. HusA would be expressed at high levels to hijack haem and certain intermediates of the de novo haem synthesis pathway. The expressed HusA would then offer an avenue for the deuteroporphyrin IX linked metronidazole (P8) to be captured and uptaken by P.
gingivalis, thereby achieving a selective killing effect against P. gingivalis inside the gingival epithelial cell.

Protein purification for HusA and mutants
His-tagged HusA/mutant proteins were purified through Ni-NTA resin and the His-tag was cleaved were obtained from chemical shift information using TALOS+ 21 25 ; the given errors were propagated from uncertainties in peak height measurements, which were calculated as the r.m.s. noise in each processed spectrum.

Correction of inner filter effect and curve fitting in tryptophan fluorescence quenching assays
The concentration of NATA was chosen such that the raw fluorescence intensity was similar to the 10 uM concentrations of HusA and mutants used. The same molarity titrants of porphyrin were titrated into NATA as were added into HusA in the control reactions. Fluorescence intensity was scaled with respect to the starting florescence and corrected for the inner filter effect. Fluorescence at 335 nm was used to fit binding curves.
The correction method is based on Fonin AV., et al. 26 , noting that a Cary Eclipse spectrofluorimeter was used.
Control titrations of porphyrins into 6 µM NATA, a fluorophore that does not bind haem, were  Overall the small ring distortions in haem, PPIX and DPIX ligands did not prevent these ligands adopting the same (or similar) sets of poses and hence the results for all ring conformers over each ligand were pooled in the analysis. The non-planar coproporphyrinogen III and uroporphyrinogen III docked against the rigid receptor with lower energies (-8.4±0.4 kcal/mol and -7.1±0.5 kcal/mol, respectively) compared to haem, PPIX and DPIX (~13 kcal/mol), and were not analysed further.

Small angle X-ray scattering SAXS
Supplementary Table 2 summarises the SAXS data acquisition, analysis and results along with details of the software used as per recommended guidelines 29 . Briefly, data were reduced to I(q) versus q using the software scatterBrain. The amplitude of the momentum transfer is I(q) = 4πsinθ/λ where 2θ is the angle between the incident and scattered X-rays and λ the wavelength. Purified apo-HusA samples were prepared by gel filtration chromatography using a Superdex-200 HiLoad 16/60 column coupled to an ӒKTA purifier system in standard buffer (50 mM Tris, 150 mM NaCl, pH 8). The HusA protein sequence includes five Cys residues, and therefore, apo-HusA samples were prepared by dialysis in standard and reduced buffer (standard buffer with 2 mM TCEP) at room temperature.
For both samples, solvent scattering blanks were recorded on the end-point dialysates. The HusA:haem complex sample was prepared by adding freshly prepared haemin at a molar ratio of 1:1 to apo-HusA after dialysis in standard buffer. Given the K d estimate of 7.3 µM for the HusA:haemin complex, it follows that, at ~140 µM HusA and haemin (i.e., 3 mg/mL of HusA), there should be minimal free haem and thus the standard buffer dialysate served as the solvent blank. Data were analysed using the ATSAS program package 8 , with the specific programs used listed in Supplementary Table 2. Guinier parameters show no significant concentration dependence for apo-HusA recorded under reduced buffer, these data were therefore merged to improve statistics. The small concentration effects for data from samples in standard buffer were minimised by using the lowest concentration data for further analysis. Molecular weight estimates were calculated from Guinier I(0) values using the method of Orthaber et. al. 30 to yield estimates that are higher than expected for a purely monomeric sample (14−17% for HusA and 29% for HusA:haem). Making the most conservative assumption that all self-associations giving rise to this overestimate are dimeric, this result means that all samples are at least ~95% monomeric. P(r) transforms of each data set gave a bell-shaped profile consistent with the folded globular protein and a tail to high r values, likely arising from flexibility at the N and C-termini and the small amount of aggregate in the sample (Supplementary Fig 5D). Values for contrast and partial specific volumes were determined using the program MULCh 7 . Size-exclusion chromatography (SEC)-SAXS was not an option at the Australian Synchrotron for this system. SEC runs carried out in-house on a SEC-multi angle laser light scattering (MALLS) system with HusA:haem as well as HusA:haem analogues at a range of molar ratios and under several buffer conditions resulted in the haem/analogue being stripped from the complex during the sizing run. The haem/analogue was observed to have immobilised at the top of the SEC column and could not be removed with buffer. This was accompanied by increased column pressure and deteriorating column performance.

Reverse ferrochelatase activity assay
The reverse ferrochelatase activity assay was conducted to explore the ability of HusA removing iron from the haem molecule using the modified protocol described by Leob 31 . Briefly, the reaction mixture consisted of 50 mM Tris, pH 8, 0.5 mM DTT (Sigma-Aldrich), 2.5 µM haemin, 0.24 mM ferrozine (Sigma-Aldrich), 5.0 µM NADH (Sigma-Aldrich), 1% Tween 80 (Sigma-Aldrich), and approximately 0.15 mg protein in a volume of 0.83 mL. The protein concentration was determined by Bradford assay. The mixture was incubated at 37 o C for 90 min. All reagents were prepared freshly.
The mixtures were centrifuged at 8,000 × g for 5 min at the end of incubation. The supernatants were applied to measure the porphyrin fluorescence using SpectraMax i3 with excitation at 410 nm. The emission was recorded from 600 nm to 650 nm. The mixture without protein was used as blank.