Unveiling the versatility of the thioredoxin framework: Insights from the structural examination of Francisella tularensis DsbA1

Abstract

In bacteria the formation of disulphide bonds is facilitated by a family of enzymes known as the disulphide bond forming (Dsb) proteins, which, despite low sequence homology, belong to the thioredoxin (TRX) superfamily. Among these enzymes is the disulphide bond-forming protein A (DsbA); a periplasmic thiol oxidase responsible for catalysing the oxidative folding of numerous cell envelope and secreted proteins. Pathogenic bacteria often contain diverse Dsb proteins with distinct functionalities commonly associated with pathogenesis. Here we investigate FtDsbA1, a DsbA homologue from the Gram-negative bacterium Francisella tularensis. Our study shows that FtDsbA1 shares a conserved TRX-like fold bridged by an alpha helical bundle showcased by all DsbA-like proteins. However, FtDsbA1 displays a highly unique variation on this structure, containing an extended and flexible N-terminus and secondary structural elements inserted within the core of the TRX fold itself, which together twist the overall DsbA-like architecture. Additionally, FtDsbA1 exhibits variations to the well conserved active site with an unusual dipeptide in the catalytic CXXC redox centre (CGKC), and a trans configuration for the conserved cis-proline loop, known for governing DsbA-substrate interactions. FtDsbA1's redox properties are comparable to other DsbA enzymes, however, consistent with its atypical structure, functional analysis reveals FtDsbA1 has a high degree of substrate specificity suggesting a specialised role within F. tularensis' oxidative folding pathway. Overall, this work underscores the remarkable malleability of the TRX catalytic core, a ubiquitous and ancestral protein fold. This not only contributes to broadening the structural and functional diversity seen within proteins utilising this core fold but will also enhance the accuracy of AI-driven protein structural prediction tools.

Keywords: AlphaFold; Crystal structure; Disulphide bond: Oxidoreductase; Redox biology; Thioredoxin protein.

Graphical abstract

Fig. 1

Comparison of FtDsbA1 to EcDsbA. (A.) Domain organisation of model TRX-like proteins compared to predicted and experimental domain organisation of FtDsbA1. EcTRX consists of a canonical thioredoxin motif containing the βαβ and ββα domains linked by an 18-residue α helical motif. Monomeric Dsb oxidases like EcDsbA contain a long ∼90-residue α helical insertion linking the βαβ and ββα elements, while oligomeric Dsb isomerases like EcDsbC have a N-terminal oligomerisation domain preceding the TRX fold and contain a ∼59-residue alpha helical insertion between the βαβ and ββα elements. FtDsbA1 was predicted to contain both an N- and C- terminal extension. The experimental domain organisation of FtDsbA1 however includes a 45-residue N-terminal extension encompassing two alpha helices flanked by loops, and three α helical insertions into the core of the TRX fold. (B.) Topology schematics of the FtDsbA1 (left) and EcDsbA (PDB ID 1FVK (right) structures with cylinders representing α-helices and arrows representing β-sheets. Catalytic cysteines are shown as yellow circles. FtDsbA1 features a Class II β-sheet topology (1−3-2−4-5) while EcDsbA features a Class I β-sheet topology (3−2-4−5-1) . (C). Cartoon depictions of FtDsbA1. Secondary elements are labelled and coloured according to B. FtDsbA1 features an extended N terminus with two helices (α_a-b), two additional helices between β₄ and β₅ (α_c-d), and a third additional helix between β₅ and α₇ (α_e). (D.) Cut-outs of the additional helices α_a-e of FtDsbA1. (E.) Cartoon depictions of reference protein EcDsbA (PDB ID 1FVK [22]). Secondary elements are labelled and coloured according to B. (F.) Superimposition of FtDsbA1 (light blue) and EcDsbA (dark blue). The two proteins superimpose with a high RMSD value of 3.13 Å over 118 equivalent Cα atoms indicating significant structural differences. (G&H.) Electrostatic surface representation of FtDsbA1 (left) and EcDsbA (right). EcDsbA features an overall hydrophobic surface with a hydrophobic patch above the active site and a hydrophobic groove below. FtDsbA1 features an overall basic surface with a positively charged patch above the active site and a broad basic cleft enveloping the catalytic cysteines. The electrostatic potential was calculated with APBS in PyMOL showing positive charges in blue (saturating at 5 kT/e) and negative charges in red (saturating at −5 kT/e).

Fig. 2

Comparison of FtDsbA1 Structure and Active Site with Structural Homologues. (A.) Superimposition of FtDsbA1 and BcfH (PDB ID 7JVE [83]). BcfH (lavender) and FtDsbA1 (light cyan) superimpose with a RMSD of 1.8 Å across 161 equivalent Cα atoms indicative of broad structural similarities across the N terminus, α-helical insertion domain and additional TRX-like elements. (B.) Superimposition of FtDsbA1 and WpDsbA2 (PDB ID 6EEZ [88]). WpDsbA2 (dark cyan) and FtDsbA1 (light cyan) superimpose with a RMSD of 2.3 Å across 140 equivalent Cα atoms reflecting larger differences in their α-helical insertion domains and WpDsbA2’s lack of additional inserted TRX helices. (C&D.) Comparison of the (C.) β₄ β₅ loop and (D.) N-terminus in FtDsbA1 (light cyan), BcfH (lavender) and WpDsbA2 (dark cyan). Like other DsbA-like proteins, WpDsbA2 lacks any secondary structural elements in the TRX ββα motif, rather has a short loop to connect β₄ and β₅. In contrast, BcfH uniquely features a half helical turn in this region and FtDsbA1 features two helices. All three DsbA-like proteins feature an elongated and crooked N terminus which, in BcfH and WpDsbA2, facilitate their trimerisation. (E.) Cartoon representation of, from left to right, FtDsbA1, BcfH (PDB ID 7JVE [83]), WpDsbA2 (PDB ID 6EEZ [88]) and EcDsbA (PDB ID 1FVK [22]) with domains coloured as per Fig. 1. (F.) Comparison of the active sites of, from left to right, FtDsbA1, BcfH (PDB ID 7JVE [83]), WpDsbA2 (PDB ID 6EEZ [88]) and EcDsbA (PDB ID 1FVK [22]). FtDsbA1 and BcfH feature trans-proline loops which orient the carbonyl of the pro-1 residue (G₁₇₃ in FtDsbA1 and T₂₁₆ in BcfH) away from the catalytic cysteines. In comparison to EcDsbA and WpDsbA2 which harbour cis-proline loops where the Pro-1 residue carbonyl group (V₁₅₀ in EcDsbA and T₂₂₁ in α-WpDsbA2) is oriented towards the catalytic cysteines.

Fig. 3

Redox Characterisation of FtDsbA1. (A&B.) Overlay of AUC continuous standardised sedimentation coefficient (s20,w) distribution of (A.) oxidised and (B.) reduced FtDsbA1 at 2.5 (solid line), 1 (dashed line) and 0.25 (dotted line) mg/mL. Residuals from the c(s) distribution best fits plotted as a function of radial position from the axis of rotation. (C.) FtDsbA1 redox potential. FtDsbA1 was incubated with various concentrations of reduced glutathione (GSH) overnight and was analysed fluorometrically. The fraction of reduced protein was plotted against log([GSH]²/[GSSG]) to determine the Keq (1.85 ×10⁻⁴ M) and redox potential was then calculated (−129 mV). Data represents one biological replicate with three technical replicates that is representative of three independent replicates. (D.) Thermal melt of oxidised (open circles ○) and reduced (closed circles •) FtDsbA1. CD thermal melts are plotted as a fraction of α-helical content (based on molar ellipticity [θ] at 222 nm) and temperature (°C). Thermal melts represent three technical replicates.

Fig. 4

Disulphide Oxidase Activity of FtDsbA1. (A.) In vitro disulphide reductase activity. Increase in absorbance (650 nm) indicates the dissolution of insulin following disulphide cleavage. Neither FtDsbA1 (closed circles •) nor EcDsbA (open squares □) show disulphide reductase activity to insulin while EcDsbC (an isomerase) (closed squares ■) shows activity. A buffer control containing GSH only, showing no activity, was omitted for clarity. (B.) In vivo thiol oxidase activity. Fluorescence curves taken at 340 nm indicate oxidation of the peptide substrate ASST by EcDsbA (open squares □), FtDsbA1 (closed circles •) or buffer control (closed squares ■). FtDsbA1 shows little catalytic activity towards ASST and reacts comparably to the buffer control. (C.) Real time oxidation of the FtPilA peptide by EcDsbA and FtDsbA1. Change in fluorescence (λex 295 nm, λem 330 nm) indicates the reduction of DsbA enzymes (2 μM) resulting from the oxidation of the peptide substrate (2 μM). EcDsbA (light grey) reacts rapidly with the FtPilA peptide, becoming fully reduced within 4 s while FtDsbA1 (black) remains fully oxidised, indicating no activity towards the peptide. (D.) Initial rates of reaction of FtDsbA1 (closed circles •) and EcDsbA (open squares □) against the reductant DTT plotted to give rate constants. Both DsbA enzymes catalyse the oxidation of DTT with second order kinetics at comparable rates, indicating FtDsbA1 plays an oxidative role. (E.) Swimming motility assay. Optical density curves taken at 600 _nm indicate the migration of cells over time. Cells containing FtDsbA1 (closed circles •) did not migrate due to flagella mediated motility and displayed migration comparable to an empty vector control (omitted from graph for clarity) while EcDsbA transformed cells (open squares □) did display flagella mediated motility. (F.) Mucoidal morphology assay. The production of colanic acid results in morphological changes which can be seen as a haze surrounding cells. Cells expressing EcDsbC display this haze, indicating in vivo isomerase activity while both isolates of FtDsbA1 transformed cells (only one shown for clarity) show no difference in phenotype compared to an empty vector control (VC).

Fig. 5

Modifications within the TRX-fold lead to functional diversity.