Building on our recent structural and functional findings about the C3d^17C dimer (14), we set out to determine the structure of a complex between dimeric C3d^17C and Sbi-IV. The C3d^17C:Sbi-IV dimer complex ( Figure 1 , data collection and refinement statistics can be found in Supplementary Tables S1, S2 ) shows two molecules of Sbi-IV binding to the opposing concave binding surfaces on the C3d^17C dimer ( Figure 1A ) via interactions identical to those observed in the previously reported C3d^17A:Sbi-IV structure (16). An additional binding mode observed in the C3d^17A:Sbi-IV structure, with Sbi-IV interacting with the convex surface of C3d close to the thioester region ( Supplementary Figure S2 ), is absent in the C3d^17C:Sbi-IV dimer complex, prevented by the tight dimer interactions. Intriguingly, and in clear contrast to the unliganded C3d^17C dimer, the Sbi-IV bound C3d^17C dimer has undergone a 3D domain swap, whereby both the N-terminal α1 helices have been reciprocally exchanged between the C3d^17C molecules within the dimer ( Figure 1A ). As detailed in Figure 1B , the helix α1 region M1 – G16 (residue numbering based on the crystal structure), located N-terminally from C17, has undergone a full polypeptide chain exchange in both C3d^17C molecules, recreating identical interactions with the opposing C3d recipient without causing appreciable differences in relative orientation when compared to both monomers in the unliganded C3d^17C dimer (PDB accession code 6RMT; main-chain RMSDs: 0.55 Å, residues 20-290 in Chain A; 0.15 Å, residues 20-290 in chain B). 2Fo-Fc electron density maps for the critical loop regions of the swapped α1 helices for each chain are shown in Supplementary Figure S1 ). This unique structural 3D domain swap phenomenon is highlighted in Figure 1C , showing a comparison between the unliganded C3d^17C dimer and the C3d^17C:Sbi-IV dimer complex. Superpositioning of the A chains in the unliganded C3d^17C dimer and the C3d^17C:Sbi-IV dimer complex reveals a slight but significant change in overall relative orientation of both monomers in the C3d^17C:Sbi-IV dimer and highlights the resulting change in conformation of the dimer disulfide bond as a result of the N-terminal helix swap ( Figure 1D ).

In addition to stabilising the C17-C17 interchain disulfide, the helix α1 swap causes a substantial increase (640%) in joint buried surface area from 698 Ų in the unliganded C3d^17C dimer to 4474 Ų in the Sbi-IV-bound helix-swapped C3d^17C dimer. This expansion in interface area is supported by 13 additional hydrogen bonds and 4 ionic interactions which are absent in the unliganded dimer ( Supplementary Figure S3 ).

To further investigate the potential role of the interaction between Sbi-IV and helix α1 on the concave face of C3d ( Supplementary Figure S2 ) we employed near-UV (320-250 nm) circular dichroism (CD) spectroscopic analyses of C3d^17C and C3d^17A collected as a function of temperature, where the absorption of aromatic side chains and disulfide bonds, absent in Sbi-IV, provides information about changes in C3d’s tertiary structure (28) and binding energy of complex formation with Sbi-IV ( Figure 2 ). After analysis of the sigmoidal fits of thermal denaturation curves, Sbi-IV was found to increase the melting temperature (ΔT_m) of C3d^17C by 7.4°C from 51.8 ± 0.1°C in the absence of Sbi-IV to 59.2 ± 1.5°C in the presence of Sbi-IV ( Figure 2A ; averaged ΔT_m at 280 and 296 nm), reflecting the binding energy of the complex. In comparison, the presence of Sbi-IV raises the ΔT_m of C3d^17A by 9°C, indicating a slightly stronger interaction energy for the binding between Sbi-IV and monomeric C3d^17A ( Figure 2B ; averaged DT_m at 280 and 296 nm).

Interestingly, whilst monomeric C3d^17A is thermally stable, it becomes unstable in the presence of wild-type Sbi-IV and unlike dimeric C3d^17C undergoes precipitation at temperatures above 65°C ( Figure 2C ). Intriguingly, the precipitation of C3d^17A does not occur in the presence of an N-terminally truncated Sbi-IV mutant (V198-D208del; residue numbering based on UniProt: Q2FVK5) lacking key residues that were previously shown to interact with helix α1 and its linker on the convex surface of C3d by crystal structure and NMR analyses of the C3d^17A:Sbi-IV complex (16) ( Supplementary Figure S2 ). Whilst the ~7°C increase in T_m ( Figure 2D ) indicates that Sbi-IV^V198-D208del is still able to bind C3d^17A, the lack of a temperature-induced precipitation in the presence of Sbi-IV^V198-D208del, suggests a potential role for the convex C3d binding mode of Sbi-IV in the displacement of the N-terminal α1 helix of C3d, whereby binding of Sbi-IV destabilises the helix, leading to aggregation and precipitation

We further explored the potential role of Sbi-IV’s convex C3d binding mode in the helix swap observed in the C3d^17C:Sbi-IV dimer complex by examining single amino acid substitutions within Sbi-IV, which were chosen based on the Sbi-IV residues identified in the convex C3d binding mode of Sbi-IV, observed in the crystal structure of the C3d^17A:Sbi-IV complex and confirmed by subsequent NMR analyses (16) ( Supplementary Figure S2 ). The Sbi-IV mutants were evaluated using near-UV CD analyses in the presence of monomeric C3d^17A. Similar to the N-terminally truncated Sbi-IV mutant (Sbi-IV^V198-D208del), alanine substitutions at positions S199, I204 and S219 in Sbi-IV exhibited no observable precipitation or aggregation ( Figures 3A-D ), suggesting that these residues could be important for Sbi-IV’s convex C3d binding mode. E201A and K212A show a reduction of the levels of precipitation ( Figures 3E, F ), indicating that these residues are important for convex surface binding but likely enhance the binding stability once the protein is initially in position. In contrast, substitutions at D208A and D243A exhibited comparable levels of precipitation and aggregation to wild-type Sbi-IV ( Figures 3G, H ) and therefore presumably have less significant roles in the convex surface binding mode.

To investigate whether the helix swap observed in C3d^17C dimers is unique to Sbi-IV, we also included Sbi-III-IV in our analyses. Because the aromatic amino acids present in Sbi-III-IV make this construct unsuitable for the near-UV CD temperature melt analyses we used for Sbi-IV, we developed a bespoke tryptophan fluorescence spectroscopy method that exploits the fluorescence quenching effect on tryptophan brought about by histidine residues positioned in close proximity (29). Several C3d^17C mutants were generated to either substitute a natural tryptophan (W41, located on a loop between helices α2 and α3) with an alanine or introduce a histidine at position 4 close to W41. As detailed in the Figure 4A schematic, solutions of C3d^A4H or C3d^W41A are expected to form homodimers and show no change of tryptophan fluorescence in the event of an N-terminal helix swap, whilst mixed solutions of C3d^A4H with C3d^W41A can form heterodimers that should display an increase in tryptophan fluorescence as H4 no longer quenches W41 when the helix swap occurs. As expected, solutions of either C3d^A4H or C3d^W41A containing homodimers showed slight changes in fluorescence in the presence of Sbi-IV^V198-D208del, Sbi-IV or Sbi-III-IV ( Supplementary Figure S4 ). However, adding these Sbi constructs to mixed solutions of C3d^A4H and C3d^W41A, expected to contain approximately 50% heterodimers of the two mutants, showed various changes in fluorescence. In the presence of Sbi-IV or Sbi-III-IV ( Figures 4B, C ) a clear increase in tryptophan fluorescence (27% and 17%, respectively) was observed when compared to their respective homodimers. These results indicate that W41 in C3d^A4H is no longer being quenched by its structural neighbour H4, which is suggestive of the occurrence of a Sbi-induced helix swap event. This is further supported by the reduced change in fluorescence intensity (2.5%) of the C3d^A4H-C3d^W41A heterodimer observed in the presence of Sbi^V198-D208del ( Figures 4B, C ), emphasising that W41 remains quenched by H4 as the α1 helices cannot exchange when Sbi is unable to interact with the convex surface of C3d.

The DNA sequences of human C3d (residues 1-310) with a cysteine to alanine substitution at position 17 (C3d^17A) was previously cloned into the pET15b expression plasmid (33). To reproduce the wild-type sequence, the alanine at position 17 of the C3d^17A construct was reverted to a cysteine (C3d^17C) using site-directed mutagenesis (14). Single amino acid mutants of C3d^17C with an alanine to histidine substitution at position 4(C3d)/997(pre-pro C3) (C3d^A4H) or with a tryptophan to alanine substitution at position 41(C3d)/1034(pre-pro C3) (C3d^W41A) were generated through site directed mutagenesis following the protocol detailed in the QuikChange II site-directed mutagenesis kit (Agilent) using Pfu polymerase (Promega) and the primers detailed in Supplementary Table S3 .

All C3d constructs were expressed in the Escherichia coli Shuffle T7 cells (NEB) and purified using cation exchange chromatography (column: 1 mL HiTrap SP Sepharose HP [Cytivia], binding buffer: 50 mM MES, pH 5.5, elution buffer: 50 mM MES, 500 mM NaCl, pH 5.5) followed by size exclusion chromatography (column: HiLoad 16/600 Superdex 200 prep grade [GE Healthcare], buffer: 20 mM Tris, 150 mM NaCl, pH 7.4).

The DNA sequences of the Sbi constructs composed of domain IV (150-266 of full-length Sbi) and domains III and IV (198-266 of full-length Sbi), both bearing N-terminal 6x His-tags were previously cloned into the pQE-30 Xa plasmid [Sbi-IV (14); Sbi-III-IV (16)]. The DNA sequence of a truncated Sbi-IV construct lacking 11 residues located at the N-terminus of helix α1 (198-208 of full-length Sbi) (Sbi-IV^V198-D208del) was produced using PCR with designed primers ( Supplementary Table S3 ) and cloned into the pET-28a plasmid. Various single amino acid mutants of Sbi-IV where residues at positions 199, 201, 204, 208, 212, 219 and 243 were substituted to alanines were produced through site-directed mutagenesis as described above for C3d using primers detailed in Supplementary Table S3 .

All Sbi constructs were expressed in E. coli BL21(DE3) (Sigma Aldrich) cells and purified using nickel-affinity (column: 1mL HisTrap FF [Cytivia], binding buffer: 50 mM Tris, 15 mM NaCl, 20 mM Imidazole, pH 7.4, elution buffer: 50 mM Tris, 150 mM NaCl, 500 mM Imidazole, pH 7.4) and size exclusion chromatography (details as described above for C3d).

Crystallisation was performed at 18°C using the hanging drop vapour diffusion method. For the C3d^17C dimer-Sbi-IV complex structure, needle-like co-crystals at a 1:1 molar (288 µM) ratio of Sbi-IV (2.66 mg ml^-1) and C3d^17C (10 mg ml^-1) appeared in the condition containing 0.2 M Sodium citrate tribasic dihydrate, 20% PEG 3350 (PACT premier condition E11, Molecular Dimensions) (measured pH: 8.07).

Crystals were mounted on loops, flash-frozen in liquid nitrogen and X-ray diffraction data collected on the I04 beamline Dectris PILATUS 6M pixel detector at the Diamond Light Source synchrotron (Oxfordshire, UK) (See Supplementary Table S1 for data collection statistics). Integration of diffraction images and data reduction were performed using Xia2-DIALS and AIMLESS respectively (C3d^17C dimer-Sbi-IV complex structure: resolution: 2.4 Å, images 1601-3600 excluded). The automated BALBES pipeline and COOT were used for molecular replacement and model building. Refinement was carried out in REFMAC and Phenix (refinement statistics can be found in Supplementary Table S2 ) and UCSF Chimera was used for superpositioning and generation of images. Bond numbers and buried surface area were calculated with PDBePISA. The structure has been submitted to the Protein Data Bank with PDB submission code 6RMU (C3d^17C dimer-Sbi-IV complex).

For tertiary structure thermal melt analyses, proteins were buffer exchanged into 10mM sodium phosphate pH 7.4 using Zeba™ Spin desalting columns (ThermoFisher) according to the manufacturer’s instructions. Buffer-corrected CD measurements of 87.5 µM C3d^17C or C3d^17A alone or in the presence of an equimolar concentration of wild-type Sbi-IV, Sbi-IV^V198-D208del and Sbi-IV mutants were acquired at wavelengths between 250 and 320 nm in 1 nm increments, and at temperatures ranging from 5°C to 85°C in 5°C increments followed by a final reading at 20°C. All measurements were obtained on a Chirascan spectrometer (Applied Photophysics) with a bandwidth of 2 nm and 2 second time per point. Millidegree values were converted to units of mean residue ellipticity, and deconvolution of secondary structural data was performed using DichroWeb. Melting temperature values were estimated using sigmoidal fits of thermal denaturation curves at different wavelengths performed using the Boltzmann function in GraphPad Prism (version 9.3.1). See Figures 2A, B respectively for C3d^17C and C3d^17A sigmoidal fits of thermal denaturation curves used for T_m calculations.

The C3d^A4H and C3d^W41A proteins were initially buffer exchanged into 10 mM sodium phosphate pH 7.4 as detailed above. Prior to fluorometric analysis, the two mutant proteins at a concentration of 1 mg mL^-1 were mixed together and treated with tris (2-carboxyethyl) phosphine (TCEP) (Merck) in TNE buffer (1 M Tris, 150 mM NaCl, 5 mM EDTA, pH 7.5) for 1 hour at 4°C, to partially reduce the proteins and cause any homodimers to dissociate. TCEP was subsequently removed allowing the formation of heterodimers by dialysis overnight at 4°C into 10 mM sodium phosphate pH 7.4 using Slide-A-Lyzer™ dialysis cassettes (ThermoFisher).

Following dialysis the C3d^A4H/C3d^W41A protein solution was concentrated and used for fluorometry experiments. Samples containing, 100 µM of the C3d^A4H/C3d^W41A solution alone or in the presence of an equimolar concentration of Sbi-IV, Sbi-III-IV and Sbi-IV^V198-D208del were excited at the 295 nm wavelength and emissions between the 300 and 450 nm wavelengths were recorded. All measurements were obtained on an LS50B Luminescence Spectrometer (PerkinElmer) at room temperature using 6nm slits and a 250 ms scan speed. For analysis, the emission spectra were visualised in GraphPad Prism (version 9.3.1) as XY scatter representations. The Sbi construct intensity change was calculated for individual emission spectra by comparing the intensity reading for the C3d^A4H/C3d^W41A peak alone with the Sbi intensity reading at the corresponding wavelength. These intensity changes were calculated as percentages and visualised as column representations that displayed the individual points, as well as the mean and standard deviation.