Molecular mechanism of complement interference by Streptococcus pyogenes nuclease A

Abstract

Bacteria deploy virulence factors to subvert host immunity; yet the molecular details of these interactions often remain limited. Here, we reveal the structure and host interactome of the Streptococcus pyogenes nuclease A (SpnA). We characterize the structure and dynamics of SpnA using hydrogen-deuterium exchange mass spectrometry and single-particle electron cryo microscopy, yielding the first structural insights to SpnA. This allowed us to identify an additional oligonucleotide-binding domain, whose flexible structure may play an important function in nucleolytic activity. Affinity-pulldown mass spectrometry identified the complement system membrane attack complex (MAC) C5b67 components as key interactors. Cross-linking mass spectrometry combined with integrative modeling identified the binding interfaces between SpnA and C5b67. These interfaces are conserved among genetically diverse S. pyogenes strains. Interaction between SpnA and C5b67 is suggested to prevent the assembly of a functional MAC. Our findings uncover a novel function of SpnA in complement inhibition, and identifies new potential targets to prevent and treat S. pyogenes infections.

Structural and proteomic analyses reveal that Streptococcus pyogenes nuclease A binds complement C5b67 via conserved interfaces, blocking membrane attack complex assembly and uncovering a new mechanism of immune evasion.

Introduction

The complement system is a fundamental part of the innate immunity that rapidly identifies and neutralizes invading microbes. Although triggered through distinct pathways (classical, lectin, or alternative), all converge to amplify common responses. These include labeling pathogens for phagocytosis, triggering inflammation and assembling the membrane attack complex (MAC) to disrupt target membranes. The MAC forms through the sequential assembly of complement components C5b, C6, C7, C8 and polymeric C9. Following cleavage of C5 into C5a and C5b by C5-convertase, C5b initiates complex formation by binding C6 and C7, enabling membrane insertion. Subsequent recruitment of C8 and insertion of multiple C9, forms transmembrane pores that compromise membrane integrity and potentially lead to (pathogen) cell lysis. In the host, a soluble but inactive, fluid-phase version of the MAC (sMAC), appears in low amounts under healthy conditions. The sMAC contains complement proteins C5b6789 in complex with the chaperones clusterin and vitronectin¹; the function of the latter two is to prevent MAC insertion into host membranes. The effectiveness of the complement responses is shaped by pathogen specific features of the microbial surface such as capsule or cell wall-associated proteins². While Gram-negative bacteria are highly susceptible to MAC-mediated lysis, Gram-positive species are generally thought to be resistant due to their peptidoglycan cell walls and array of diverse immune evasion strategies, with complement thought to act primarily through opsonization. However, evidence suggests that this resistance may not be absolute, prompting a re-evaluation of the actual scope of mechanisms of complement system active against Gram-positive organisms^3,4.

Streptococcus pyogenes (Group A Streptococcus, GAS) is a human specific Gram-positive bacterium most often causing mild and self-resolving infections of the skin and oropharynx⁵. However, S. pyogenes can also cause severe infections with high mortality, including necrotizing soft tissue infections, bacteremia, and streptococcal toxic shock syndrome, as well as autoimmune sequalae, such as acute rheumatic fever and glomerulonephritis^6,7. Globally, S. pyogenes causes over 600,000 invasive infections each year: with estimated 160,000 deaths⁶. The pathogenesis of S. pyogenes infections has been extensively studied, and drivers of the broad spectrum of symptoms are due to secreted, surface-associated, and intracellular virulence factors. These directly and indirectly interact with both the innate and adaptive human immune systems^8,9. One such virulence determinant is the well-studied M protein, a highly variable surface molecule that plays a pivotal role in complement inhibition. The M protein binds to human factor H, a negative regulator of the alternative complement pathway, thereby reducing C3b deposition and subsequent opsonophagocytosis^10,11. Additionally, the hypervariable regions of M proteins and the streptococcal protein H also interact with C4b-binding protein (C4BP), further attenuating activation of the classical pathway^12–14. S. pyogenes secretes the streptococcal inhibitor of complement (SIC) and the complement evasion factor (CEF), which interfere with MAC formation and neutralizes the antimicrobial activity of extracellular histones^15–17. Beyond complement inhibition, other virulence factors contribute to immune evasion and tissue colonization. These factors range from adhesins (including the M protein family, fibronectin- and collagen-binding proteins, and pili)^13,18,19, cytolysins (streptolysins O and S: SLO and SLS)²⁰, and several immunoglobulin-degrading enzymes, such as IdeS (immunoglobulin G-degrading enzyme of Streptococcus pyogenes) and EndoS, the streptococcal cysteine proteinase B (SpeB)^21–24, to superantigens and deoxyribonucleases (DNAses)^9,25.

S. pyogenes expresses several extracellular DNAses, either chromosomally encoded or associated with prophages^26,27. These DNAses have described functions mainly in the degradation of host DNA associated with neutrophil extracellular traps (NETs) leading to immune evasion. Some have also been described to dampen the host immune responses by suppressing Toll-like receptor 9 (TLR9) mediated IFN-α and TNF-α production, hence decreasing macrophage bactericidal activity²⁸. Of the streptococcal nucleases described to date, the S. pyogenes nuclease A (SpnA) is the only cell-wall anchored DNAse²⁹. SpnA is a 99.8 kDa protein, with a C-terminal Mg²⁺-dependent endo/exonuclease domain (NUCL) involved in host NET degradation, and three N-terminal oligonucleotide-binding (OB) domains, two of which are required for substrate binding during catalysis³⁰. In addition to host DNA degradation, we have recently described a function of SpnA in re-binding the secreted, active IdeS to the surface of the bacteria promoting IgG-cleavage at the bacterial surface and thereby protecting the bacteria against phagocytic killing³¹. While this is thought to primarily occur in the oropharynx, host interactions beyond DNA-binding and degradation have not been fully elucidated to date. Previous studies using both blood bactericidal assays and mouse infection models have, however, indicated that SpnA has a central role in streptococcal pathogenesis, as a spnA-knockout strain displayed reduced virulence when compared to the wildtype strain²⁶.

Here, we have used quantitative and structural proteomics together with single-particle electron cryo microscopy (cryoEM) to determine the host plasma interactome and structure of SpnA. Our resolved structure of SpnA reveals a previously uncharacterized fourth OB domain that precedes the NUCL. We also demonstrate that SpnA interacts with human complement, specifically targeting the MAC C5b67 assembly intermediate, preventing the insertion of C5b67 into cell membranes. This protects against complement mediated lysis, providing an additional line of defense to the action of SIC and CEF^15,16. Besides expanding the structural knowledge on SpnA, our findings discover a previously unknown function through which SpnA targets the host immune system and advances our understanding of S. pyogenes pathogenesis.

Results

SpnA exists both as a monomer and as a dimer in solution

Chang et al.³⁰ have originally described a model of SpnA, consisting of three oligonucleotide-binding domains (OB1, OB2 and OB3) at the protein N-terminus followed by a C-terminal NUCL and a cell wall-anchor (CW) (Fig. 1A). The OB2 and OB3 domains have been demonstrated to be essential for (host) DNA-binding and -cleavage³⁰. The existing AlphaFold model of SpnA (AlphaFold ID: AF-Q9A0J7-F1) (Fig. 1B–D) depicts a plausible structure, with extended disordered regions both at the N-terminus preceding the OB-domains as well as between the OB2 and OB3 domains. We found further support for this flexibility in hydrogen-deuterium exchange mass spectrometry (HDX-MS) data (Fig. 1D, F–I). The HDX-MS data collected of SpnA alone (apo) (Fig. 1F, J) identified other regions susceptible to high deuterium exchange and large changes in the deuterium uptake over time, indicating dynamic and flexible regions. The AlphaFold prediction in combination with the HDX-MS data supports the presence of a previously unidentified domain between OB3 and NUCL (Fig. 1B–D). This suggested fourth OB domain, here termed OB4 (Fig. 1E), is also supported by both the DALI Protein Structure Comparison server³² and The Encyclopedia of Domains³³. The HDX data shows that the potential OB4 has high degree of flexibility or disorder, as compared to the other OBs (Fig. 1D).

Fig. 1. Domain organization and structural dynamics of SpnA.

A Domain organization of SpnA as described by Chang et al.³⁰. The different domains of SpnA are indicated as follows: signal sequence (SS, gray), oligonucleotide-binding domains (OB, shades of green), endo/exonuclease domain (NUCL, dark blue), cell wall attachment LPXTG motif (CW, pale blue). The amino acids residues numbers composing each domain are indicated under the schematic representation of the SpnA structure. B Updated domain organization of SpnA supported by AlphaFold modeling and HDX-MS data. C AlphaFold model of SpnA (AlphaFold ID: AF-Q9A0J7-F1). The AlphaFold per-residue confidence score (pLDDT, see color key) is indicated with higher confidence regions indicated in shades of blue and lower confidence regions indicated in shades of orange and yellow. D AlphaFold model of SpnA in complex with Ca²⁺ - and Mg²⁺ -ions colored according to deuterium uptake propensity at 9000 s. The HDX-MS data pinpoints to a high degree of flexibility as indicated by shades ranging from yellow to red. Regions in green and blue correspond to regions with lesser deuterium uptake propensity, hence indicating more protected regions. Regions for which no sequence coverage was obtained are indicated in gray. E Suggested functions for the proposed OB4 domain of SpnA based on the top 100 PDB hits as determined by the DALI Protein Structure Comparison server⁵⁵. Around 25% of the hits point to ssDNA-binding function as indicated by the blue regions. F HDX-MS deuterium uptake plot of SpnA alone (apo) and G in the presence Ca²⁺ - and Mg²⁺ - ions. The data is shown at time points 30, 300, 3000 and 9000 s. Selected residue positions are indicated. The deuterium uptake propensity (% Deuteration) is shown in (H). HDX-MS deuterium uptake difference plots (delta % deuteration) of I SpnA apo compared to SpnA in presence of Ca²⁺ -ions and J SpnA apo compared to SpnA in presence of Mg²⁺ -ions only. The deuterium uptake propensity (delta % deuteration) is shown in (K). Panels (A) and (B) were created in https://BioRender.com.

To gain more insight into the structural organization and flexibility of SpnA, we determined the structure of SpnA by single-particle cryoEM. For the cryoEM data collection, we used a recombinantly expressed full-length construct of SpnA spanning residues 25–876 (Fig. 1B, Supplementary Fig. 1A, B). Image classification revealed the particles as a mixture of monomers and dimers (Supplementary Fig. 1C, E, F). Reconstruction of particles representing the monomer yielded a 2.60 Å consensus map (Fig. 2, Supplementary Fig. 1D, E, Table 1). For the monomer, the resolved residues span 288–853 of SpnA consisting of OB3, OB4 and NUCL. The most populated dimeric assembly resulted in a 2.80 Å map and model (Supplementary Fig. 1 F-G, Supplementary Fig. 2A-B, Table 1). For the dimeric SpnA, the resolved residues span 290–852, the root mean square deviation (RMSD) between the AF-Q9A0J7-F1 model to chain A is 0.655 Å and to chain B 0.745 Å, as determined by rigid body alignment in pyMOL (Schrödinger, L. & DeLano, W., 2020. PyMOL, available at: http://www.pymol.org/pymol). The structure of the dimeric assembly is maintained by hydrogen bonds between residues at the interface, as determined by SpotOn analysis^34,35 (Supplementary Fig. 2A, C, Supplementary Table 1). The domains OB1-OB2 are separated from the OB3-OB4 domains by a 15-residue long loop (Fig. 1B, C) and are not resolved in our cryoEM models (Fig. 2, Supplementary Figs. 1, 2), likely due to a high flexibility of the entire N-terminal domain (residues 1–306) in relation to OB3, OB4 and NUCL.

Fig. 2. SpnA structure.

Monomeric structure of SpnA obtained by cryoEM. A top tilted view, B front view and C side view. The endo/exonuclease domain (NUCL) is indicated in dark blue, the oligonucleotide-binding domain (OB) 4 in dark green and OB3 in light green. D The suggested active site with the residues involved in catalysis³⁷ are indicated in pink. According to Chalmers et al.³⁷, R696 and N769 bind the nucleic acid substrate, H716, D767 and D810 are essential for catalysis, and E592 coordinate the Mg²⁺ -ion. Distances in the active site to the Mg²⁺ -ion are indicated. E The Ca²⁺ -ion (red) coordinating residues are indicated in green. Whereas the backbone of the loop is resolved in our structure in opposed in BP DNAse I (PDB: 1ATN), the sidechains of residues D686-N689 could not be modeled here. The threshold for the map (dark blue) is 0.35 in panels (D, E). In panels (D) and (E) the metal ions are modeled based on PDB: 1ATN.

Table 1.

CryoEM data collection, refinement and validation statistics

	SpnA monomer	SpnA dimer
	(PDB 9FS8)	(PDB 9FS9)
Data collection and processing
Magnification	215,000	215,000
Voltage (kV)	300	300
Electron exposure (e–/Å2)	60	60
Defocus range (μm)	1.0–2.2	1.0–2.2
Pixel size (Å)	0.57	0.57
Symmetry imposed	None	None
Initial particle images (no.)	9,810,516	9,810,516
Final particle images (no.)	134,130	126,766
Map resolution (Å)	2.60 (0.143)	2.80 (0.143)
FSC threshold
Refinement
Initial model used (PDB code)	AF-Q9A0J7-F1	AF-Q9A0J7-F1
Model composition Non-hydrogen atoms	4313	7559
Protein residues	562	1077
Ligands	None	None
R.m.s. deviations
Bond lengths (Å)	0.003	0.007
Bond angles (°)	0.531	1.108
Validation
MolProbity score	1.48	2.73
Clashscore	6.45	11.53
Poor rotamers (%)	1.07	7.76
Ramachandran plot
Favored (%)	97.50	92.22
Allowed (%)	2.50	6.52
Disallowed (%)	0	1.26

The table summarizes microscope settings and refinement settings for both monomeric and dimeric SpnA.

The endo/exonuclease domain of SpnA is similar to that of the bovine pancreatic (BP) DNAse³⁰ and the catalytic activity of both BP DNAse³⁶ and SpnA³⁰ is Mg²⁺-ion dependent. The cryoEM data revealed a clearly resolved active site within the endo/exonuclease domain, defined by catalytically conserved residues H716, D767 and D810 as reported by Chalmers et al.³⁷ (Fig. 2D, Supplementary Fig. 2B). As the Mg²⁺-ion is present in the structure of BP DNAse (PDB ID:1ATN), we have modeled a corresponding ion for SpnA (Fig. 2A–D, Supplementary Fig. 2C). Chalmers et al.³⁷ have further described E592 and D842 as coordinating a density consistent with a bound Mg²⁺-ion, supporting their role in metal-dependent phosphodiester bond cleavage. In our cryoEM structure, these residues are located too far away to carry out this function (Fig. 2D, Supplementary Fig. 2B), and we suggest that D767 carry out this task. Chang et al.³⁰ have demonstrated that Ca²⁺-ions are required for SpnA nuclease activity, and that the Ca²⁺-ions increase the structural stability of SpnA, as with BP DNAse³⁶. In the absence of Ca²⁺-ions, the SpnA OB1 and OB2 domains are released in limited proteolysis experiments³⁰; an observation supported by our HDX-MS data. The addition on Ca²⁺-ions stabilizes the SpnA structure, as indicated by an increased protection against deuterium uptake, whereas the addition of Mg²⁺-ions to the buffer has little to no effect on the deuterium uptake propensity (Supplementary Fig. 3). Interestingly, the increased protection against deuterium uptake is more prominent in the NUCL and OB3-domain, rather than merely the Ca²⁺-ion coordinating loop (Supplementary Fig. 3), either suggesting additional Ca²⁺-ion binding pockets, or more overarching conformational rearrangements. The findings that SpnA requires Ca²⁺-ions for structural stability and Mg²⁺-ions for catalytic activity are further supported by our circular dichroism (CD) spectroscopy data, which reveal that the addition of Ca²⁺-ions alter the conformation of SpnA, whereas the addition of Mg²⁺-ions does not (Supplementary Fig. 3K–M). In BP DNAse, the Ca²⁺-binding site has been determined to reside in a loop region incompletely resolved³⁶. In our cryoEM model of SpnA, this loop is intact. Based on the Ca²⁺-ion present in the structure of BP DNAse, we have modeled a corresponding ion for SpnA (Fig. 2A–C, E).

An additional OB domain in SpnA and the conserved and distinctive features of the OBs

Based on our cryoEM and HDX-MS data, we propose that SpnA has four OB domains instead of three (Fig. 1B, D, E). Considering the results of Chang et al.³⁰, this indicates that domains OB2-4, but not OB1, are required for host DNA binding during effective catalysis. To get a more in-depth view into the secondary structure organization of the different OB domains as determined by AlphaFold, we compared their secondary structure elements to each other (Fig. 3). The OB domains increase in length according to distance from the N-terminus, with OB1 having 80 residues (Fig. 3A), OB2 95 residues (Fig. 3B), OB3 103 residues (Fig. 3C) and OB4 104 residues (Fig. 3D). Domains OB1-OB3 can be aligned to each other (RMSD between OB1 and OB2 1.572 Å; between OB1 and OB3 4.814 Å; and between OB2 and OB3 7.998 Å), whereas OB4 has a more distinctive 3D structure. The structural difference between OB4 and OB1-3 can be largely attributed to the significant disorder in domain OB4 (55% loops) (Fig. 3D) compared to the other domains (OB1 35%, OB2 26% and OB3 41% loops, as calculated by the number of residues in loop regions divided by the number of residues in regions with ordered secondary structure). Importantly, the high flexibility of the OB4 domain in the AlphaFold model is supported by our cryoEM map and our HDX-MS data with an RMSD between the two of 0.357 Å, compared to the monomeric SpnA. Despite differences in length, number and order of structural elements, all four OB domains share one N-terminal α-helix (here termed α1), and five antiparallel β-sheets (termed β1-β5) forming a barrel-like structure in the center of the respective domains (Fig. 3A–D).

Fig. 3. Oligonucleotide-binding domains (OB) of SpnA.

Sequence of the A OB1, B OB2, C OB3, and D OB4 domains from the AlphaFold model with secondary structure elements (α-helices and β-strands) indicated (left) and the structures with secondary structure elements (right). The figures representing the sequences with overlaid secondary structure elements were generated using ENDscript (https://endscript.ibcp.fr)^56,57.

SpnA-host-plasma interaction network reveal specific interactions with complement

Chalmers et al.³⁷ suggested that the C-terminal nuclease activity of SpnA is not the only function contributing to virulence, and an additional virulence function could reside in the N-terminal parts of the protein. To expand our knowledge on SpnA-host interactions and virulence mechanisms, we used affinity-pulldown mass spectrometry (AP-MS)³⁸ to determine host plasma proteins interacting with SpnA. For this, recombinant SpnA spanning residues 25–876, the same construct as in the cryoEM model, was produced with a C-terminal affinity-tag. The superfolder green fluorescent protein (sfGFP) was used as a negative control³⁸. In pooled normal human plasma, we identified 35 high-confidence SpnA interactions when filtered against sfGFP (Fig. 4A, Supplementary Data 1). The significantly enriched proteins can be grouped into four main categories: complement system proteins, immunoglobulins, apolipoproteins and other plasma proteins. Intriguingly, most of the complement proteins targeted belong to the MAC^1,39 – proteins C5, C6, C7, C8α, C8β, C8γ, C9 (Fig. 4A), products of the activation of all three complement cascade pathways (Fig. 4B). In addition to MAC components, C3 as well as the complement inhibitors clusterin and vitronectin were significantly enriched to SpnA. To further study the interactions between SpnA and complement, we used complement depleted human serum individually depleted of either C3, C5, C6, C7, C8 or C9. In AP-MS experiments in C3 depleted serum, the other complement proteins are not significantly enriched together with SpnA (Fig. 4C). This trend was similarly noted in experiments using C5, C6 and C7 depleted serum, where the enrichment in C3, C5, C6, C7, C8 and C9 were no longer significant (Fig. 4D–F). In experiments using C8 depleted serum, only the enrichment of C3 and C9 was lost (Fig. 4G), as was the case in experiments using C9 depleted serum where only the enrichment of C3 was lost (Fig. 4H). Collectively, these results indicate that SpnA interacts with the MAC assembly intermediate C5b67 in human plasma and serum. The binding of MAC components is human specific, since no interactions were found between MAC proteins in pooled mouse plasma and SpnA (Supplementary Fig. 4).

Fig. 4. Human plasma interactome of SpnA.

A Volcano plot comparing the human plasma interactome of SpnA to GFP. Significantly enriched proteins are indicated on the sides of the plot in Log2 foldchange (FC) order. Baits and key human complement interactors marked in bold. B Schematic overview of the three complement pathways, classical, alternative and lectin all converging to the terminal pathway of membrane attack complex formation. Proteins identified as significantly enriched to SpnA in normal plasma are indicated in green. C–H Volcano plot comparing the human plasma interactome of SpnA to complement depleted plasma (C3–C9) as compared to pooled normal human plasma. Significantly enriched proteins are indicated on the sides of the plot in Log2 FC order. The schematic representation on the right-side hand of the graph indicates the complement system activation terminal pathway. The complement protein depleted in the respective sera is indicated in pink, non-significant protein interactions in gray, and significant protein interactions in blue. Non-significant complement proteins are indicated in green. Significantly enriched proteins are indicated on the sides of the plot in Log2 FC order. For AP-MS assays, n = 5 independent pulldowns were performed.

In our affinity-purification experiments, several immunoglobulins were also found to interact with SpnA (Fig. 4A). The major IgG subclass associated with SpnA is IgG1, in line with what we have previously found for S. pyogenes^13,38,40. The presence of immunoglobulins in the samples could indicate that SpnA triggers the classical pathway of the complement cascade, mediated by immunoglobulins and the C1-complex. Hence, we repeated our affinity-pulldown experiments in plasma where the IgG molecules had been cleaved by the streptococcal enzyme²³ into F(ab)2‘- and Fc-fragments, or depleted using protein-G beads, abolishing downstream activation pathways. We observe that the complement protein binding pattern in IdeS-digested plasma is similar to normal pooled plasma, whereas IgG-depletion impacts the interactome more (Supplementary Fig. 5). As IgA- nor IgM-molecules are cleaved by IdeS or depleted by protein G, our data suggests that SpnA might trigger either alternative complement pathway or the classical pathway mediated by IgA/IgM complexes. Of importance is, however, that in neither IdeS-digested nor protein G-depleted plasma any immunoglobulins are statistically significantly enriched by SpnA, whereas several components of the MAC are (Supplementary Fig. 5).

Identification of the SpnA domains interacting with the complement system

To pinpoint which regions of SpnA mediate the interactions with the MAC assembly intermediate C5b67, we performed cross-linking mass spectrometry (XL-MS) with purified components in solution⁴¹. Here, recombinantly expressed SpnA was crosslinked to either the purified C5b6 intermediate or to C7 using disuccinimidyl suberate (DSS). These experiments confirmed that SpnA OB2-domain β3-sheet, the loop between OB2 and OB3, and the NUCL-domain interact directly with the C5b6 complex (Fig. 5A). For the interaction with C7, the interfaces on SpnA are more dispersed, including the disordered N-terminal region, previously lacking an identified function in SpnA mediated pathogenesis, the OB2-domain β3-sheet, the newly identified OB4-domain, and the NUCL-domain (Fig. 5B). The OB2-domain β3-sheet seems to be crucial for the interaction with these complement components. Despite the β3-sheet structure being conserved in all four OB domains (Fig. 3), there is no apparent sequence similarity between these. These results directly imply a novel function for SpnA in host interactions, namely that of binding to and likely interfering with complement system functions, similarly to what has been described for other streptococcal proteins, such as SIC¹⁶, CEF¹⁵ and possibly SLO²⁰. Importantly, this proposes a role in pathogenesis for the N-terminal domain of SpnA, spanning the N-terminal disordered region and the domains OB1-2 (Fig. 1B), which has been shown to have limited to no importance in DNA degradation³⁰.

Fig. 5. Interaction interfaces between SpnA and complement system factors C5b67.

Cross-linked peptide pairs identified when cross-linking SpnA to the A C5bC6 and the B C7 complexes. Amino acid residues are indicated in numbers, and the crosslinked spectrum matches (CSM) are indicated by the thickness of the connecting edge. Sequence conservation logos of the SpnA peptides mediating the interactions to the C5bC6 and C7 complexes are shown on the left. The crosslinked lysine (K) residue is indicated. Non-conserved residues are indicated with asterisks (*). For XL-MS assays, n = 2 technical replicates.

For this work, we used the Uniprot ID: Q9A0J7 sequence of SpnA as a starting point when designing the recombinantly expressed construct. To determine if our findings are relevant to the broader diversity of SpnA sequences present in different S. pyogenes isolates, we performed an SpnA sequence variation analysis of public genome assemblies comprising 150 distinct emm types, including the entire collection of 20,580 samples analyzed previously⁴², representing isolates recovered globally from invasive and non-invasive S. pyogenes infections. This analysis identified 602 DNA alleles encoding 433 distinct protein alleles (Supplementary Fig. 6, Supplementary Table 2), showing substantial diversity of SpnA. However, the 16 alleles representing each >1% of the genomes, accounted for 71.5% of the entire dataset, meaning that most alleles were present in only a minority of the isolates. We compared the sequences of these 433 alleles to that of Q9A0J7, focusing on the interfaces we identified as crosslinked to the C5b67 complex (Fig. 5A, B). The Q9A0J7 allele is the most frequent allele, present in 19.1% (n = 3933) of the genomes, and it presented a single change relative to the consensus in the regions identified as interacting with the complement components (R729C). This same change is present in another three alleles among the most frequent and in a total of 34.9% of the dataset. Another four changes in the regions interacting with the C5b67 complex were detected among the most frequent alleles: S52T, A236T, T390I, and A690V (Supplementary Fig. 6, Supplementary Table 2). Despite the diversity of SpnA in circulating S. pyogenes strains, the regions interacting with the C5b67 complex are largely conserved. Therefore, the behavior of the SpnA Q9A0J7 allele should reflect that of other frequent alleles suggesting that our findings are broadly applicable to S. pyogenes isolates.

To visualize the interaction interfaces between SpnA and the complement components, these were docked to each other using HADDOCK 2.5⁴³ based on the identified cross-linked interfaces on the respective partners. For the interaction of the SpnA OB2- and NUCL-domains with C5b6, we used a truncated version of the AlphaFold model of SpnA missing the N-terminal and C-terminal disordered regions. The loop between OB2 and OB3 was modeled as flexible. This version of SpnA was docked to the C5b6 model (PDB: 4A5W)⁴⁴, to the C7 AlphaFold model (AlphaFold ID: AF-P10643-F1), and to the C5b67 complex extracted from the sMAC model (PDB: 7NYD)¹ (Fig. 6). All resulting complex models were superimposed on the 7NYD structure, and a proposed interaction surface was determined by highlighting MAC residues within 3 Å of any docked SpnA model (Fig. 6A). This residue set was contained in a section of the MAC assembly facing towards the internal parts of the presumed assembled pore, highlighted with C8 units in purple (Fig. 6D). Approximately 54% of these residues belonged to C5b, 36% to C6, and 10% to C7. The SpnA orientation was then queried by placing centroids representing the N-terminal OB1 and OB2 or the C-terminal OB3-4 and NUCL domains. Overlap between the potential interaction surface residues and centroids of these models indicate that the rightmost placement of the C-terminal domains constitutes a more stable interface. A final top model was then determined by assessing adherence to the DSS distance constraints (Fig. 6B, C) and placed SpnA at the proposed stable interface. Based on these results, SpnA binds to the C5b67 assembly intermediate in a way that inhibits further C8 and C9 MAC assembly (Fig. 6D), indicating a novel mechanism of MAC inhibition.

Fig. 6. SpnA-mediated complement assembly inhibition model.

A SpnA binding interface to the MAC assembly intermediate C5b67 as identified by XL-MS. MAC component interface residues potentially mediating interaction (within 3 Å) are indicated in gray (C5b), blue (C6) and green (C7). Non-interface residues are indicated in black. The structures are derived from PDB: 7NYD. B The highest scoring HADDOCK model of SpnA (front view) docked to the C5b67 interface based on XL-MS derived distance constraints is shown, and a view tilted at 45° below. C5b, C6 and C7 are shown in shades of gray and indicated with labels. The AlphaFold model of SpnA excluding the disordered N- and C- terminal regions is shown in dark blue (NUCL), dark green (OB4), light green (OB3), loops (orange) and pink (OB1-OB2 (color key in panel (C)). C Cross-links supported by the MS data and the HADDOCK docking between SpnA and the C5b67 complex, cross-linked lysines (K) are indicated in black. The cross-links are listed, with distances in Å indicated. D The SpnA-C5b67 complex rotated as shown in panel (B)) displayed together with C8αβγ showing the steric hinderance (indicated in black) posed by SpnA preventing further MAC assembly.

SpnA inhibits the formation of the MAC and prevents hemolysis

The streptococcal proteins SIC and CEF have been demonstrated to inhibit the MAC by blocking insertion of C5b67 complexes into membranes^15,16. To determine if our identified mode of interaction similarly inhibits the insertion of C5b67 complexes, we used a hemolysis assay modified from Fernie-King et al.¹⁶. using human erythrocytes and purified complement components. The addition of SpnA resulted in a significant, concentration dependent reduction in hemolysis when compared to the BSA control (Supplementary Fig. 7A). To study the mechanism of MAC interaction and hemolysis inhibition further, we used affinity-tagged SpnA. SpnA was added to the reaction mixtures sequentially to determine at which stage we observe a decrease in lysis inhibition, indicating that the SpnA interaction interface is masked by the progression of the MAC assembly (Supplementary Fig. 7B). Here, we observe a statistically significant reduction in inhibition if SpnA is added after C8 has already been assembled into the C5b67 complex. We could also demonstrate that SpnA does not associate with the host membranes, suggesting that the inhibitory mechanisms occur in the liquid phase prior to C5b67 membrane insertion (Supplementary Fig. 7D–F). Mutagenesis of key residues (H716G involved in DNA-degradation, C729S; the residue with the lowest degree of sequence conservation at the interaction interfaces (Fig. 5) or deletion of the N-terminus together with all OB-domains) did not affect the inhibition of hemolysis (Supplementary Fig. 7C), consistent with our model of multiple different sites on SpnA being required to mediate the C5bC7 interactions (Fig. 5) and subsequent downstream MAC inhibition.

Discussion

The complement system is an important part of the innate immune system. It is a significant immunological sensor and effector machinery which, in concert with other parts of the innate and adaptive immune systems, defend the host against infection by neutralizing and destroying invading pathogens and regulating the inflammatory response. This response includes opsonophagocytosis, inflammation, pro-inflammatory chemokine secretion and cytokine production. S. pyogenes has evolved to produce several enzymes and proteins inhibiting the function of the complement system by cleaving C3 and C5, inhibiting progression of complement system activation. CEF interferes with the classical complement pathway by binding C1r and C1s in a glycan-dependent manner, impairing C3 convertase activity and C5 cleavage¹⁵. The cysteine protease SpeB degrades C3 and its active fragment C3b, thereby reducing opsonization and the generation of the anaphylatoxins C3a and C5a⁴⁵. Both SpeB and the C5a peptidase ScpA further inactivate C3a and C5a, disrupting neutrophil recruitment to the site of infection^45,46.

Although Gram-positive bacteria have been considered to be protected against the pore-forming action of the MAC due to their thick peptidoglycan layer, several Gram-positive bacteria secrete small proteins that specifically target this multicomponent complex and inhibit its formation. In S. pyogenes, SIC¹³ and possibly the immunogenic secreted protein (ISP)^38,47 perform this function by binding clusterin, a plasma protein acting as a MAC complex regulator. Interestingly, SLO also interacts largely with the same set of plasma proteins as ISP and with a subset of those interacting with SIC, targeting components of the complement system terminal pathway and complement regulators, such as clusterin²⁰. Moreover, it was shown that in Gram-positive bacteria, complement activation leads to specific C3 independent deposition of the MAC on the bacterial surface and that in S. pyogenes this deposition is localized close to the division septum⁴. Our previous results supports these findings and additionally shows that the assembled MAC can take an active, pore-shaped conformation on the surface of S. pyogenes⁴⁸.

Here, we determine that SpnA is a new complement system inhibitor of S. pyogenes. Unlike SIC and ISP, both of which mainly target clusterin - an interaction which for SIC has been demonstrated to prevent the insertion of the C5b67 assembly intermediate into membranes¹⁶, SpnA directly targets the C5b67 complex and sterically hinders its assembly into the pore-shaped MAC via the addition of C8 and C9 (Figs. 4–6, Supplementary Fig. 7). Since SpnA is a cell wall attached protein, this function is suggested to occur primarily in the vicinity of the bacterial surface. We provide a possible mechanism explaining unidentified virulence functions, in addition to DNA degradation, previously suggested to be located on the N-terminal part of SpnA³⁷, since several key residues mediating the interaction with complement components C5b67 reside in this region, both in the extended disordered region including the first 100 residues and in the OB2 domain (Fig. 5). The overall conservation of the SpnA regions interacting with the C5b67 complex highlights their evolutionary importance in modulating this pathogen’s interactions with host immune defenses (Fig. 5, Supplementary Fig. 6). Similarly, the newly defined OB4 domain is conserved even outside of the region interacting with complement, supporting its importance for the nucleolytic function of SpnA (Supplementary Fig. 6).

In addition to functional studies, we provide the first experimentally determined structure for SpnA. Through a combination of cryoEM and HDX-MS we give insights into the structural organization of SpnA and its functional dynamics. Our findings expand upon the initial characterization of SpnA³⁰, where three OB domains (OB1-OB3), a C-terminal nuclease domain and a cell wall-anchor domain were described. By resolving most of SpnA at high resolution we identified a previously unrecognized OB4 domain situated between OB3 and the nuclease domain. The flexible structure of OB4 could be a reason why this domain was overlooked but it is supported here by conclusive data from AlphaFold, HDX-MS and cryoEM (Figs. 1–3).

Although the nuclease domain of SpnA is highly similar to that of BP DNAse³⁶, while the BP DNAse is catalytically active even in the absence of additional nucleotide binding domains, these have been proved essential for SpnA³⁰. It is noteworthy that SpnA requires this extra safeguarding mechanism by coupling DNA degradation and binding to separate domains. The reason for this remains unknown and warrants further investigation. SpnA is the second S. pyogenes nuclease for which the structure has been determined, the first being the substantially smaller Sda1 (44 kDa; PDB: 5FGW)⁴⁹. Like SpnA, Sda1 has been shown to promote streptococcal neutrophil resistance via the degradation of host NETs⁵⁰. However, unlike SpnA but similarly to BP DNase, Sda1 is active without the presence of additional OB domains.

We describe mechanistically the streptococcal inhibition of the assembly of the complement system MAC. While previous studies have identified SIC, CEF and SLO as targeting components of the MAC^15,16,20, and have demonstrated that SIC prevents the insertion of C5b67 intermediates into the membranes, the respective structural mechanisms have remained elusive. Our data demonstrate that a possible mechanism through which SpnA inhibits MAC formation is by binding to the C5b67 assembly intermediate preventing further recruitment of the C8 and C9 complement components (Fig. 6, Supplementary Fig. 7). It is noteworthy that SpnA did not show affinity for MAC components in mouse plasma, in contrast to the observations in human plasma. This underscores the highly specific evolutionary adaptation of S. pyogenes to humans, its only known natural host for both colonization and infection.

Materials and methods

Proteins

The ORFs encoding wt SpnA (UniProt ID: Q9A0J7)³¹, the ΔNSpnA H716G and C729S mutants, IdeS (UniProt ID: Q7DAM2)³¹ and sfGFP³⁸ were obtained as synthetic constructs from GenScript. For the ΔNSpnA H716G and C729S mutants, the constructs contained SpnA residues 97-859, with the respective residues mutated. The constructs were cloned into a pET-26b(+) vector with a C-terminal 6xHis-HA-StrepII-TEV (histidine–hemagglutinin–StrepII–Tobacco Etch Mosaic Virus protease recognition site)-tag, whereas sfGFP was cloned into a pNIC28-Bsa4-based expression vector carrying the same affinity tag at the N-terminal³⁸. All proteins were cloned, expressed and purified at the Lund Protein Purification Platfrom (LP3) in Lund, Sweden.

IdeS³¹ was expressed in Escherichia coli TURNER(DE3) cells in Luria Bertani (LB) broth supplemented with 50 µg/ml kanamycin at 18 °C. At OD₆₀₀ = 0.8 the protein expression was induced with 1 mM IPTG for 21 h. The cells were harvested and resuspended in phosphate buffer (50 mM sodium phosphate pH 8.0, 300 mM NaCl and 20 mM imidazole, supplemented with EDTA-free Complete Protease Inhibitor tablets (Roche). The cells were lysed using a French pressure cell at 18,000 psi. The lysate was cleared by ultracentrifugation (Ti 50.2 rotor, 244,000 × g, 60 min, 4 °C) and subsequently passed through a 0.45 μm filter prior to loading on a HisTrap HP column (GE Healthcare). After washing with 20 column volumes (CV) of phosphate buffer, IdeS was recovered by elution using a 0–500 mM imidazole gradient in phosphate buffer. The IdeS fractions were pooled and incubated with TEV protease at a TEV:IdeS mass ratio of 1:12 in the presence of 1 mM DTT overnight at 16 °C. Cleaved IdeS was separated from uncleaved protein and TEV protease by a second HisTrap HP purification step, collecting the flow-through fraction. The final purification step was performed by size-exclusion chromatography on a HiLoad 26/600 Superdex 200 pg column (GE Healthcare) equilibrated in 1x PBS pH 7.4.

Recombinant sfGFP³⁸ was expressed in E.coli TURNER (DE3) in Terrific Broth (TB) (BD Difco) at 18 °C. At OD₆₀₀ = 0.5–0.7 the protein expression was induced with 1 mM IPTG for 18 h. The cells were harvested and resuspended in phosphate buffer (25 mM sodium phosphate, pH 8.0, 300 mM NaCl, and 20 mM imidazole) supplemented with EDTA-free Complete Protease Inhibitor tablets (Roche). Following cell lysis by French pressure cell disruption as above, the lysate was clarified and purified using a HisTrap HP column (GE Healthcare) as above including TEV cleavage. The purified protein was subsequently dialyzed against 1xPBS pH 7.3.

All SpnA constructs were expressed in E. coli TUNER (DE3) cells in TB (BD Difco) supplemented with 50 μg/ml kanamycin at 18 °C. At at OD₆₀₀ = 1.5 for protein expression was induced with 0.5 mM IPTG for 18 h. The cells were harvested and resuspended in phosphate buffer (50 mM sodium phosphate pH 8.0, 300 mM NaCl, 10% glycerol, 0.5 mM TCEP, 20 mM imidazole) supplemented with EDTA-free Complete Protease Inhibitor tablets (Roche). The cells were lysed using a French pressure cell at 18,000 psi. The lysate was cleared by ultracentrifugation (Ti 50.2 rotor, 244,000 × g, 60 min, 4 °C) and subsequently passed through a 0.45 μm filter prior to loading on a HisTrap HP column (GE Healthcare). The column was washed with 20 CVs of phosphate buffer, and bound protein was eluted using a 20 to 500 mM imidazole gradient in phosphate buffer. The SpnA fractions were pooled and incubated with TEV protease at a TEV:SpnA mass ratio of 1:20, and incubated overnight at 16 °C. Cleaved SpnA was purified on a HisTrapHP column as aforementioned with the flowthrough fraction collected. Both affinity-tagged and protease-cleaved SpnA were further purified by size-exclusion chromatography using a Hi load 26/600 Superdex 200 pg column (GE Healthcare) in 50 mM Tris-HCl pH 8.0, 150 mM NaCl, 10% glycerol, 0.5 mM TCEP, resulting in a monodisperse main peak. The cloning and expression of the mutant containing only the NUCL domain has been described³⁸.

Bovine serum albumin used as a control in the hemolysis assay was from Abcam. The purified complement system proteins used for XL-MS and in the hemolysis assays were from Complement Technology, Inc. (Texas, USA).

SEC-MALS

For the wt SpnA used for single-particle cryoEM, SEC-MALS was used to determine the monodispersity. SpnA was injected into an OMNISEC system (Malvern Panalytical). The system included the OMNISEC RESOLVE module, equipped with either a Superdex 200 Increase 10/300 GL or Superose 6 Increase 10/300 GL column (Cytiva), and the OMNISEC REVEAL, an integrated multi- detector module comprising right-angle light scattering (RALS) at 90°, low-angle light scattering (LALS) at 7°, a differential refractive index detector, a viscometer, and a diode- array-based UV/VIS spectrometer. Data were collected and analyzed using OMNISEC v11.41 software, with the flow rate maintained at 0.5 mL/min.

Plasma

Pooled human plasma from healthy donors and pooled BALB/c mouse plasma was purchased from Innovative Research (MI, USA). Normal human serum individually depleted of complement C5, C6, C7, C8 or C9 was purchased from Complement Technology, Inc. (Texas, USA). In some experimental setups the IgG-molecules in the pooled normal human plasma were digested into F(ab’)₂ and two Fc fragments using IdeS (5 mg/ml) for 4 h at 37 °C, prior to affinity-purification. In other experimental setups, the IgG-molecules were depleted using Protein G Sepharose 4 Fast Flow (Cytiva) according to the manufacturer’s instructions.

Blood collection and preparation

Informed consent was obtained prior to blood collection from healthy donors into 0.1 M Na₃ citrate (BD) as an anticoagulant. Ethical approval was obtained from the local ethics committee (approval 2025-05426-01), and the study was conducted according to the Declaration of Helsinki. The blood was centrifuged at 3000 × g for 10 min to collect the erythrocytes.

Affinity-purification experiments

The affinity-purification experiments have been described³⁸. Briefly, Strep-Tactin beads (IBA Life Sciences) were equilibrated in PBS-buffer (1X PBS pH 7.4, 1 mM CaCl₂, 5 mM MgSO₄) and charged with 100 µg of recombinant, affinity-tagged SpnA. Affinity-tagged sfGFP was used as a negative control. Pooled normal human plasma (100 µl), IdeS-digested pooled normal human plasma (100 µl), IgG depleted pooled normal human plasma (100 µl), BALB/c mouse plasma (100 µl) or complement component depleted human serum (100 µl) was incubated with the protein-charged beads at 37 °C, 1000 rpm, 30 min. The beads were washed with 10 mL ice-cold PBS-buffer, and proteins eluted with 180 µl of 5 mM biotin in PBS-buffer. The samples were precipitated with a final concentration of 25% trichloroacetic acid (TCA), washed with acetone and dried in a speedvac prior to sample preparation for mass spectrometry.

Cross-linking

For cross-linking, 5 μg of SpnA was mixed with 5 μg of C5bC6 or C7 in a final reaction volume of 30 μl in 1X PBS buffer complemented with 1 mM CaCl₂ and 5 mM MgSO₄, and incubated for 15 min, 37 °C, 1000 rpm in a thermoblock for the proteins to bind to each other. Disuccinimidyl suberate (DSS-H12/DSS-D12, Creative Molecules Inc., 001S) was added to a final concentration of 1 mM and the samples further incubated for 30 min, 37 °C, 1000 rpm. Non-cross-linked samples were kept as controls. The cross-linking reaction was quenched with a final concentration of 50 mM of Tris-HCl pH 7.4 for 15 min, 37 °C, 1000 rpm.

Sample preparation for mass spectrometry

The affinity-purified or cross-linked samples were resuspended in 8 M urea-100 mM ammonium bicarbonate (ABC), cysteine-bonds reduced using 5 mM tris(2-carboxyethyl)phosphine hydrochloride (TCEP) (37 °C, 60 min, 800 rpm) and alkylated with 5 mM iodoacetamide (22 °C, 30 min). The samples were subsequently diluted with ABC to a final urea concentration of 1.5 M. For the cross-linked samples, 2 µg of lysyl endopeptidase (LysC; Wako) was added, and the samples digested (37 °C, 2 h, 1000 rpm). Sequencing grade trypsin (Promega) was added to all samples for digestion (37 °C, 18 h, 1000 rpm). The samples were acidified with 10% formic acid (FA) to a final pH <3.0 and purified using C18 reverse phase spin columns according to manufacturer’s instructions (The Nest Group, Inc.). The purified peptides were dried in a speedvac and reconstituted in of 2% FA in 2% acetonitrile prior to mass spectrometric analysis.

Liquid chromatography tandem mass spectrometry (LC–MS/MS)

All peptides were analyzed on an Eclipse mass spectrometer connected to an ultra-high performance Ultimate3000 liquid chromatography system (Thermo Scientific). Approximately 150 ng of peptides were separated on a Thermo EASY-Spray column (Thermo Scientific 25 cm column, column temperature 45 °C) operated at a maximum pressure of 900 bar. For affinity-pulldown samples, a linear gradient of 3% to 38% of 80% acetonitrile in aqueous 0.1% formic acid was run for 90 min at a flow rate of 300 nl/min. For crosslinked samples, a linear gradient of 5–25% of 80% acetonitrile in aqueous 0.1% formic acid for 100 min followed by a linear gradient of 25–45% of 80% acetonitrile in aqueous 0.1% formic acid for 20 min. One full MS scan (resolution 120,000 for a mass range of 350-1400 m/z (affinity-pulldown samples or 400–1600 m/z (cross-linked samples)) was followed by MS/MS scans (resolution 15,000 m/z). The cycle time was 3 sec. For cross-linked samples, precursors with an unknown charge state, a charge state of 1, 2, or above 9 were excluded. The precursor ions were isolated with 1.6 m/z isolation window and fragmented using higher-energy collisional-induced dissociation (HCD) at a normalized collision energy of 30 for affinity-pulldown samples or using stepped HCD at a normalized collision energy of 21, 26, 31 for cross-linked samples. The dynamic exclusion was set to 45 or 60 s.

Data analysis of affinity-purification experiments

The mass spectrometry data of the affinity-purified human samples were analyzed in MaxQuant (v 2.6.7.0) against an in-house generated database containing the reviewed Homo sapiens Uniprot proteome (proteome ID: UP000005640), where the complement component C3 and C5 sequences were split into C3a and C3b, as well as C5a and C5b respectively. The database was complemented with the sequences for SpnA and sfGFP. Fully tryptic digestion was used allowing two missed cleavages. Carbamidomethylation (C) was set to static and oxidation (M) to variable modifications. Mass tolerance for precursors was set to 10 ppm and for fragment ions to 0.02 Da. The protein false discovery rate was set to 1%. Proteins identified by two or more peptides and one or more proteotypic peptides were considered as relevant. The label-free quantification (LFQ) intensities were Log2 transformed and median normalized in Perseus (v 2.0.11.0). Any missing values were imputed from the normal distribution (width 0.3, down shift 1.8). For the generation of volcano plots of the affinity-purification samples the FDR was 0.05 and S.O 0.01.

Data analysis of cross-linking experiments

All spectra from cross-linked samples were analyzed using pLink 2 (version 2.3.11)⁵¹. Prior to cross-linked peptide identification, the proteins present in the SpnA-C5bC6 or SpnA-C7 samples were analyzed using MaxQuant as described above, and the output containing SpnA-C5bC6 or SpnA-C7 as well as any other proteins present originating from the complement protein preparations, were used to generate a sequence database. pLink2 was run using default settings for conventional HCD DSS-H12/D12 cross-linking, with trypsin as the protease and up to three missed cleavages allowed. Peptides were selected with a mass between 600 and 6000 Da, and a length between 6 and 60 amino acids. Precursor and fragment tolerance were set to 20 and 20 ppm, respectively. For pLink2, crosslink identifications were filtered by requiring 10 ppm mass accuracy, false discover rate (FDR) <  5%, and an E-value <  0.01.

Docking of complexes

For cross-linking guided docking with HADDOCK 2.5⁴³, SpnA was trimmed to include only OB1-NUCL, residues 97–850 (Fig. 1B) and docked to the C5bC6 model (PDB: 4A5W)⁴⁴, to the C7 AlphaFold model (AlphaFold ID: AF-P10643-F1), and to C5b67 complex extracted from the soluble MAC model (PDB: 7NYD) with allowed K-K distances of 12–35 Å as determined by the spacer arm length of DSS. All HADDOCK output complexes were superposed on 7NYD to the corresponding MAC protein (C5bC6 or C7), and subsequently analyzed, using PyMOL (v3.1.3.1) and Python (v3.12.9) to assess modeled cross-linking adherence. The RMSD was on average 1.8, 0.6, and 3.3 for SpnA-4A5W/7NYD-C5b67/AF-P10643-F1 respectively. The SpnA-7NYD interface residues were selected with a 3 Å cut-off for heavy atoms between the interactors, and centroids were calculated based on coordinates for residues 89-363 (OB1-2) and 379-938 (OB3-4 and NUCL). The top model was subsequently selected by scoring each model based on the number and gaussian weighted length of the satisfied constraints (as described in ref. ⁴¹). The maximum linker length was set to 32 Å.

Hydrogen–deuterium exchange mass spectrometry (HDX-MS)

All chemicals were from Sigma Aldrich. The pH measurements were made using a SevenCompact pH-meter equipped with an InLab Micro electrode (Mettler-Toledo), a 4-point calibration (pH 2,4,7,10) was made prior to all measurements. The HDX-MS analysis was made using automated sample preparation on a LEAP H/D-X PAL™ platform interfaced to an LC-MS system, comprising an Ultimate 3000 micro-LC coupled to an Orbitrap Q Exactive Plus MS.

HDX measurements were performed on SpnA using four different buffering solutions. Apo runs were performed using 10 mM PBS, pH 7.5, 150 mM NaCl and dPBS. For SpnA with Ca²⁺ -and/or Mg²⁺-ions, these were added to the buffers at concentrations of 1 mM CaCl₂ and 5 mM MgSO₄. The four buffer combinations were: Apo, Mg²⁺, Ca²⁺ and Mg²⁺ and Ca²⁺. For each protein-buffer combination and labeling time point, triplicate samples were run. Deuterium labeling was carried out for t = 30, 300, 3000, 9000 s at 4 °C.

In all experiments 3 μl of 25.3 μM SpnA samples were diluted with 27 μl of 10 mM PBS, pH 7.5, 150 mM NaCl, or HDX labeling buffer of the same composition prepared in D₂O, pH_(read) 7.1. The labeling reaction was quenched by dilution of 30 μl labeled sample with 30 μl of 1% TFA, 0.4 M TCEP, 4 M urea, pH 2.5 at 1 °C. For analysis, 60 μl of the quenched sample was directly injected and subjected to online pepsin digestion at 4 °C using a Nepenthesin-2/pepsin column (Affipro, CZ). The online digestion and trapping were performed for 4 min using a flow of 50 µl/min 0.1% formic acid, pH 2.5. The peptides generated by pepsin digestion were subjected to on-line SPE on a PepMap300 C18 trap column (1 mm × 15 mm) and washed with 0.1% FA for 60 s. Thereafter, the trap column was switched in-line with a reversed-phase analytical column, Hypersil GOLD, particle size 1.9 µm, 1 ×50 mm, and separation was performed at 1 °C using a gradient of 5–50% B over 8 min and then from 50 to 90% B for 5 min, the mobile phases were 0.1% formic acid (A) and 95% acetonitrile/0.1% formic acid (B). Separated peptides were analyzed on a Q Exactive Plus MS, equipped with a HESI source operated at a capillary temperature of 250 °C with sheath gas 12, aux gas 2 and sweep gas 1 (au). For HDX analysis MS full scan spectra were acquired at 70,000 resolution, with an AGC target of 3e6, a maximum injection time of 200 ms and a scan range 300–2000 m/z. For identification of generated peptides separate undeuterated samples were analyzed using data dependent MS/MS with HCD fragmentation. A summary of the HDX experimental detail is reported in Supplementary Table 2.

HDX-MS data analysis

PEAKS Studio X Bioinformatics Solutions Inc. (BSI, Waterloo, Canada) was used for peptide identification after pepsin digestion of undeuterated samples. The search was done against the SpnA sequence, using a mass error tolerance of 15 ppm and a fragment mass error tolerance of 0.05 Da, allowing for fully unspecific cleavage by pepsin. Peptides identified by PEAKS with a peptide score value of log P > 25, and no modifications were used to generate a peptide list containing peptide sequence, charge state and retention time for the HDX analysis. HDX data analysis and visualization was performed using HDExaminer, version 3.4.2 (Sierra Analytics Inc., Modesto, US). The uptake analysis was using only charge states found in the peptide search, allowed only for EX2 kinetics and the first two residues of a peptide was assumed unable to hold deuteration. The presented deuteration data is the average of all high and medium confidence results. The allowed retention time window was ±0.5 min. The difference heatmaps were drawn using an automatically calculated significance based on replicate variance. For visualization, we have assumed 75% deuteration as the endpoint for all peptides.

Circular dichroism

To analyze changes in the secondary structure of SpnA, circular dichroism (CD) was used. All reactions were performed using 5 µM SpnA in buffer containing 150 mM Tris-HCl, complemented with 5 mM CaCl₂, 10 mM MgSO₄ or 5 mM CaCl₂ and 10 mM MgSO₄. All measurements were performed on a Jasco J-810 spectropolarimeter (Jasco) equipped with a Jasco CDF-426S Peltier set to 25 °C. For the measurements, 1 mm quartz cuvettes (HellmaAnalytics) were used; spectra were recorded at 190–260 nm with a scan speed of 20 nm/min. The raw spectra were corrected for their respective buffers set as baseline and converted to mean residue ellipticity, θ (mdeg cm² dmol⁻¹).

Hemolysis inhibition assays

To test the effects of SpnA as an inhibitor of complement mediated cell lysis, SpnA and BSA (negative control) were diluted in 1x PBS, pH 7.4 complemented with 1 mM CaCl₂ and 5 mM MgSO₄ (reaction buffer). The complement components C5b6, C7, C8 and C9 were all diluted separately to 10 µg/ml in the reaction buffer. For the assay, all experiments were prepared as three technical replicates. Here, we used 50 µl of 5% erythrocytes in reaction buffer, mixed with 0.1, 1.0 or 10 µM of SpnA or BSA and 2.5 µg of each complement component added sequentially, and incubated at 37 °C, 30 min. Erythrocytes mixed with water, with reaction buffer or with the complement components in reaction buffer only were used as controls. The plates were centrifuged to pellet intact cells and cell membranes, and 200 µl supernatant from each well was transferred to a fresh plate to measure hemoglobin release. To determine at which step of the MAC assembly process SpnA inhibits further complex assembly, 10 µM of affinity-tagged SpnA and 2.5 µg of each complement component was added to erythrocytes as follows: SpnA-C5b6789, C5b6-SpnA-C789, C5b67-SpnA-C89, C5b678-SpnA-C9 or C5b6789-SpnA. For western blot analysis, the pellets were washed twice in reaction buffer prior to analysis. The percentage of lysis was calculated according to Fernie-King et al.¹⁶.

Western blot analysis

For assessing the human plasma IdeS-mediated IgG cleavage, the samples were loaded onto SDS-PAGE gels and electrophoresed for 35 min at 150 V. Subsequently, proteins were transferred onto a 0.2 µm PVDF membrane (Trans-Blot Turbo Mini format, Bio-Rad Laboratories) using the Trans-Blot Turbo Transfer System for 5 min according to the manufactures’s instructions. The membrane was blocked overnight at 4 °C in PBS containing 2% BSA PBS-T. Following the blocking, membranes were incubated for 1.5 h at RT with affinity-purified protein G-HRP conjugate (Bio-Rad Laboratories) diluted in 2% BSA PBS-T according to manufactures instructions. Membranes were washed three times in PBS-T for 5 min. Signal detection was performed using Clarity Western ECL substrate (Bio-Rad Laboratories) and imaged using a chemiluminescence imaging system.

To determine if SpnA is associated with host membranes upon complement inhibition (see above), washed erythrocytes were mixed with SDS-PAGE loading buffer and resolved on 7.5% SDS-PAGE gels for 60 min at 150 V. Control samples consisting of 1 µg of SpnA were included. Assuming 100% SpnA membrane association, the analyzed washed erythrocytes were estimated to contain 5 μg of affinity-tagged SpnA. Proteins were then transferred onto a 0.2 µm PVDF membrane (Trans-Blot Turbo Mini format, Bio-Rad Laboratories) using the Trans-Blot Turbo Transfer System at a constant current for 5 min, 25 V, 2.5 A. The membrane was blocked for 1 h at RT in 3% BSA PBS-T and subsequently 6x-His Tag Polyclonal Antibody, HRP (Invitrogen) was added in a 1:5000 dilution and incubated for 1 h at RT. The membrane was then washed three times in PBS-T for 5 min. Signal detection was performed as above.

Cryo-EM data collection and data processing

The recombinantly expressed SpnA was concentrated to a final concentration of 10 mg/ml in PBS buffer containing 1 mM CaCl₂, 5 mM MgSO₄. For grid preparation sample was diluted to 0.5 mg/ml in the same buffer and 4 µl of the sample was applied to glow-discharged Quantifoil 1.2/1.3 300 grids. Vitrification was performed on a Vitrobot MarkIV (Thermo Fisher, Eindhoven, Netherlands) for 5 s at 4 °C and 100% using a blot force of −5 and a blotting time of 5 s.

Data was collected on Titan Krios G2 300 kV cryo-TEM, equipped with a Falcon 4i direct electron detector (4k x 4k pixels) and a Selectris energy filter (Thermo Fisher, Eindhoven, Netherlands). Movies were recorded in electron event representation (eer) format at a nominal magnification of 215,000×, corresponding to a pixel size of 0.57 Å/pixel. The total electron dose was 60 e⁻/Ų, with a defocus range of −1.0 to −2.2 µm and exposure time of 2.05 s. Data acquisition was performed using EPU data collection software (version 3.2.0, Thermo Fisher, Eindhoven Netherlands).

A total of 10 648 micrographs was collected in a single session. All image processing steps were performed using cryoSPARC (version 4.6.2). Motion correction and dose weighting were carried out using Patch Motion Correction, followed by Patch CTF estimation for contrast transfer function (CTF) determination. Particles were picked using blob picker, yielding 9 810 516 particles.

Subsequent 2D classification was used to discard poorly aligned or contaminant particles, resulting in 3,458,128 particles across the highest-quality classes for 3D analysis. To resolve structurally distinct species from the same dataset, including both monomeric and dimeric forms, an iterative strategy of multiple rounds of 2D classification and ab initio reconstruction was employed. These steps enabled separation of heterogeneous populations and improved model convergence. The dataset was subjected to heterogeneous refinement, resulting in clear partitioning into monomeric and dimeric classes. The dimeric complex was refined directly using non-uniform refinement, producing a final reconstruction at 2.80 Å resolution. The monomeric class, in contrast, required an additional round of heterogeneous refinement to further improve homogeneity before final refinement, yielding a final reconstruction at 2.60 Å. Initial model building was performed using Phenix Predict and Build⁵² with a trimmed AlphaFold model AF-Q9A0J7-F1 as the starting template, followed by manual curation and structure validation on COOT (version 0.9.8.92)⁵³, guided by local map quality and geometry validation metrics.

SpnA sequence variability analysis

To investigate the SpnA variability in circulating S. pyogenes isolates, the sequence of the spnA gene was analyzed in a dataset of 20,580 assemblies from public whole-genome sequencing data of isolates recovered worldwide from invasive and non-invasive infections⁴². The spnA allelic variants and respective protein variants were obtained from the S. pyogenes whole-genome multilocus sequence typing schema (locus wgMLST-00047545). Alleles and multiple sequence alignments were obtained with chewBBACA⁵⁴.

Statistics and reproducibility

Mass spectrometry data were processed and analyzed using MaxQuant (v 2.6.7.0). Protein identification and quantification parameters, including database searching and filtering criteria are described in the corresponding methods section.

Statistical analyses and graphical representation were performed using Perseus (v 2.0.11.0) and Graphpad Prism (v 10.5.0). Statistical tests were selected according to the experimental design and are indicated in the corresponding figure legends. Data are presented as mean ± SD as indicated in the figure legends. The exact number of replicates used for each analysis is reported in the corresponding figure. Statistical significance was defined as P < 0.05 unless otherwise stated.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

I.A.B., I-M.F., L.B. and L.J.H. conceptualized the manuscript. The methodology was jointly designed by I.A.B, J.S, R.M, A.F., M.R., A.S., S.E. and L.J.H. I.A.B, A.N., M.H., A.S. and L.J.H. conducted the laboratory experiments and data collection. Data analysis and interpretation were carried out by I.A.B., J.S., R.M., A.N., A.F., A.S., S.E. and L.J.H. Computational modeling was designed and conducted by J.S., and bioinformatic analyses were performed by J.S., R.M. and A.F. I.A.B., J.S., R.M., A.N., A.F., A.S., S.E. and L.J.H. made the figures. L.J.H. supervised the work with input from M.C. and L.M. and secured the funding. I.A.B, I-M.F, L.B. and L.J.H wrote the initial draft of the manuscript, with critical input from J.S., R.M., A.N., A.F., M.R., M.H., M.C., L.M., A.S. and S.E. All authors have reviewed and approved the final version of the manuscript.

Peer review

Peer review information

Communications Biology thanks the anonymous reviewers for their contribution to the peer review of this work. Primary Handling Editors: Dr. Dana Reichmann and Dr. Nilanjan Banerjee.

Funding

This work was supported by grants from the Swedish Research Council (2022-03860), the Royal Physiographic Society in Lund, Stiftelsen Clas Groschinskys Minnesfond, and the Foundations of Åke Wiberg, Alfred Österlund and Crafoord to LJH. R.M. was supported by the Fundação para a Ciência e Tecnologia (FCT) (grant 2020.08493.BD). Open access funding provided by Lund University.

Data availability

Coordinates of the SpnA monomer and dimer structures and the corresponding cryoEM map have been deposited to the Protein Data Bank under accession codes 9SF8 and 9SF9, and EMDB entry codes EMD-54816 and EMD-54817, respectively. The affinity-pulldown and XL-MS data have been deposited to the ProteomeXchange consortium via the MassIVE partner repository https://massive.ucsd.edu/ with the dataset identifier PXD076782. The HDX-MS data and HDExaminer analysis files have been deposited to the ProteomeXchange Consortium via the MassIVE partner repository with the dataset identifier PXD076781. Source data for findings presented in main figures is presented as Supplementary Data 1. The SpnA DNA sequence alleles and their changes in regions interacting with C5b7 complex is presented as Supplementary Data 2. The plasmids generated in this study have been deposited into the Addgene repository with the following IDs: 259974 (ΔNSpnA C729S), 260160 (ΔNSpnA H716G) and 260161 (wt SpnA).

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s42003-026-10956-9.