Human Serum Albumin Binds Streptolysin O (SLO) Toxin Produced by Group A Streptococcus and Inhibits Its Cytotoxic and Hemolytic Effects

The pathogenicity of group A Streptococcus (GAS) is mediated by direct bacterial invasivity and toxin-associated damage. Among the extracellular products, the exotoxin streptolysin O (SLO) is produced by almost all GAS strains. SLO is a pore forming toxin (PFT) hemolitically active and extremely toxic in vivo. Recent evidence suggests that human serum albumin (HSA), the most abundant protein in plasma, is a player in the innate immunity “orchestra.” We previously demonstrated that HSA acts as a physiological buffer, partially neutralizing Clostridioides difficile toxins that reach the bloodstream after being produced in the colon. Here, we report the in vitro and ex vivo capability of HSA to neutralize the cytotoxic and hemolytic effects of SLO. HSA binds SLO with high affinity at a non-conventional site located in domain II, which was previously reported to interact also with C. difficile toxins. HSA:SLO recognition protects HEp-2 and A549 cells from cytotoxic effects and cell membrane permeabilization induced by SLO. Moreover, HSA inhibits the SLO-dependent hemolytic effect in red blood cells isolated from healthy human donors. The recognition of SLO by HSA may have a significant protective role in human serum and sustains the emerging hypothesis that HSA is an important constituent of the innate immunity system.


INTRODUCTION
Group A Streptococcus (GAS, Streptococcus pyogenes) is a Grampositive bacterium that causes a variety of diseases, ranging from pharyngitis to severe invasive infections, associated with a possible poor prognosis despite adequate antibiotic therapy. Its clinical impact is also related to the development of non-suppurative sequelae such as acute rheumatic heart fever and poststreptococcal glomerulonephritis (1). In the current therapy of severe streptococcal infection, a combination of clindamycin with a b-lactams molecule is used with the rationale to lower toxin production through protein synthesis reduction (clindamycin effect). Apart from a somatic constituent, a plethora of extracellular streptococcal toxins are involved in the pathogenesis of the severe invasive GAS diseases (2); thus, specific anti-toxin treatments are likely to have an impact on the clinical outcome. Among these extracellular products, the cytolysin streptolysin O (SLO) is produced by almost all GAS strains. SLO has a toxic activity toward red blood cells (RBCs), polymorphonuclear leukocytes, and platelets, as well as isolated mammalian and amphibian hearts. SLO has a tissue-destructive activity and is rapidly lethal when injected intravenously into mice or rabbits as a purified compound (3)(4)(5). At sublethal doses in vivo, SLO causes dermal necrosis, venous congestion, increased vascular permeability, and neurological abnormalities before death (6). More recently, SLO has been demonstrated to be essential in the pathogenesis of severe invasive infection in animal models (7).
SLO belongs to the cholesterol-dependent cytolysin (CDC) proteins, a conserved family of b-barrel pore-forming toxins (PFTs) that are secreted by Gram-positive bacteria (8,9). PFTs are produced by many pathogenic bacteria from the genera Escherichia, Staphylococcus, Clostridioides, Streptococcus, Listeria, Bacillus, and Arcanobacterium, being important components of their virulence (10,11). The CDCs share the undecapeptide and cholesterol binding motif, as well as the ability to form pores on host cell membranes via a cholesteroldependent mechanism (9,12). Although membrane cholesterol is required for the cytolytic mechanism of CDCs to form a complete transmembrane pore, also glycans receptors define the cellular tropism and are required for efficient deposition of PFTs, including CDCs, into cell membranes (8,(13)(14)(15)(16). Following secretion from bacteria as soluble monomers, CDCs bind to host cell membranes through the C-terminal domain 4 (D4), undergo oligomerization, and form the pre-pore complex, which then collapses into the membrane to form the b-barrel pore that permeabilizes target cells and causes lysis and death (8,(17)(18)(19)(20).
Human serum albumin (HSA) is the most abundant protein in plasma (∼7.0×10 −4 M) and represents the main modulator of fluid distribution between body compartments. By acting as a depot and carrier for many endogenous and exogenous compounds, HSA affects the pharmacokinetics of many drugs, and accounts for most of the antioxidant capacity of human serum (21)(22)(23)(24). HSA is a 66 kDa all-a-helical protein built by three homologous domains (labeled I, II, and III). Fatty acids (FAs) represent the primary physiological ligands of HSA (24). HSA acts as a self-defense mechanism toward Clostridioides difficile (C. difficile) infection (CDI) by binding both toxins A (TcdA) and B (TcdB); indeed, HSA induces a conformational change that promotes toxins autoproteolysis outside the intestinal epithelial cells. This event prevents toxins internalization and the consequent cytotoxic and cytopathic effects, both in vitro and in vivo. This provides an explanation for the clinical correlation between CDI severity and hypoalbuminemia (25,26).
Here, the capability of HSA to bind SLO and to modulate its pathogenic actions is reported by combined in silico, in vitro, and ex vivo approaches. We showed that HSA binds SLO with high affinity through a non-conventional binding site located in domain II. This binding inhibits SLO cytotoxicity in vitro in human epithelial type 2 (HEp-2) and human lung carcinoma (A549) cells and significantly reduces its hemolytic effects in red blood cells. Overall, these results support the assumption that HSA protects human cells from SLOdependent toxicogenic effects.

Ethical Standards
Human RBCs were collected from four of the co-authors of the present work, who voluntarily, without constraints, and freely donated blood samples for the study purposes. Therefore, Institutional Review Board approval was not requested. Blood has been collected in accordance with the World Medical Association's Declaration of Helsinki.

Commercial Proteins
SLO isolated from GAS (S5265; Merck KGaA, Darmstadt, Germany) was dissolved in cold deionized water to obtain a stock solution of at least 125 U/mL (according to the manufacturer's units indication), corresponding to 2.2×10 −6 M. SLO has been always activated by incubation with 2.0×10 −2 M dithiothreitol (DTT; Merck KGaA) for 15 minutes at 37°C. Fatty acid-and globulin-free HSA (99.5%; A3782; Merck KGaA) and fatty acid-free bovine serum albumin (BSA) (A4612; Merck KGaA) were dissolved in deionized water at a final concentration of 2.0×10 −3 M, corresponding to 132 mg/mL. All commercial proteins were of reagent grade and used without further purification. SLO, HSA, and BSA concentrations were determined spectrophotometrically using the following values of e 280nm : 71.280, 39.310, and 43.824 M −1 cm −1 , respectively (http:// www.expasy.org). SLO, HSA, and BSA were handled following the manufacturer's instruction and using established good laboratory practices under biosafety cabinets with installed HEPA filters to avoid lipid contamination.

Protein-Protein Docking
Docking simulations of the SLO three-dimensional structure (PDB ID: 4HSC) bound to HSA (PDB ID: 1AO6) were performed using ZDOCK 3.0.2 server (28). ZDOCK implements a Fast Fourier Transform algorithm and a scoring system based on a combination of shape complementarity, electrostatics, and statistical potential terms. The top 216 docked structures for the SLO-HSA complex were rescored using ZRANK (29), which uses a more detailed potential including electrostatics, van der Waals, and desolvation terms. The best pose was finally refined by Rosetta Docking 2 server that uses the Rosetta force field to evaluate the energy of the complexes (30).

Spectrofluorimetric Binding Assay
Values of the apparent dissociation equilibrium constant for binding of SLO to the recombinant wt-HSA or to the L305A/ F374A-HSA (i.e., K d ) were determined by mixing the toxin solution (final concentration 2.0×10 −8 M) with the HSA solution (final concentration ranging from 0 M to 4.0×10 −8 M). The formation of the HSA:SLO complex was monitored spectrofluorimetrically between 300 nm and 400 nm. The excitation wavelength was 280 mm and the slit width was 5 nm. Values of K d were obtained from the dependence of the fluorescence intensity change (DF) on the HSA concentration (i.e., [HSA]), according to Eq. 1: where DF tot is the total fluorimetric change.

ELISA Binding Assay
The amount of HSA required to efficiently coat wells of the ELISA plate (NUNC MaxiSorp ™ flat-bottom, Thermo Fisher Scientific, Waltham, MA, USA) was determined using 1.

Ex Vivo Red Blood Cell Hemolysis Assay
The RBC cell hemolysis assay was performed according to previously published protocols (8,31). Briefly, human whole blood was centrifuged at 500×g for 5 min, in order to separate the RBCs from plasma.

Molecular Docking of SLO Binding to HSA
SLO consists of four domains (D1-D4), among which D3 provides the transmembrane spanning regions of the toxin and D4, which contains the highly conserved undecapeptide sequence (Glu529-Arg539), takes part in the initial interactions with the membrane, including direct recognition of cholesterol (18,(32)(33)(34). The SLO residues Gln476, Trp503, Trp537, and Trp538, of which Trp537 and Trp538 are located within the undecapeptide, play a key role in the recognition and binding to the host cell membrane (8,19). The docking analyses predicted that the interaction between SLO and HSA involves the D4 of SLO and the domain II of HSA ( Figure 1A). In particular, the binding interface is lined by residues Leu480, Trp503, Ala534, Trp537, and Trp538 of SLO, and residues Leu305, Ala306, Val310, Tyr341, Phe374, Phe377, and Val381 of HSA. These residues are located nearby each other, suggesting a hydrophobic nature of the interaction. Remarkably, HSA residues involved in the interaction with SLO are the same as those involved in the recognition of Clostridioides difficile toxins (26). The HSA-SLO complex is stabilized by a hydrogen bond between Glu311 and Lys378 residues of HSA and Gln476 and Glu536 side chains of SLO, respectively. Interestingly, Trp503, Trp537, and Trp538 of SLO involved in the recognition of HSA are located in the undecapeptide region, which is a signature sequence in CDCs that plays a key role in the interaction with host membrane cholesterol (8).

SLO Binding to HSA
On the basis of docking results, Leu305 and Phe374 residues of HSA were mutated to evaluate their role in SLO recognition. We showed that recombinant wt-HSA and L305A/F374A-HSA bind to SLO in a saturating dose-dependent manner ( Figure 1B). According to Eqn. 1, data analysis indicated that the affinity of wt-HSA for SLO (K d = [2.5 ± 0.2]×10 −9 M) is higher than that of L305A/F374A-HSA (K d = [7.3 ± 0.5]×10 −9 M). SLO binding to HSA was confirmed by ELISA assay that was performed by coating wells with HSA ( Figure S2) and testing SLO binding. Results obtained indicated a dose-dependent increase of the absorbance values, with a 2-fold (P<0.01) and a 3.8-fold (P<0.001) increase at 1.0×10 −8 M and 2.0×10 −8 M SLO compared to HSA-coated well without SLO addition (i.e., 0 M SLO) ( Figure 1C).  Pull-down experiments using recombinant wt-HSA bound to Ni 2+ -magnetic beads further supported HSA:SLO interaction as revealed by the presence of the 69 and 55 kDa bands of SLO (36) in the elution fraction ( Figure 1D, lane + HSA). To confirm wt-HSA binding to the beads, the presence of HSA was detected by Western blot (Figure S4).

HSA Exerts a Neutralizing Effect Toward SLO-Induced Cytotoxicity in Epithelial HEp-2 and A549 Cells
The cytotoxic activity of SLO in vitro is significantly increased in the presence of reducing agents (e.g., DTT) (36,37). To achieve SLO activation without inducing cytotoxic effects, preliminary experiments were performed using HEp-2 and A549 cells to find out the optimal concentration of DTT to be used. Results showed that 2.0×10 −3 M DTT was sufficient to induce SLO activation without affecting cell viability ( Figure S5).
As the docking analysis revealed that HSA recognizes part of the undecapeptide region of SLO (Figure 1), which is required to bind membrane cholesterol of the host cells and to exert SLO cytolytic activity (8), we studied whether HSA was able to protect HEp-2 cells from SLO-dependent permeabilization. Cells staining with the PKH67 fluorescent cell linker dye solution allowed visualization of cell membranes, whereas propidium iodide (PI) staining, which is not permeant to live cells, binds the nucleic DNA of permeabilized dead cells. Results obtained showed that SLO induced cell permeabilization (100% PIpositive cells), whereas the presence of HSA in the medium completely inhibited the SLO-dependent permeabilization (Figure 3).

HSA Protects From SLO-Induced Cytotoxicity in Red Blood Cells
The protective effect of HSA toward SLO-induced cytotoxicity was also evaluated ex vivo by a hemolytic assay (Figure 4). First, a dose-response analysis of SLO hemolytic effect in RBCs was performed ( Figure 4A). Results obtained showed a percentage of hemolysis proportional to the units of SLO added, with a 40% hemolysis in RBCs treated with 5 U SLO compared to the negative control. The pre-incubation of SLO for 15 minutes with either 1.0×10 −5 M or 1.0×10 −4 M HSA caused a reduction of RBCs hemolysis to 20% (P<0.001) and 9% (P<0.0001), respectively, compared to HSA-untreated cells exposed to SLO. However, 1.0×10 −4 M BSA did not inhibit the SLO-dependent hemolysis indicating that the neutralizing effect is specific for human albumin ( Figure 4B). This is not unusual; indeed, it has been reported that the binding sites of human, bovine, equine, and leporine albumin are not conserved and do not always coincide (39). Of note, HSA alone did not exert any hemolytic effect at both concentrations tested (1.0×10 −5 M and 1.0×10 −4 M) ( Figure S7).

DISCUSSION
In the last years, the hypothesis that HSA is a player in the innate immunity "orchestra" has emerged (25,26,40). Previously, we demonstrated that HSA binds C. difficile TcdA and TcdB toxins that reach the bloodstream after being produced in the colon (25,26). Here, we report the in vitro and ex vivo capability of HSA to neutralize the effects of the PFT produced by GAS, i.e., SLO. HSA binds SLO with high affinity through a non-conventional binding site located in domain II. This interaction, involving the same domain of HSA that has been previously demonstrated to interact with the TcdA and TcdB toxins of C. difficile (26), protects epithelial HEp-2 and A549 cells as well as ex vivo RBCs from the cytotoxic effects of SLO. The HSA:SLO interaction takes place outside cells impairing the SLO-depe ndent permeabilization of cell membranes.
Human cells put in place several protective mechanisms against PFTs such as changes in membranes receptors, membranes repair, activation of cell stress responses, and increased nutrient (e.g., glutamine) demand for immune cells metabolism (10,(41)(42)(43)(44)(45)(46). Results here reported contribute in defining the relevance of HSA in human physiology and pathophysiology that goes far behind its major role of regulator of fluid distribution in body compartments. Indeed, HSA is capable to bind and inhibit the hemolytic activity of the SLO toxin produced by GAS by binding to domain II. Mutagenesis experiments indicate a direct interaction between SLO and HSA, as the replacement of Leu305 and Phe374 residues located in the binding interface of HSA caused a decrease in the affinity for SLO. This non-canonical binding site seems to be specifically devoted to the recognition of toxins and proteins expressed by bacteria. Several Gram-positive bacterial species, including human pathogens, express surface proteins that interact with host proteins like HSA and IgG with high specificity and affinity   (24,47). In 1979, Kronvall and coworkers first described the binding of HSA to bacterial surface structures and found that groups A, C, and G streptococci specifically absorbed HSA from plasma (48). Subsequently, some strains of Finegoldia magna were also found to bind HSA. In the case of group C and G streptococci, protein G is responsible for HSA binding. Some isolates of F. magna bind HSA to their surface; the molecule responsible for the HSA:F. magna complexation is called poly (A)-binding protein (PAB) (49), which is the homologue of streptococci protein G (50,51). Protein G and PAB share the GA module whose exact function is unknown (50,51). The crystal structure of HSA in complex with the GA module of F. magna PAB highlighted that the interaction involves domain II of HSA (52). However, differently from the protective effects of HSA Propidium iodide (PI) staining, which is not permeant to live cells, allowed to detect dead cells. Cells were immediately analyzed and acquired using the LCS Leica confocal microscope (Leica Microsystems, Heidelberg, Germany).

A B
FIGURE 4 | Effect of HSA and BSA on the SLO toxin-mediated red blood cells (RBCs) hemolysis. RBCs were isolated from the human whole blood of four healthy donors and treated with (A) 0, 1, 2.5 and 5 U of activated SLO for 1 h at 37°C. (B) Considering that 1 U of activated SLO causes a 50% lysis of 50 mL of 2% human RBCs, we used 5 U of activated SLO to cause the 50% lysis of 190 mL of 2% human RBCs. RBCs were treated with the toxin for 1 h at 37°C, in the absence or presence of 1.0×10 −6 M to 1.0×10 −4 M HSA. The protective effect of 1.0×10 −4 M BSA was tested. As negative control, RBCs were incubated with PBS, in the absence of activated SLO. As positive control, RBCs were disrupted with 2% Triton X-100. The concentration of released hemoglobin was quantified by measuring the absorbance at 541 nm. Results were represented as the percentage of lysed RBCs (assuming as 100% the positive control) derived from three independent experiments ± SD (One-way ANOVA, ***P < 0.001 and ****P < 0.0001 compared with cells treated with 600 U of activated SLO in the absence of HSA).
toward C. difficile TcdA and TcdB (25,26) and S. pyogenes SLO toxins (data here reported), binding of HSA to the GA module of F. magna could provide growing bacteria with FAs and, possibly, other nutrients transported by HSA (52,53). Overall, the selective recognition of bacteria proteins and toxins through domain II of HSA represents a clear example of host-microbe adaptation at the molecular level. Our work demonstrates that HSA protects from SLO-induced cytotoxicity. Since this effect is the result of the extracellular binding between HSA and SLO, from a clinical point of view the demonstration of a neutralizing effect itself rather than the protective effect on different types of adopted cellular models is the most relevant result. In vivo studies clearly demonstrated that the inhibition of SLO activity was significantly associated with a disease severity reduction in necrotizing fasciitis, necrotizing myositis, bacteremia, and soft tissue infection models (54,55). Therefore, we hypothesize that HSA levels could affect streptococcal diseases severity.
In conclusion, here for the first time the interaction between HSA and SLO exotoxin has been reported. Even if further studies are needed to confirm in vivo interaction, we demonstrated that HSA is able to partially inhibit the cytolytic and hemolytic activity of SLO, which are the two main toxic effects of this extracellular streptococcal product (3)(4)(5). As SLO has a prominent role in the pathogenesis of severe invasive infection in animal models (7), the recognition of SLO by HSA may have a significant protective role in human serum and sustains the emerging hypothesis that HSA is a player in the innate immunity system.

DATA AVAILABILITY STATEMENT
All datasets generated for this study are included in the article/ Supplementary Material.

ETHICS STATEMENT
Ethical review and approval was not required for the study. The patients/participants provided their written informed consent to participate in this study.

AUTHOR CONTRIBUTIONS
ADM, GDS, GMV, LL, and PA performed and analyzed the biochemical assays. ADM, DM, GDS, GMV, and LL performed the in vitro experiments. LL performed the bioinformatic analyses. ADM, AG, PA, RL, and SDB conceived and designed the experiments. ADM, FM, and SDB wrote the paper. All authors contributed to the article and approved the submitted version.