Patients with S. dysgalactiae bacteraemia were identified through the Laboratory for Clinical Microbiology in the county of Skåne, Sweden between 2017 and 2018. Patients were included after informed and written consent. Acute sera were obtained <5 days after admission and at follow-up 6 weeks after admission convalescent sera were obtained. Epidemiological, clinical parameters and microbiological results for each patient were collected through the medical records.

Blood isolates of S. dysgalactiae were collected from the Laboratory for Clinical Microbiology, Lund University Hospital Sweden and cultured on blood agar plates in 37°C and 5% CO₂. The isolates were species determined utilising Microflex MALDI-TOF MS (Bruker, Bremen, Germany) with the direct transfer protocol (Bizzini et al., 2010) and the software FlexControl and MBT Compass 4.1 with reference database DB-8468 MSP 2019 (MALDI Biotyper; Bruker) where a score value ≥ 2.0 was needed for species determination. emm typing based on the sequence of the emm gene was performed as described at¹. The emm gene sequences were compared with those in the CDC emm sequence database. An isogenic mutant lacking the IgG-binding protein protein G (G45ΔproteinG) was used as a control for reducing non-antigenic binding (Nitsche-Schmitz et al., 2007). G45ΔproteinG was cultivated in Todd-Hewitt broth (TH), with a final concentration of 1 μg/ml erythromycin.

Fragments of the M protein, corresponding to the mature protein, from the two most common emm types in the study [stG2078: aa 42–514, (sequence ID WP_148848339.1) and stG62647 aa 42–457, (sequence ID WP_084917006.1)] were synthesised and ordered from GenScript. The genes were codon optimised for expression in Escherichia coli. The constructs, containing a C-terminal His-tags for simplified purification, were delivered in pET26 b vectors and expressed recombinantly in E. coli TUNER (DE3). Purification was conducted using Ni-affinity chromatography according to standard protocols at the Lund Protein Production Platform LP3.

Quantitative analysis of IgG antibodies was performed at the Department of Clinical Chemistry in Skåne, Sweden using turbidimetry. Antibodies toward IgG produce agglutinate together with the antigen. The absorbance was measured biochromatic at 340 and 700 nm, following routine clinical procedures.

IgG in acute and convalescent sera was quantified using enzyme-linked immunosorbent assay (ELISA). Plates were coated either with bacteria or recombinant M protein.

For coating with bacteria, the relevant isolate was cultured on blood agar plates at 37°C and 5% CO₂. Three or four colonies were inoculated in 20 mL coating buffer (1.69 g Na₂CO₃ and 2.94 g NaHCO₃ in 1 L H₂O, pH 9.6) mixed vigorously, and dispensed into a 96-well microtitre plate Nunc Maxisorp (ThermoFisher Scientific). Bacteria were allowed to bind to the plate overnight at 37°C, 5% CO₂, before removing the coating buffer and heat-killing the bacteria at 80°C for 15 min. Alternatively, recombinant M proteins of stG2078 and stG62647 were diluted in coating buffer to a concentration of 50 ng/mL and added into each individual well on the microtitre plate and the plate was incubated overnight at 4°C. The wells were blocked using 5% bovine serum albumin and 0.05% Tween20 in PBS (blocking buffer) and incubated at 37°C and 5% CO₂ for 1 h. Since S. dysgalactiae expresses protein G which binds IgG in a non-immune manner, serum, diluted 1:100 in blocking buffer, was preincubated with IdeS, (1:40 enzyme: substrate) at 37°C and 5% CO₂ for 50 min to generate F(ab′)₂ and Fc fragments of IgG (Rungelrath et al., 2017). The mixture was then added into each well of a microtitre plate coated as described above. The plate was incubated at 37°C and 5% CO₂ for 1 h and washed three times in PBST. Protein L with covalently linked HRP (GenScript), which specifically binds to the Fab-fragment of IgG, was then diluted 1:3000 (0.17 μg/ml) in blocking buffer, added to each well and the plate was incubated at 37°C and 5% CO₂ for 1 h. This was followed by three wash steps in PBS with 0.05% Tween20. ABTS 2% 2,2′-azino-di-(3-ethylbenzthiazoline sulfonic acid) dissolved 1:20 in a substrate buffer (21 g citric acid monohydrate and 17.8 g Na₂HPO₄ 2 H₂O in 1 L H₂O, pH 4.5) was mixed in combination with 30% H₂O₂ and incubated in the dark in room temperature for 15–30 min. The absorbance was read in iMark Microplate reader (Bio-Rad) at 415 nm.

An overnight culture of S. dysgalactiae bacteria was diluted in TH, 1:20, and grown to early mid-log phase (OD₆₂₀ 0.2) and diluted to approximately 2 × 10³ cfu/mL in TH. Patient serum (25 μl) was mixed with freshly obtained human heparinised whole blood (125 μl) from a healthy donor and preincubated with 1U β-lactamase (β-lactamase Blend, recombinant expressed in E. coli, L7920, and Sigma-Aldrich) to degrade antibiotics possibly present in the patient sera. The mixtures with or without IdeS (1:40 enzyme: substrate) were incubated at 37°C and 5% CO₂ for 20 min. Bacterial suspension (25 μl) was added to the mixture and incubation continued by gentle rotation at 37°C and 5% CO₂ as originally described by Lancefield (Lancefield, 1957). Aliquots of the mixture were taken at different times and plated in soft agar on TH-plates. Colony forming units (cfu) were counted after overnight incubation at 37°C and 5% CO₂. Paired acute and convalescent sera from patients were applied in the experiments. As a control, serum from the healthy donor of whole blood was utilised.

Phagocytosis assay was performed and analysed using the PAN-method as described (23). Briefly, bacteria were grown to exponential growth, heat-killed (80°C, 5 min) and stained with 5 μM Oregon Green 488-X succinimidyl ester (Invitrogen) and CypHer5E (General Electric) 7 μg/ml. Opsonisation occurred at 37°C for 30 min.

THP-1 cells (ATCC TIB-202, male) were maintained in Roswell Park Memorial Institute medium 1640 (Sigma-Aldrich) supplemented with 10% fetal bovine serum (Gibco) and 2 mM GlutaMAX (Life Technologies). The phagocytosis was performed with 100,000 cells in a final volume of 150 μl with multiplicity of prey (MOP) between 0 and 400, at 37°C or on ice as control for internalisation. Phagocytosis was haltered by putting the samples on ice for at least 15 min before data acquisition. Four independent experiments were performed.

Flow cytometric acquisition was performed using CytoFLEX (Beckman-Coulter, lasers: 488 nm, 638 nm, filters: 525/40, 660/10). At least 5,000 cells were acquired and then analysed using FlowJo version 10.2 (Tree Star), see Supplementary Material for detailed information on settings and gating strategy. THP-1 cells positive for Oregon Green-signal were defined as associating cells, and of those THP-1 cells, cells positive for CypHer5E-signal were defined as internalising cells.

Phagocytosis data were analysed using Prism 8.2.1 (GraphPad) using the PAN-template v 1.1 available at Nordenfelt Lab (zenodo version link). Normalisation for association was performed by interpolating the MFI corresponding to the MOP which evoked half of the maximal response (MOP₅₀). The fluorescent signal of a prey unit was determined by measuring the MFI of free bacteria by flow cytometry. The number of prey per phagocyte, PxP, could be quantified by dividing each signal (associated: Oregon Green, internalised: CypHer5E) with the corresponding prey unit signal at pH 5. Streptococci is not typically present as a single bacterium so a prey unit most likely represents a chain.

For categorical data Fisher’s exact test was applied and comparisons of continuous variables were performed utilising Mann–Whitney U test. For comparison of paired observations Wilcoxon matched pairs signed rank test was used. Significance was defined as a p value ≤ 0.05. Microsoft Excel 2016 (Microsoft Corporation) was used for data collection and GraphPad Prism, version 8.2.1 (GraphPad Software) was utilised for statistical assays. In phagocytosis assay the non-parametric paired Friedman test with Dunn’s multiple comparisons test was used to test if there was a significant (alpha 0.05) difference between the episodes in association and internalisation at MOP₅₀.

Nineteen patients were initially included of whom two patients were excluded due to inability to obtain convalescent sera. One patient deceased prior to convalescent serum was obtained. During the study period, 2017-2018, 129 cases of S. dysgalactiae bacteraemia occurred in the three hospitals (Helsingborg, Lund, and Malmö) where patients were enrolled. Median age of these patients was 76 years (IQR 69-87) and 55% were male.

Sixteen patients with S. dysgalactiae bacteraemia were finally included in the study. The most common emm types were stG62647 (n = 5; carrying group C antigen), stG2078 (n = 3), and stG480 (n = 2; carrying group G antigen; Table 1). Table 1 summarizes the clinical characteristics of each patient. The most common focus of infection was soft tissue (69%). One patient, denoted PW had three episodes of bacteraemia with the same emm type during 1 year (emm 2078).

Paired acute and convalescent sera from patients were tested for the presence of antibodies against S. dysgalactiae. To avoid the non-immune binding of IgG to protein G, a mutant S. dysgalactiae strain lacking protein G (G45ΔproteinG) was used to coat the plates. There were no significant differences in levels of absorbance toward G45ΔproteinG between acute and convalescent patient sera, [Wilcoxon matched-pairs signed-rank test, (p = 0.052)] (Figure 1A). Total levels of IgG in serum, however, were significantly higher in convalescent than in acute sera, [Wilcoxon matched-pairs signed-rank test, (p < 0.0001)] (Figure 1B). Figure 1C, displays absorbance levels from the ELISA (Figure 1A) divided by the concentration of total IgG (Figure 1B). Adjusting for total IgG-levels, acute sera displayed a stronger signal in ELISA compared to convalescent sera [Wilcoxon matched-pairs signed-rank test, (p = 0.008)].

Since the M protein is believed to be a major antigenic structure of S. dysgalactiae and is variable between isolates, we next tested the levels of antibodies in acute and convalescent sera against the patient-specific isolate. Using this experimental set-up, we observed a median increase of 0.18 in absorbance between acute and convalescent sera [Wilcoxon matched-pairs signed rank test, (p = 0.0006)] (Figure 2A). However, this difference was not observed when adjusting for total levels of IgG (Figure 2B).

To investigate the role of the M protein in the immune response to S. dysgalactiae, we analysed the binding of serum IgG to recombinant M proteins, representing the two major emm types in the study (e.g., stG2078 and stG62647), both in acute and convalescent sera from some of the patients (Table 2). To gain insight relating to cross-reactivity between M proteins, serum was pre-incubated with either stG2078, stG62647, or PBS (control) before the interaction between antibodies in serum and the immobilised M protein was measured. Antibody binding was often increased in the convalescent serum, as compared to the acute serum. Patients infected with emm types other than stG2078 and stG62647 showed no proof of major IgG reactivity with any of the two recombinant M proteins.

The patient, PW, had three episodes of bacteraemia with stG2078 during the study period and we had access to serum drawn before the second episode and after the third episode. A high level of antibody binding toward M2078 and M62647 was observed in both sera. In the serum drawn before the second episode this reactivity could be inhibited either by the addition of soluble M2078 or M62647 to the reaction (Table 2) indicating that this serum contained antibodies directed toward epitopes shared between the two M proteins. In the serum drawn after the third episode, however, the reactivity toward M2078 was inhibited by M2078 but not by M62647. This indicates that type-specific antibodies had been formed in this patient.

The paired sera from PW and the formation of type-specific antibodies were next tested for their ability to confer killing of the stG2078 from the same patient in non-immune blood from a healthy donor. Since the sera contained antibiotics, they were first treated with β-lactamase. Figure 3 summarises results from the bactericidal assay. Unexpectedly, more bacteria were retrieved after incubation of the blood with serum containing type-specific antibodies (convalescent serum) than after incubation with serum lacking such antibodies. Preincubation with IdeS, cleaving and inactivating IgG, resulted in increased levels of cfu/mL in both acute and convalescent serum.

Serum from the patient, PW, which showed seroconversion between the second and the third episode, was assessed with respect to phagocytosis with an in-depth flow cytometric assay that takes into account both phagocyte association and internalisation (de Neergaard et al., 2019). Figure 4A visualizes the association capacity of the phagocytes with the percentage of phagocytes interacting (adhered or internalised) with at least one bacterium on the y-axis and the MOP on the x-axis. An increase in association capacity results in a left shift of the curves. This can be quantified by comparing the MOP needed to reach 50% association (MOP₅₀), which resulted in a small decrease from 97 ± 14 (SD) in average to 70 ± 10 (SD) with a significant increase in phagocyte association (p = 0.04) between the first and third serum (Figure 4B). The same trend can be seen in the percentage of phagocytes internalising at least one prey (Figure 4C) and at MOP₅₀ there is a small significant increase (p = 0.04) from 14 ± 8.3% to 22 ± 8.6% (SD) internalisation between the first and third serum (Figure 4D). We also wanted to see if there was a difference once a phagocyte had internalised a bacterium. The number of prey associated per phagocyte decreases from on average 14 ± 3.1 to 11 ± 2.2 (SD; Supplementary Figure 1A) between the first and third serum and the number of prey internalised per phagocyte from 7.5 ± 1.6 to 7.0 ± 0.8 (SD; Supplementary Figure 1B). There was no significant difference (p = 0.07 association, p = 0.1 internalisation) between the sera, but the data indicate a decrease in association and internalisation ability of internalising phagocytes. Taken together with the population assessment above, this means that with the serum containing type-specific antibodies, more phagocytes associate with bacteria, leading to an overall small increase in phagocytosis, but on an individual level they do not become better at associating with or internalising their prey.