The S. pyogenes serotypes AP1, AP4, and AP12 were obtained from the WHO collaborating center for Reference and Research on Streptococci, Prague, Czech Republic. S. pneumoniae F23 was obtained from the Gothenburg Cooperation Lab and E. coli K12 from the Institute of Microbiology, Technical University of Braunschweig, Germany. For the experiments with S. aureus bacteria the strain Newman was used which is a coagulase- and clumping factor-positive strain (Boden and Flock, 1989). AP1 M-protein mutant MC25, with a non-covalently bound M1 protein was kindly provided by Prof. Mattias Collin, Department of Clinical Sciences, Lund Division of Infection Medicine, Lund University, Sweden. It should be noted that this strain lacks the ability to convert SpeB, a secreted streptococcal cysteine proteinase, into its enzymatically active form (Kihlberg et al., 1995). Because this strain does not generate a protein layer around the bacterial wall, this strain was not used when testing the effect of antimicrobials. All bacteria, with the exception of the E. coli K12 strain, were first cultured overnight on blood agar plates from glycerol stocks stored in -80°C cryotubes. The E. coli K12 strain was cultured on THY plates. For cultivation of MC25, kanamycin (75µg/ml) was plated directly on the blood agar plates and later added to the THY media in the liquid culture. All bacteria were then cultured in 10ml Todd-Hewitt broth supplemented with yeast extract (30g/l Todd-Hewitt broth, 10g/l yeast extract) at 37°C in a 5% CO₂ in 15ml cultivation screwable tubes.

Starting culture in THY-media was grown till OD_{600 =} 0.1 and then split and refilled with new media to enable equal growth of cultures to late exponential phase of OD₆₀₀ of 0.45. Bacteria were gently centrifuged at 1.000 x rcf for 10min and washed twice with 1 x PBS. After the last resuspending step, the OD₆₀₀ was adjusted to a starting CFU-count around 1 x 10⁵ bacteria/ml and centrifuged again. This time bacterial pellets were resuspended in assay buffer (150mM NaCl, 20mM Tris, pH 7.6) and then equally incubated in the same volume of 10% thawed prewarmed human plasma or in assay buffer. Tubes were placed horizontal in the incubator for 5 or 30min at 37°C, 5% CO₂. Bacteria were again gently centrifuged at 1.000 x rcf for 10min, washed twice with assay buffer and then used for further assays. Assay buffer-incubated bacteria were used in all experiments as negative controls. For the killing assays in solution 100µl of the bacterial solution, either buffer or plasma incubated, were mixed with 100µl of the testing solution.

Human plasma derived from human peripheral blood, was collected from healthy volunteers (IRB approved protocol Dnr 2017/728) in a 2.7ml vacutainer containing citrate (0.3ml 0.109 M sodium citrate, BD Vacutainer system). Vacutainer tubes with collected blood were inverted carefully to mix blood and anticoagulant. Samples were centrifugated immediately for 10min at 1300 x rcf, at room temperature. Plasma was aliquoted and frozen at -80°C and thawed for a maximum of two times.

For the transmission electron microscopy (TEM), bacterial solution (200µl), was incubated for different timepoints (5, 30, 60 and 120min). The samples were immediately mixed with 1ml EM-fix (2.5% glutaraldehyde in 0.2M Na-cacodylate buffer, pH 7.2) to preserve the formed surface structures. Samples were then left at room temperature for 24h and further processed for TEM. Embedding and negative staining was done by IQ-Platform, Lund University.

Assay buffer (500µl) was treated with cold atmospheric plasma (CAP) with the KINPen^® MED (Settings:30s, 2.5bar, 3l/min) in a 6-well plate. The KINPen^® MED was kept at a stable distance throughout. This distance was based on the length of an adapter for skin treatment added by the supplier. The CAP-buffer was used 5-10min after preparation. The bacteria were resuspended and incubated (5min) with CAP buffer (500µl) after washing. The bacterial survival assay was performed as mentioned below.

After the washing step, bacteria were resuspended in assay buffer. CAP incubation was followed by LL-37, calf thymus histones (CTH) or tetracycline incubation. Bacteria were resuspended in CAP-treated buffer as described above before adding other solutions. Final concentrations of CTH (0.1μg/ml), LL-37 (10μM) and tetracycline (30μg/ml) were added all in assay buffer. CTH was purchased from Merck Millipore (Darmstadt, Germany) and LL-37 trifluoroacetate salt (human) was purchased from Schafer-N (Copenhagen, Denmark) and both stored at -20°C. Tetracycline hydrochloride was purchased from Thermo Fisher and stored at 4°C in the dark. After the incubation time ended (30 or 60min), 100µl were taken from the reaction mixture and serial dilutions were performed in a 96-well plate followed by plating on THY plates in triplicates.

After preincubation and washing, bacterial solutions were mixed with fibrinogen in solution on a coverslip. Fibrin network formation was initiated by adding human thrombin (Merck Millipore). 25µl of the bacterial solution was mixed with 25µl of the fibrinogen stock solution (100mg/1700µl in sterile water, prewarmed for at least 24h at 37°C). The mixture was then added to rat tail collagen I coated coverslips (final concentration: 50µg/ml diluted with 0.01M HCl). The clot was gently resuspended with 200µl assay-buffer, containing testing substances. The covered clot was then incubated for 30 or 60min at 37°C, in a 5% CO₂ atmosphere. CLEM analysis involved the same procedure with the inclusion of indium titandioxid cover glasses coated with collagen I. Samples were prepared for CLEM by the IQ-Platform, Lund University.

LIVE/DEAD™ staining (LIVE/DEAD™ BacLight™ Bacterial Viability Kit from Thermo Fisher) was used to identify bacterial cells which have structural integrity deficiency in their cell wall enabling propidium iodide staining. Propidium iodide red stained cells were considered as dead. For a LIVE/DEAD™ staining the solution in the well was removed after the incubation time of 30min or 60min and 200µl assay buffer with 1/1 mixture of component A and B (total vol. of the mixture 1µl each) were added and incubated for 30min at 37°C, protected from light. Afterwards the wells where carefully washed three times with 300µl of fresh assay buffer and then incubated overnight in 4% PFA solution at room temperature to preserve the sample. All cells where stained by the DNA intercalating dye SYTO 9. Using a Nikon microscope (TI2 body) and an Andor Zyla 4.2 sCMOS camera, 10 z-stacks with a thickness of 3µm were taken from representative areas per coverslip.

Merged z-stacks images where further semi-automatic processed with ImageJ (1.52p 64-bit) using the find-maxima-function for red and the green fluorescence image analysis. For this purpose, images were converted to grey scale. For counting the prominence for maxima was set to 3 up to 4 for the TRIC-channel and to more than 10 for the FITC-channel. Counting was examined manually by zooming in and review counted signals by eye. Percentage killing was calculated by subtracting counts between the two channels, green (all) and red (dead/membrane ruptured) cells. For a better comparison of effects from LL-37 and CTH a background normalization was done for the killing observed after 30min and 60min of the only buffer and plasma incubated bacteria.

The RiboPure™-Bacteria Kit was used for the RNA-extraction. Briefly, bacteria were incubated as described before for the bacterial survival assay. After 0, 30 and 60min, samples were centrifuged at 1.000 x rcf for 10min and resuspend in RNAwiz and transferred to the kit tubes with Zirconia Beads. Samples were processed according to the Kit protocol. Extracted nucleic acids were measured with the NanoDrop™ 2000 spectrophotometer and genomic DNA was first eliminated before converting the remaining extracted RNA to cDNA. Both steps were done with the QuantiTect Reverse Transcription Kit, purchased from Qiagen (Hilden, Germany). Afterwards cDNA amount and quality were determined using the NanoDrop™ 2000 spectrophotometer. 300ng was used as a template for each Q-PCR master mix. Expression of virulence genes, M-protein, SpeB, HasB and DNA gyrase, Sortase A, FhuB, and HtrA of AP1 were determined. A modified PCR master mix was prepared, following the standard Thermo Fisher protocol ( Table S1 ). The following thermal cycling conditions were used in a QuantStudio™ 7 Flex Real-Time PCR Instrument: UNG incubation for 2min at 50°C, polymerase reaction for 10min at 92°C and denatured at 95°C for 1min with an annealing/extending time of 20sec at 60°C. The primers and probes ( Table 1 ) were designed based on published target sequences and by using the NCBI primer BLAST program along with the online qPCR Primer & Probe Design Tool from Eurofins (von Pawel-Rammingen et al., 2002). Expression fold change (EFC) was calculated by using the 2-ΔΔCt method (Akesson et al., 1996). DNA gyrase, a known housekeeping gene (Frick et al., 2011), was chosen as the housekeeping gene.

Statistical analyses were performed using Prism (8.3.0). The data were first tested for normal distribution. In case not all data sets followed a Gaussian distribution a Mann-Whitney-test was performed to evaluated, if there were significant differences between treatments. For the clot assay the buffer and plasma incubation with N=18 a Dunn’s multiple comparison was performed and for the LL-37 and CTH clot assays a Mann-Whitney test was used. In all other cases a one-way ANOVA (Tukey’s multiple comparisons test) was chosen. Rt-qPCR-experimental data were analyzed using a non-parametric Student’s t-test for significance, with no significance detected. The pre-formed clot experiments were done as biological duplicates with each completed in technical duplicate. All other experiments were based on technical quadruplicates and biological triplicates.

Previous work has shown that upon contact with plasma, group A streptococci are covered with a proteinous layer around their outer surface (Naudin et al., 2015). To test whether the formation of the proteinous layer is part of an immune response or a species-specific bacterial escape mechanism, we incubated human plasma with Streptococcus pneumoniae, Staphylococcus aureus, or Escherichia coli bacteria and analyzed whether similar shell formation can be detected as seen with S. pyogenes bacteria. To this end, the four species were treated with plasma for a short (5min) or a longer (30min) incubation time. This was followed by a washing step and subsequent examination by transmission electron microscopy (TEM). Figures 1A–H shows that only S. pyogenes bacteria are able to surround themselves with a protein layer, while all other species lack this ability. These results suggest that the formation of the layer is driven by S. pyogenes rather than host immune response.

In the next series of experiments, we explored the proteinous shell morphology of S. pyogenes AP1 bacteria after their incubation with human plasma. In line with previous observations, we found in the absence of plasma, the bacteria are covered with hair-like fibrils built up by M1 protein and protein H (Collin and Olsen, 2000; Herwald et al., 2003) ( Figures 1I–L ). When incubated with human plasma, the newly formed protein cover was attached to the extension of the hair-like structures, suggesting that the layer is bound to the outer bacterial surface proteins with a thickness between 40nm and 80nm ( Figures 1K, L ).

When performing a time kinetic study we found that S. pyogenes AP1 bacteria were already covered 5 minutes after being added to human plasma ( Figures 1M, N ). No major changes of thickness of the proteinous layer were seen upon longer incubation times ( Figures 1M, O, Q, S black arrows), except that the diameter size decreased at some septum formation sites ( Figures 1M, O, Q, S white arrows). At lower magnification ( Figures 1N–T ) the images also show that some bacteria have lost their cytosolic content ( Figures 1M, N, P, R, T marked with a black star), suggesting that not all bacteria can survive within the plasma protein shell. Subjecting the S. pyogenes strains AP4 (an M4 serotype, Figures 2E–H ) and AP12 (an M12 serotype, Figures 2I–L ) to the same experimental conditions, a similar shell formation was noted, though less dead bacteria were seen. An isogenic AP1 mutant strain (MC25) was employed as a control, that lacks surface-bound M1 protein (Kihlberg et al., 1995). As depicted in Figures 2M–P , plasma proteins did not completely cover the surface of the MC25 strain and a gap formation was not noted as seen for the other streptococcal strains. Also the protein layer did not completely cover the bacterial surface of the MC25 strain ( Figures 2N–P black arrows).

The TEM results suggest that streptococci of the AP1 strain ( Figures 2A–D ) are more susceptible to bacterial killing, when embedded in the plasma protein shell compared to the other streptococcal strains tested. This observation prompted us to examine whether shell formation can influence bacterial viability of streptococcal strains AP1, AP4, AP12, and MC25, respectively. Figure 3A shows that all S. pyogenes wild type strains tested were able to grow and proliferate when surrounded with a plasma protein layer (with the exception of the AP1 strain where survival after 60min was decreased compared to 30min). In contrast, the isogeneic AP1 mutant MC25, lacking surface-bound M-protein on its surface, was not able to form a layer that covered the entire bacterial surface ( Figures 2M–P ). Though the growth rates were for all strains statistically significant ( Figure 3A ), bacterial proliferation and the relative survival ( Figure 3A ) were considerably lower for the AP1 strain compared to the other strains, which is in line with the findings of the TEM experiments ( Figures 2A–P ). E. coli K12, a non-pathogenic Gram-negative enterobacterium was significant eradicated after the plasma preincubation compared with the buffer incubation. The S. pneumoniae strain F23, which was isolated from a meningitis patient in Gothenburg showed similar growth behavior after the plasma preincubation as AP4 and AP12.

Previous results have shown that activation of the coagulation system and subsequent formation of a fibrin network not only crosslinks the bacteria within a formed fibrin clot (Loof et al., 2011), but also leads to the generation of fibrinogen-derived antimicrobial activity (Pahlman et al., 2013). Both findings emphasize the importance of the coagulation system in the early innate immune system. Considering this role, the next experiments were set up to study whether the formation of the plasma protein shell around the streptococcal surface is a bacterial defense mechanism that can rescue immobilized streptococci of the AP1 strain from an otherwise lethal immune attack. To prove this hypothesis bacteria were preincubated in the presence or absence of human plasma. Bacteria were then added to a fibrinogen solution directly on a collagen I coated coverslip. The mixture was then treated with thrombin to allow the generation of a fibrin network for 5min before the clot was covered with buffer including antimicrobials (see also workflow shown in Figure 3B ).

After 30min and 60min incubation bacterial survival was microscopically determined using a LIVE/DEAD™ staining kit. Figure 3C depicts that under these experimental conditions bacterial survival is enhanced for the 30min, but decreased at the 60min time point, when AP1 bacteria were preincubated with plasma. Figure 3C also show that inside a fibrin clot, the plasma shield prevents the streptococci from killing by LL-37 or CTH. For LL-37 at the 60min timepoint, more bacteria survived after buffer pre-incubation than after the plasma incubation ( Figure 3C ). CTH-treated samples had at both time points significantly increased survival for the plasma incubated samples ( Figure 3C ).

Having shown that streptococci can proliferate and escape the host immune response, when covered with a plasma protein layer, we next tested whether the formed shell will also protect the bacteria against other antimicrobial peptides. To this end, the three strains (AP1, AP4, and AP12) were incubated with buffer or human plasma for 30min and exposed for 30min and 60min to sublethal doses of LL-37 and calf thymus histones (CTH). As shown in Figures 4A–C the treatment with LL-37 led to a decrease in survival of all three strains, when the bacteria were buffer preincubated, while the preincubation with human plasma made the bacteria more resistant to the LL-37 attack with exception of the 30min time point for AP12. The protective effect was most significant in the streptococcal AP4 strain ( Figure 4B ), whereas the differences in survival of the streptococcal AP12 strain was, though statistically significant, less apparent ( Figure 4C ).

Like LL-37, also histones are considered as endogenous antimicrobial peptides. They are released for example from necrotic cells and have been shown to directly kill bacteria, fungi, and other parasites (Hoeksema et al., 2016). As depicted in Figure 4A–C , when CTH were tested under the same experimental conditions, similar findings were obtained as seen for LL-37. Together these data show that the shield formed by plasma proteins protects the bacteria from an endogenous antibacterial immune response.

Our previous results revealed that the plasma protein shield protects streptococci from an innate immune attack caused by endogenous antimicrobials. In the next series of experiments, we examined whether the activity of exogenous antimicrobial substances can additionally be abolished by the same mechanism. Notably, penicillin is still the drug of choice to treat S. pyogenes infections in hospitals. However, in cases of allergic reactions towards penicillin, antibiotics such as tetracycline, macrolides, and lincosamides are used as replacement therapy. While streptococcal resistances towards penicillin do not constitute a clinical problem, resistances to tetracycline and the other mentioned antibiotics have started to emerge (Sanson et al., 2019). Considering this aspect, we decided to focus on tetracycline throughout the remaining part of this study. Notably, tetracycline is below 0.5 kDa in size which may allow the antibiotic to penetrate the shielding layer. Thus, once it has passed the plasma layer and has been translocated inside the bacterium, the antibiotic should interfere with ribosomal translation processes.

As depicted in Figure 5A , our experiments show that in contrast to LL-37 and CTH the plasma protein shield was not able to block the effect of sublethal doses of tetracycline. Thus, no significant differences in streptococcal proliferation were noted under these experimental conditions. In search for alternatives to traditional antibiotic treatment, cold atmospheric plasma (CAP) has recently attracted considerable attention (Brany et al., 2020). CAP is considered the fourth state of matter apart from solids, liquids, and gases. It is generated by applying an electric field to inert gases, typically helium or argon (Heinlin et al., 2011). During this process the gas becomes ionized and can build reactive radicals with a broad antibacterial activity (Brany et al., 2020). When applying sublethal doses of CAP-treated buffer instead of tetracycline to the bacteria, streptococcal survival was significantly decreased after 30min and even further decreased after 60 min ( Figure 5A ). At both time points survival was increased when the bacteria were covered with a plasma protein layer, suggesting that the shield caused a delay in antimicrobial activity of CAP-treated plasma.

When using a combined approach starting with a CAP treatment followed by the addition of LL-37 or tetracycline, we observed that CAP modified the two antimicrobials differently. Figure 5B show that a combined CAP and LL-37 treatment significantly decreased bacterial survival. On the other hand, when a combined CAP and tetracycline treatment was employed, bacteria survival increased, suggesting that CAP limited tetracycline antimicrobial activity ( Figure 5C ).

Our previous results demonstrated that plasma protein shield formation allows the bacteria to proliferate in an otherwise hostile environment. These findings prompted us to investigate whether the gene regulation of six selected streptococcal virulence factors or proteins involved in metabolism are affected when bacteria were treated with tetracycline, CAP or in combination with both ( Table 1 ). As tetracycline inhibits bacterial proteins synthesis by blocking the interaction between aminoacyl-tRNA with the bacterial ribosome (Chopra and Roberts, 2001), the transcription of genes should not be affected. Subsequent RT-qPCR show that the gene-regulation of M1 protein was not influenced when the bacteria were incubated with plasma followed by a treatment with tetracycline or a combination with tetracycline and CAP ( Figure 6A ). Other virulence factors, however, such as the hyaluronan synthase (HasB) and the streptococcal cysteine proteinase (SpeB) were found upregulated, especially when applied in combination ( Figures 6D, E ). Furthermore, an upregulation of the ferrichrome ABC transporter (FhuB, Figure 6C ) was noted, while the genes for the transpeptidases sortase A (SortA) and the putative serine protease HtrA were only up regulated when treated with tetracycline. When a combined treatment was applied, both genes were initially (after 30min) upregulated followed by a downregulation (60min) as depicted in Figures 6B, F . Together these data show that different antimicrobial treatments trigger distinct gene regulation within the proteinous layer, though more analysis is required to unravel the molecular mechanisms more precisely.