Proteome analysis of PBMCs challenged with biofilms

To investigate the interaction of circulating immune cells with biofilms structures, we utilized mass spectrometry coupled with liquid chromatography (LC‒MS/MS). This approach allowed us to evaluate secretome alterations in PBMCs exposed to live plankton or biofilms of S. aureus.

As we reported in our previous study, we identified 1,666 ± 162 differentially expressed proteins (DEPs) in the biofilms-challenged group and 1,305 ± 227 in the untreated control group²². Notably, this analysis considered only proteins with a molecular weight greater than 10 kDa because of the filtration step. Thus, chemokines, such as those in the CXCL10 subfamily, which are between 8 and 10 kDa, were not captured. On the basis of the Benjamin‒Hochberg adjusted p value (known as q value), proteins with a q value < 0.05 and an abundance fold change (FC) > 2 compared with those in the control groups were identified as having significantly different abundances (Fig. 1A). In PBMCs challenged with biofilms, 321 proteins with significantly higher or lower abundances in cell culture supernatants were identified (for a detailed list of the proteins, see the Supplementary Material, Table S1).

A Volcano plot (-logq versus logFC) showing the difference in the abundance of secreted proteins in PBMCs exposed to biofilms (right). Proteins with a twofold higher (logFC > 1) or lower (logFC < -1) abundance (indicated by the vertical black lines) and BH-corrected p values < 0.05 (indicated by the horizontal red lines) are highlighted (blue and orange, respectively). The significantly differentially abundant proteins were further explored for GO terms and KEGG pathway analysis and visualized. B Protein‒protein interaction (PPI) network of proteins differentially secreted by PBMCs upon exposure to biofilms (right). The gene-concept network depicts the linkages of genes and biological concepts (e.g., GO terms or KEGG pathways) as a network. The gene-concept network was constructed to visualize the most relevant nodes engaging in hubs. Edges confidence correlates with the strength of the PPI.

Following enrichment analysis²³ of DEPs in the biofilms-challenged samples, 44 edges were visualized using a gene concept network plot. The gene-concept network depicts the linkages of genes and biological concepts (e.g., Gene Ontology (GO) terms or KEGG pathways) as a network²⁴. For the secreted proteins associated with biofilms challenge in PBMCs (Fig. 1B), several highly connected protein nodes (hubs) were identified. Among these hubs, two were significantly associated with the biological process of chemokine production regulation and neutrophil chemotaxis, as indicated by a significant q value for the corresponding GO term.

Chemokine secretion of PBMCs challenged with S. aureus biofilms

CXCL10-like cytokines have been previously shown to directly bind to the S. aureus cell, with CXCL-9 binding to the cell membrane and CXCL10 to both the cell membrane and the cell wall¹⁸. We observed the chemotaxis pattern in the proteomics experiment, while we could not capture chemokines such as those in the CXCL10 subfamily, which are smaller than 10 kDa. Therefore, we investigated whether the secretion of CXCL10 subfamily chemokines of PBMC was similarly induced by planktonic S. aureus bacteria and their biofilms. The PBMCs were challenged for 24 h with planktonic S. aureus suspensions or with one-day-old S. aureus biofilms. The time-dependent secretion of the chemokines CXCL9, CXCL10, and CXCL11 was assessed at 2, 4, 8, and 24 h using a bead-based multiplex flow cytometry assay and compared to that of unchallenged PBMCs cultured under the same conditions (control group) (Fig. 2).

A CXCL9, B CXCL10 and C CXCL11. The values are presented as the means and standard errors of the means (SEMs) of 6 independent experiments.

In contrast to CXCL9 and CXCL11, which were secreted at very low levels by untreated PBMCs, CXCL10 secretion increased after 8 h and remained stable until 24 h at a level of 72.26 ± 16.57 pg/mL (Fig. 2). When exposed to S. aureus, independent of the planktonic or biofilms stage, the secretion of all three chemokines increased. Considering the FC in secretion under stimulation compared with that in untreated PBMCs, both biofilms and planktonic S. aureus resulted in similar increases in CXCL9 levels (Fig. 2A), whereas CXCL11 was more strongly stimulated by planktonic bacteria within the first 8 h (Fig. 2C). CXCL10 secretion was greater in the first few hours of stimulation by the biofilms compared to stimulation by the planktonic bacteria, but due to the stimulus-independent increase in the CXCL10 levels in the untreated controls, the stimulus-dependent CXCL10 levels decreased and evened out (Fig. 2B).

These results indicate that the secretion of CXCL10 is stimulated relatively strongly by S. aureus biofilms, although the effect cannot be easily differentiated from the stress caused by planktonic S. aureus, which also triggers the secretion of chemokines. Nevertheless, we next asked whether a specific component of the biofilms plays a role in this effect. Since PNAG is one of the most important biofilms matrix components, we determined how this polysaccharide compound impacts the secretion of CXCL10 chemokine family members in comparison to other S. aureus virulence factors such as SpA, the most abundant and highly conserved virulence factor of S. aureus, which is a cell wall-associated protein (but can also be secreted) that binds to immunoglobulins²⁵.

Effects of PNAG on chemokine secretion in PBMCs

Despite chemical differences among cell wall carbohydrates, such as the varying glycosidic bonds in PGN and the carbohydrate moieties of teichoic acids, they all contain N-acetylglucosamine (GlcNAc), which also forms the homopolymer PNAG. To assess whether PNAG, a component of the biofilms matrix, specifically influences chemokine secretion, we measured CXCL10 levels secreted by PBMCs following a 24-hour exposure to PNAG or the S. aureus virulence protein-A. These results were then compared to a control group of unchallenged PBMCs cultured under identical conditions.

The CXCL10 level was visibly increased by both compounds, but significance was only achieved when the PBMCs were treated with PNAG but not with SpA (i.e., PNAG in comparison with the untreated control (Fig. 3A) or compared with the SpA-challenged condition). We also examined the secretion levels of the other CXCL10 family members, CXCL9 and CXCL11, in PBMCs treated with PNAG or SpA and compared them to those in unchallenged PBMCs cultured under the same conditions. While both CXCL9 and CXCL11 secretion levels were increased in PBMCs under these treatments, significant differences were observed in the PNAG-treated group compared with the untreated control group (data not shown). Although CXCL11 secretion was significantly greater than that in the SpA-treated group, the secretion level of CXCL11 was generally lower.

A by PBMCs versus the untreated control. B Increase in chemokine secretion as a percentage of treated and untreated PBMCs (dashed line). The values are presented as the means and standard deviations (SDs) of six independent experiments. Asterisks indicate significant differences as p values * ≤ 0.05 ** ≤ 0.01 and *** ≤ 0.001. Abbreviations: PNAG: poly N-acetylglucosamine, SpA: S. aureus virulence protein A.

A comparison of the ratios of CXCL10 family chemokines between PNAG- or SpA-treated PBMCs and untreated controls (Fig. 3B) confirmed that CXCL10 secretion was preferentially induced by PNAG. To determine whether the observed CXCL10 secretion is specifically induced by PNAG or more generally by carbohydrate-containing components of the bacterial cell, PBMCs were treated for 24 h with staphylococcal PGN, the main polymer of the cell wall. No significant change in CXCL10 levels was observed compared to the untreated control (Supplementary Material, Fig. S2). This suggests that PNAG is the driver of the observed CXCL10 induction.

These results indicate that the matrix-derived product PNAG significantly affects immune cell chemotaxis by increasing the production of the chemokine CXCL10. CXCL10 was stimulated with a clear predominance by PNAG. CXCL10, which is a ligand of the CXCR3 axis, primarily regulates immune cell extravasation, differentiation, and activation, promoting the recruitment of different immune cells, such as cytotoxic lymphocytes, natural killer cells and mucosal-associated invariant cells²⁶.

Transcriptome analysis of PBMCs challenged with PNAG

Three possible signaling pathways can be controlled by MAMPs in immune cells: i) TLRs, ii) NLRs and iii) CLRs. CLRs specialize in recognizing polysaccharide residues or motifs, such as β-glucans, which are components of the fungal cell wall²⁷. β-glucan in fungi is a linear polysaccharide of glucose, whereas PNAG is a polysaccharide composed of polymeric linked N-acetylglucosamine that can be produced in some bacteria, such as S. aureus^28,29. We aimed to investigate whether PNAG, similar to β-glucan in fungi, can stimulate the expression of this receptor and related signaling processes.

PBMC were incubated for 24 h in the presence of PNAG, and the mRNAs of six genes encoding Dectin-1, C-type lectin domain family 7 member A, Syk as a signaling adapter, caspase recruitment domain-containing protein 9 (CARD9) signaling adapter mediating the inflammatory response for C-lectin pattern recognition, protein kinase C-delta (PKCδ), nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB), and CXCL10 were transcribed into complementary DNA (cDNA) using reverse transcription. Quantification via the TaqMan qPCR method was normalized to the expression of the 18S rRNA gene³⁰. The expression level of the 18S rRNA gene ranged from 7.91 ± 0.22 to 8.92 ± 0.14 cycles (means and standard deviations (SDs)), with no significant difference between the groups (Supplementary Material, Fig. S1), confirming its suitability as a reference gene in this experimental setup. The mean ΔC_t value of the control group was subtracted from all individual values, including the control group values, to visualize the variation within this group. Additionally, the 2^-ΔΔCt values of all the groups were calculated to illustrate the expression patterns of all the target genes (Fig. 4).

The values are presented as the means and standard errors (SE) of 6 independent experiments. Significance is indicated as follows: * p < 0.05, ** p ≤ 0.01, **** p ≤ 0.01. Abbreviations: PNAG: poly N-acetylglucosamine, SpA: S. aureus virulence protein A.

TaqMan qPCR confirmed the results observed for cytokine protein secretion, revealing a greater than 10-fold increase in the mRNA expression of CXCL10 after PNAG treatment compared with that in the SpA or control conditions. When the effects of PNAG and SpA on the expression of the Dectin-1-Syk-CARD9 signaling pathway were compared, a specific and large increase (FC) of 4.81 ± 0.69) in NF-κB expression by PNAG was detected upon PNAG challenge. Compared with the control, PNAG also significantly elevated the expression of Dectin-1 (FC 2.12 ± 0.53), CARD9 (FC 2.83 ± 0.55), and Syk (FC 2.46 ± 0.54). PKCδ expression remained unchanged compared with that in the untreated control.

Given the observed PNAG-mediated signaling, we assessed whether S. aureus PGN could also activate this pathway. PBMCs were treated with PGN for 24 h and the same targeted transcriptomics was measured. Transcriptomic analysis showed a significant upregulation of NF-κB (FC 2.1 ± 0.69) but no significant changes in the other four target mRNAs examined (Supplementary Material, Fig. S3A). Pretreatment of PBMCs with an anti Dectin-1 blocking antibody did not abrogate PGN induced NF-κB upregulation (Supplementary Material, Fig. S3B), suggesting that PGN activates NF-κB independently of Dectin-1. We further investigated whether PGN induced NF-κB activation affects CXCL10 mRNA expression. We observed that PGN treatment did not induce CXCL10 expression (Supplementary Material, Fig. S3C), indicating that the PGN activated NF-κB does not converge on CXCL10 expression.

NF-κB is a master regulator of the inflammatory response to pathogens, and its expression can be stimulated through a variety of pathways involving diverse pathogen recognition receptors and chemokines, including the Dectin-1-Syk- pathway³¹. Thus, these results suggest that the major staphylococcal biofilms matrix PNAG stimulates NF-κB expression via this pathway. Moreover, the expression of the essential signaling components of the Dectin-1-Syk-CARD9 cascade was also increased in the presence of PNAG. Dectin-1 expression is activated by various cytokines and chemokines¹¹, but how this process is related to PNAG is still unclear³². As shown above, PNAG has a significant effect on the proteome of PBMCs and influences the expression of chemokines, thereby modulating immune chemotaxis.

Chemokine secretion in PBMCs after dectin-1 blockade or SyK inhibition

We blocked Dectin-1 with a blocking antibody and then measured the secretion levels of the chemokine CXCL10 by PBMCs in response to a 24 h treatment with the S. aureus biofilms component PNAG or SpA and compared these levels with those of unchallenged PBMCs cultured under the same conditions. Compared with that in PBMCs without antibody blockade under the same treatment, the CXCL10 level in PBMCs under PNAG treatment significantly decreased after antibody blockade against Dectin-1; however, this difference was not significant for SpA-treated Dectin-1-blocked PBMCs compared with PBMCs without blockade antibody (Fig. 5A). The level of CXCL10 secretion was altered in unchallenged PBMCs after treatment with an antibody blockade against Dectin-1, but the difference was not significant (Supplementary Material, Fig. S4A).

A Secretion of CXCL10 with/without antibody blockade of Dectin-1 in PBMCs treated with PNAG or SpA. B Levels of secreted CXCL10 in PBMCs with/without inhibition of the protein tyrosine kinase SyK. The values are presented as the means and standard deviations (SDs) of three independent experiments. Asterisks indicate significant differences as p values * ≤0.05 ** ≤0.01 and *** ≤0.001. Abbreviations: AD1_PNAG or SpA: Antibody-blocked Dectin-1 PBMCs treated with PNAG or SpA, ASyK_: SyK inhibited PBMCs treated with PNAG or SpA, PNAG or SpA: PBMCs treated with PNAG or SpA.

Afterward, we inhibited protein tyrosine kinase Syk in PBMCs via piceatannol, followed by a 24-hour challenge with either PNAG or SpA. We then measured the CXCL10 secretion levels in these treated PBMCs and compared them to those in unchallenged PBMCs cultured under the same conditions. The results revealed a significant decrease in CXCL10 secretion in Syk-inhibited PBMCs treated with PNAG compared with that in PBMCs without Syk inhibition under the same treatment (Fig. 5B). The secretion of CXCL10 by PBMCs tended to increase in the SpA-treated group but was not significant. Similar to dectin-1 antibody blockade, the levels of CXCL10 secretion in unchallenged PBMCs after piceatannol inhibition did not change significantly compared to those not treated with piceatannol (Supplementary Material, Fig. S4B).

This finding suggests that Dectin-1 and SyK signaling adapters may play a direct role in the immune chemotaxis induced by the biofilms matrix-derived product PNAG on immune cells during chronic infection. As discussed in the previous section, PNAG predominantly induces the secretion of CXCL10 through the Dectin-1 receptor and its corresponding signaling pathway. We observed that this predominance of CXCL10 is a distinct feature of PNAG, involving Dectin-1 and its corresponding signaling pathway. The CXCL10/CXCR3 axis is involved in regulating immune cell chemotaxis, leading to the infiltration of specific immune cell subtypes, such as T cells, which may aid in the clearance of pathogens.

Proteome analysis of monocytes treated with PNAG

We detected a secretome change in isolated monocytes exposed to PNAG and identified 1,447 ± 207 or SpA 1,353 ± 201 DEPs in comparison with the untreated control group. We further visualized the differential proteomic results between the two treatment groups on the basis of the Benjamini‒Hochberg adjusted p value (q value < 0.05) and an abundance FC > 2 as a volcano plot (Fig. 6A). In the secretome of monocytes exposed to PNAG or SpA, 24 protein groups were detected with a significant differential abundance, with 16 proteins higher (N = 16, red) in the PNAG group and eight higher in the SpA group (for a detailed list of the proteins, see Supplementary Material, Table S2).

A Volcano plot (-log_q versus log_FC) showing the difference in the abundance of secreted proteins in monocytes exposed to PNAG. Proteins with significantly differential secretion expressed as q values < 0.05 (indicated by the horizontal red line) and twofold higher (log_FC > 1, red) or lower (log_FC < -1 blue) secretion ₍indicated by the vertical red lines) are highlighted. B Gene-Concept network plot of proteins differentially secreted by monocytes upon exposure to PNAG. The gene-concept network was constructed to visualize the most relevant nodes engaging in hubs. Edges confidence correlates with the strength of the PPI. Abbreviations: PNAG: poly N-acetylglucosamine, SpA: S. aureus virulence protein A.

The significantly DEPs were further explored by GO term analysis and visualized as a Gene-Concept network (Fig. 6B)^23,24. A total of 33 edges were identified with two major hubs and many poorly connected nodes under the PNAG challenge. Cluster one was composed of 15 proteins involved in cytokine activity and regulation, leukocyte chemotaxis, and the inflammatory response. The proteins with the most significant and greatest increase in secretion were CCL4, CCL3, CXCL8, tumor necrosis factor α, and IL-1β. Within this cluster, three downregulated proteins were also involved: i) tetratricopeptide domain 25 (TTC25), a component of the outer dynein arm-docking complex subunit 4 that is involved in ciliary and flagellar motility, whose ciliary dysfunction is often associated with recurrent respiratory infections^33,34; ii) NOTCH3, which plays a role in the activation of NF-κB³⁵; and iii) calreticulin (CALR), a calcium-binding chaperone involved in the assembly of the endoplasmic reticulum (ER)³⁶. The second cluster was related to stress and defense responses but contained primary proteins downregulated in the presence of PNAG (the hemoglobin subunits HBB and HBA and orosomucoid 1 protein (ORM1)) (Fig. 6B).

Although our primary focus was on CXCL10-family chemokines, their small size³⁷ made them challenging to capture during LC‒MS/MS analysis, possibly due to filtration during sample preparation. Nonetheless, these results confirmed the occurrence of chemotaxis-related changes in the monocyte proteome. This observation suggests that PNAG plays a role in initiating chemotaxis, thereby recruiting immune cells to the site of infection.

Effect of PNAG on monocyte phenotypes

We aimed to further investigate how the phenotype of monocytes, the main modulators of the cytokine milieu at the site of infection^38,39, changes upon interaction with PNAG and SpA. We treated isolated monocytes with PNAG or SpA followed by 24 h of incubation and observed potential changes in monocyte phenotypes via fluorescence-based staining with monoclonal antibodies against CD14 (a cell surface marker) and intracellular CXCL10 (Fig. 7).

A The absolute cell count of CXCL10⁺ monocytes evaluated under PNAG- or SpA-treated conditions compared with the untreated control. The values are presented as the means and standard deviations (SDs) of three independent experiments. **P ≤ 0.01, ***P ≤ 0.001. B Dot plot representing the frequency of specific monocytes treated with PNAG or SpA compared with the untreated control. The X-axes correspond to the CD14+ receptor, and the Y-axes correspond to the CXCL10 signals. Abbreviations: PNAG: poly N-acetylglucosamine, SpA: S. aureus virulence protein A.

Compared with that of the untreated control, the number of CXCL10-expressing monocytes remained unchanged after SpA treatment (Fig. 7A), whereas the presence of PNAG led to a significantly increased number of CXCL10-expressing monocytes. Flow cytometric analysis revealed that up to 12.2% of the CD14⁺ CXCL10+ monocytes in the untreated control group and 17.5% of the SpA-treated monocytes were positive, whereas 39.7% of the PNAG-treated monocytes were positive in some experiments (Fig. 7B).

Flow cytometry analysis confirmed that PNAG induces the formation of CD14⁺CXCL10⁺ monocytes that migrate to the site of infection and trigger the innate immune response against S. aureus. The specific process through which CD14+CXCL10+ monocytes initiate the innate immune response remains elusive. However, it is possible that this process leads to the activation of TLRs or other PRRs and the subsequent release of proinflammatory cytokines.

Impact of CXCL10 on the growth and biofilms formation of S. aureus

On the basis of these results, we focused on CXCL10 and investigated the effect of CXCL10 on S. aureus. We added recombinant CXCL10 to a planktonic suspension of GFP-labeled S. aureus and assessed the increase in viable planktonic bacteria and biofilms mass up to 24 h at intervals of 30 min by using a heatable plate reader with orbital shaking (planktonic stage) or confocal laser scanning microscopy (CLSM, biofilms stage). The optical density at 600 nm was assessed for the planktonic bacteria. The biomass of the biofilms was quantified for images with the highest density (just above the bottom of the glass) as the sum of the fluorescence signal intensity (see the Materials and Methods section for details) and as viable bacteria (colony forming units (CFU)/mL) after microdilution and plating on agar.

CXCL10 had a weak inhibitory effect on planktonic growth, as observed for 12 h at 500 nM CXCL10 (Supplementary Material, Fig. S5A). The growth rate of planktonic was not significantly influenced by CXCL10, but the culture reached the stationary phase at a lower OD₆₀₀ than did the lower concentrations and the untreated control. Interestingly, after 24 h, the effect was reversed, and a decrease in the OD₆₀₀ was evident at 250 nM compared with 500 nM CXCL10 (Supplementary Material, Fig. S5B), while a concentration of 100 nM did not significantly change the OD₆₀₀. The increased density observed at higher CXCL10 concentrations likely reflects the development of biofilms with enhanced resistance to stress. This induction of biofilms development by CXCL10 challenge may involve specific signaling pathways leading to increased matrix production or altered bacterial aggregation.

To assess the influence of CXCL10 on biofilms formation, we quantified the biofilms mass over time using the constitutive GFP-labeled bacteria. We observed that treatment of planktonic bacterial with 500 nM CXCL10 resulted in faster initial attach to the glass surface (Supplementary Material, Fig. S5 A-B) and formation of bacterial aggregates which leading to higher GFP intensities than those in the absence of CXCL10 (Fig. 8C). To prove that the observed effect of CXCL10 is specific we treated the bacteria with heat-inactivated CXCL10 or phosphate-buffered saline (PBS). We found that neither heat-inactivated CXCL10 nor PBS (Supplementary Material, Fig. S5C, D) had a discernible effect on the initial attachment phase of planktonic bacteria to the glass surface. This lack of effect suggests that the previously observed enhancement of early biofilm formation is specifically mediated by the active form of CXCL10. The visualization of biofilm mass clearly increased in the presence of CXCL10 (Supplementary Material, Fig. S5B) compared with that of other conditions within 24 h. In the presence of CXCL10, increased bacterial aggregation was observed after 1 h compared with that of the untreated S. aureus. After that point, bacterial aggregation was less clearly recognizable, but the biofilms mass increased visibly in the presence of CXCL10.

A Growth curves of planktonic S. aureus recorded for 12 h at 37 °C in a 96-well plate and nonlinearly fitted. B Optical densities of the planktonic cultures after 24 h. C Relative GFP intensity of the biofilms layer formed on the glass coverslip over time. D Viable S. aureus (CFU/mL) fraction of the 24 h mature biofilms before and after treatment for an additional 24 h with different concentrations of CXCL10 or the untreated control for 48 h. Values are presented as the means and standard deviations (SDs) of three independent experiments. Asterisks indicate significant differences as p values * ≤ 0.05, *** ≤ 0.001, **** ≤ 0.001.

We further investigated the effect of CXCL10 on mature biofilm growth by quantifying CFUs after treatment with 500 nM and 1 µM CXCL10 (Fig. 8D) for 24 h. We increased the CXCL10 concentration because of the greater tolerance of the biofilms to chemical treatments. Prior to treatment, the mature biofilms contained approximately 1.2 × 10⁸ CFU/mL. After further incubation for 24 h (untreated controls), the biofilms reached 3.4 × 10⁸ CFU/mL. The concentration of 500 nM CXCL10 did not affect bacterial count within biofilm. A higher concentration of CXCL10 (1 µM) appears to have a negative effect on bacteria number. This is evident from the lower number of viable bacteria (1.9 CFU/ml) compared to the control and 500 nM CXCL10 (Fig. 8D). However, it remains unclear whether this apparent reduction in viable bacteria at 1 µM CXCL10 was due to direct inactivation of bacterial cells or overall inhibition of biofilm expansion. It is also unclear whether CXCL10 interacts specifically or nonspecifically with biofilm-embedded bacteria or matrix components at higher protein concentrations.

Thus, at relatively high concentrations, CXCL10 has some inhibitory effects on biofilms growth. On the other hand, the presence of CXCL10 can also trigger biofilms formation, which can be an escape strategy of S. aureus from host immune cells. However, it cannot be completely ruled out that the observed increase in the GFP signal could be a bias from increased metabolic activity due to CXCL10 challenge.