GAGs are involved in the adherence of pathogenic bacteria to lung cells
To analyze the possible involvement of GAGs in the adhesion of some common lung pathogens, biosynthesis inhibition experiments were performed in both lung fibroblasts and epithelial cells using two different compounds that disrupt synthesis at different levels, rhodamine B or genistein [32,33,34,35]. The cells treated with the inhibitors were exposed to previously labeled pathogens in individual experiments, resulting in a decrease in bacterial adherence in all cases, suggesting that GAGs are involved in the binding of the pathogens in both types of cell (Fig. 1a, b).
In epithelial cells, genistein displayed a statistically significant higher inhibitory effect with all the bacteria tested, with the exception of E. faecalis and K. pneumoniae (Fig. 1a). Similar results were obtained when MRC5 cells were analyzed, with genistein being the most effective at reducing the adherence of almost all bacteria, although in this case rhodamine B showed higher inhibition of S. pyogenes binding (Fig. 1b). When the data were analyzed as a set, however, no significant differences were observed in terms of type of inhibitor or cell line used (Fig. 1c).
Distinct GAG species are differentially involved in the binding of pathogens to lung cells
The role played by the different molecular species of GAGs in the adherence of pathogenic bacteria was studied using two strategies: the enzymatic degradation of these molecules from the cell surface, and by binding competition experiments with commercial GAGs.
The reduction of the levels of the different cell surface GAGs of the A549 and MRC5 cell lines was carried out using commercial bacterial lyases. All treatments, individual or combined, produced a decrease in bacterial adherence to lung cells (Fig. 2a, b).
In epithelial cells, the removal of both HS and CS by lyases reduced bacterial binding in all cases. The use of heparinases and the combination of all lyases showed significantly stronger effects on the adherence of all bacteria than chondroitinase alone. The combination of lyases decreased bacterial adherence to a similar extent as treatment with heparinases, except in the binding of S. marcescens, S. pneumoniae and K. pneumoniae, where the inhibition effect was greater (Fig. 2a). However, analyzing the averaged data, the degradation of CS produced a reduction in binding for all pathogens tested of 19% (±6.68), while treatment with heparinases resulted in an average reduction of 30% (±4.52), the difference between the two treatments being statistically significant (p < 0.01) (Fig. 2c). The combined digestion using both lyases decreased bacterial adherence by about 35% (±5.60) on average, which was only significantly different statistically with respect to the chondroitinase treatment (p < 0.001) (Fig. 2c). These data suggest that, in lung epithelial cells, both species are involved in adherence, but that HS seems to play a more crucial role.
In a similar vein, in the fibroblast cell line, digestion with bacterial lyases also caused a decrease in pathogen binding. The mixture of enzymes showed the strongest effect on binding in all cases, except S. marcescens and E. faecalis. Comparing the data obtained after individual use of heparinases or chondroitinase, significant differences were found in the case of the binding of S. aureus, K. pneumoniae and S. pyogenes, where the removal of HS showed an increased effect. For the remaining microorganisms, both types of enzymes decreased bacterial adherence to similar degrees (Fig. 2b). When average data were analyzed, treatment with chondroitinase ABC reduced adherence by 30% (±7.20), while HS treatment caused a decrease of around 35% (±7.51) (Fig. 2c). The combination of both lyases decreased adherence by more than 40%, although the differences between the effects obtained with lyases used individually and when combined are not statistically significant (Fig. 2c). It would thus seem that both types of GAG mediate in bacterial binding to a similar extent, but that the role of CS in the interaction between pathogen and fibroblast appears to be more important than in epithelial cells.
Comparing the effects of degradation of HS and CS between both lung cell lines, statistically significant differences were only observed in the treatment involving chondroitinase ABC (p < 0.05), which reduced adherence of bacteria more in the MRC5 than the A549 cell line.
To further establish the influence of the several species of GAGs as bacterial receptors in lung cells, diverse adherence interference experiments were performed using commercial HS, CS-A, CS-B, CS-C and an equimolar mixture of each of them, as described in the Methods section. The presence of GAG molecules decreased the binding of all bacteria tested in both the A549 line (Fig. 3a) and the MRC5 lines (Fig. 3b). Furthermore, in both lines adherence was reduced in a dose dependent manner.
In epithelial cells, when commercial GAGs were used individually, HS was the most effective interfering molecule. However, the role played by each type of CS varied, depending on the pathogen tested (Fig. 3a). When a mixture of all the GAGs was used, the effect obtained was similar to when using HS, with the exception of S. pneumoniae, which at low concentrations showed greater inhibition of adherence by HS (Fig. 3a). While no statistically significant differences between the mixture of GAGs and HS were found, significant differences were found between HS and CS-C (p < 0.05), as well as between the mixture of all GAGs and each type of CS individually (p < 0.05) (Fig. 3c).
In MRC5 cells, bacterial adherence was also reduced by the different species of GAGs used individually and by the mixture of GAGs (Fig. 3a). Comparing the individual effects of each GAG type, HS displayed the highest inhibition effect on adherence in all pathogens except S. marcescens, where CS-B showed similar values. As in the epithelial cells, the effect of the different CS species was dependent on the bacteria involved. Moreover, the effect of the equimolar mixture of GAGs was similar to HS, the only exception being S. aureus, where the combination of GAGs inhibited adherence more effectively (Fig. 3a). Taken together the results evidence that both, HS and the mixture of GAGs, showed the highest inhibitory effect, and no statistically significant difference was found between them (Fig. 3b). However, the mixture of GAGs displayed significant differences with every individual CS: CS-A (p < 0.01), CS-B (p < 0.05) and CS-C (p < 0.05); significant differences were also found between HS and CS-A (p < 0.05) (Fig. 3b).
In contrast, when the effects observed for each interfering molecule individually were compared to the impact of the combination of all of them were compared, no differences between the two types of lung cell were detected (Fig. 3c).
Differential interference of peptides with HP binding sequences in the binding of pathogenic bacteria to lung cells
Two peptides which include consensus HP binding sequences of different lengths, QKKFKN and FKKKYGKS, were designed to examine their influence on bacterial binding to lung cells. The effect of both peptides was first calibrated in our laboratory in adherence interference experiments using S. aureus and S. marcescens. The influence of both QKKFKN and FKKKYGKS was dose-dependent, resulting in a reduction in binding up to concentrations of 10–20 nM, diminishing their effects at higher concentrations (data not shown). Consequently, a concentration of 20 nM of each peptide was selected for subsequent adhesion assays. The use of either of the peptides caused a decrease in the binding of every pathogen tested in both types of lung cell (Fig. 4a, b).
In the A549 line, despite both peptides reducing bacterial binding, the long peptide was significantly more efficient in all cases except K. pneumoniae, where values were similar to those for the short peptide, and for E. faecalis binding where the short peptide showed the greater effect (Fig. 4a). Average values for the reduction of adherence for the long peptide were 40% (±9.80), while for short peptide they were 26% (±9.46). This observed difference between the effect of the two molecules, was analyzed and found to be significant only in epithelial cells (p < 0.05) (Fig. 4c).
In contrast, in fibroblasts, the effect of individual peptides depended on the pathogen in question. The long peptide had a significantly stronger effect on reducing adherence of S. pneumoniae, S. marcescens, K. pneumoniae and S. pyogenes, whereas the short peptide was more efficient in decreasing the binding of E. faecalis. Both peptides decreased adherence to a similar extent for E. coli and S. aureus (Fig. 4b). Considering all the results together, adherence was reduced by 28.5% (±7.67) with the short peptide and 68.06 (±9.53) with the long peptide, though the difference was not statistically significant (Fig. 4c).
Neither were significant differences found when comparing the results of each peptide between the two cell types (Fig. 4c).
Differential involvement of cell-surface HSPGs in the binding of pathogenic microorganisms to lung cells
To determine which species of HSPGs were expressed in both epithelial and fibroblast cell membranes, a transcriptome study of the genes encoding the core proteins of all the syndecan and glypican members was performed.
In A549 cells, transcripts for the four SDC isoforms were detected with different levels of expression, isoform 1 and 4 being the most strongly expressed, and isoform 2 showing the lowest level of expression. Regarding transcripts for GPCs, all six isoforms were also detected, but the magnitude of expression levels found varied widely. GPC-1 was the most abundant, with GPC-5 and -6 transcripts being expressed at levels about one order of magnitude, and those of GPC-2, −3 and −4 around three orders of magnitude lower than levels GPC-1 (Fig. 5a). In contrast, the arrangement of cell surface HSPG expressions in MRC5 cells was quite different: the four syndecan isoforms all appeared highly transcribed, and while the levels of the different glypican isoforms varied between themselves by about three orders of magnitude, similar to in the A549 cells, the pattern of expression was quite distinct, particularly for GPC-2 and -4, which expressed at higher levels (Fig. 5a).
To analyze the possible involvement of these families of HSPGs in pathogen binding to lung cells, different adherence essays were performed. To establish the role of GPCs in the binding of pathogens, they were removed from the cell surface using PI-PLC, an enzyme that cleaves their GPI-anchor. The results showed that this treatment only slightly decreased bacterial adherence in both A549 cells (Fig. 5b), and MRC5 cells (Fig. 5c), for all pathogens. In epithelial cells, binding was reduced by 4% (±3.52), and, in fibroblasts by 6% (±5.19). Comparing the average values obtained in each cell line, no great differences were detected between epithelial cells and fibroblasts. (Fig. 5d).
The involvement of SDCs in bacterial adherence was analyzed by blocking experiments using a combination of specific antibodies against the four isoforms of these molecules. In all cases, in both epithelial and fibroblast cells, this treatment decreased bacterial binding (Fig. 5b, c), by 27% (±10.32) in A549 cells and 28% (±6.31) in MRC5 cells (Fig. 5d), the difference not being significant.
The analysis of the differences between PI-PLC and anti-SDC treatments on each bacterium followed a general pattern, which revealed the significantly higher involvement of SDCs in binding to epithelial cells (Fig. 5b). In terms of adherence to fibroblasts, the differences between the two treatments were statistically significant for all bacteria (Fig. 5c).
Additionally, when all the data were analyzed, the differences between PI-PLC and anti-SDC treatments on bacterial adherence were significant for both cell lines (p < 0.001) (Fig. 5d).
Influence of specific N- and O- sulfations on bacterial adherence to lung cells
To analyze the influence of sulfation at specific positions on the disaccharide unit of HS chains on interaction with bacteria, a variety of adherence interference experiments were performed using 2-O, 6-O and N- desulfated HPs and compared to normal HP as control. For all the pathogens tested, the presence of any heparin-derived molecule diminished binding, both to A549, and to MRC5 cells, although the results varied depending on the pathogen (Fig. 6).
In epithelial cells, desulfated HPs were less effective in inhibiting the adhesion of all bacteria tested than fully sulfated HP, although the scale of the effect varied depending on the specific microorganism analyzed (Fig. 6a). Analyzing the effects of desulfated HPs for each bacterium the results showed similar values in S. aureus, K. pneumoniae and E. faecalis. In the adherence of E. coli and S. pyogenes 2 and 6-O-desulfated HP showed significantly higher effects than N-desulfated HP whereas, 6-O- and N-desulfated HP affected the bonding of S. marcescens even more strongly, and 6-O-desulfated HP decreased the adherence of S. pneumoniae more than other desulfated HPs did (Fig. 6a). Pooling all the data, no significant differences were observed between in inhibition values for the three forms of desulfated HP tested, although each showed significant differences when compared to normal HP (p < 0.05) (Fig. 6c).
In fibroblasts, binding was also inhibited by the presence of each of the tested molecules, but in this case the effect on the different pathogens was more homogeneous than in epithelial cells. As in the A549 cell line, both when analyzing the data of each bacterium individually, and when the set of all bacteria were considered together, native HP displayed the highest inhibitory effect on adhesion in all cases, showing statistically significant differences compared with each of the other desulfated-HPs (p < 0.001) (Fig. 6b, c). Comparing the individual effect of each desulfated HP, very different results were obtained depending on bacterium involved. While no significant differences were found between the effects of these molecules on the adherence of S. pneumoniae and S. marcescens, 6-O-desulfated HP showed the highest effect on binding for S. pyogenes, S. aureus and E. faecalis. 6-O- and N-desulfated HP reduced bonding of E. coli more significantly, and N-desulfated HP showed significant differences with 2-O-desulfated HP in the case of K. pneumoniae (Fig. 6b). However, regarding the average data, in these cells differences dependent on the type of sulfation were detected, with 6-O-desulfated HP able to compete more effectively than 2-O-desulfated HP (p < 0.05), while N-desulfated molecules produced more heterogeneous results (Fig. 6c).