Morphology of amyloid fibers in S. mutans
When observed by TEM, amyloid fibers emanating from the S. mutans could be observed, which showed various length, from 50 nm to several microns (Fig. 1a). Generally, amyloid fibers on S. mutans had two morphologies, different in width. The thick one was fuzzy and about 10–16 nm, (Fig. 1b), while the thin one was clear and about 4-6 nm (Fig. 1c). Height image (Fig. 1d) and 3D reconstruction image (Fig. 1e) of amyloid fibers in S. mutans taken by AFM were supplied. The amplified images of amyloid fibers (Fig. 1f, g) in S. mutans suggested that amyloid fibers intertwined with each other and formed like net in vitro.
Amyloid fibers promoting S. mutans biofilm formation
During biofilm formation, the ThT fluorescence intensity shared similar pattern with that of biofilm biomass. EGCG at 50 μM, 100 μM and 200 μM obviously decreased the amount of amyloid fibers and biofilm biomass at all biofilm growth stages, with a dose-dependent decrease (Fig. 2a, b). We found biofilm was fragile and easy to be washed away when treated by EGCG. We collected the washed away bacteria and amyloid fibers could rarely be observed under TEM (Fig. 2c), which indicated that amyloid fibers were the crucial structure for biofilm formation and integrity. Besides, the live/dead bacteria staining also indicated that biofilm was obviously decreased when treated by EGCG, in contrast to the untreated group (Fig. 2d).
Amyloid fibers were the universal structure in clinical isolates
To observe whether amyloid fibers universally exist in the clinical isolates of S. mutans and correlated with their biofilm formation, we randomly selected 15 clinical isolates of S. mutans from our previously separated clinical isolates (Zhou et al. 2018). We cultured these clinical isolates for 24 h for biofilm formation, and found ThT fluorescence intensity had a significant linear correlation with biofilm biomass (p < 0.05; Fig. 3), which demonstrated that amyloid fibers were also correlated with biofilm formation in clinical S. mutans isolates, and amyloid fibers were the universal structure for biofilm formation.
Few amount of amyloid fibers existed in planktonic S. mutans
Next, to learn whether amyloid fibers only appear in biofilm, we explored whether planktonic S. mutans had amyloid fibers. When planktonic S. mutans grew to the saturation period, cells were centrifuged and observed under TEM, we also found few cells had amyloid fibers, but they showed very small ratio compared with that in biofilm (Additional file 1: Figure S1A). However, planktonic S. mutans had another connection structure between cells, with large amount, also emanating from the cell surface, but not like amyloid fibers when observed by TEM (Additional file 1: Figure S1B).
Isolation and characterization of amyloid fibers
In order to know more about the characteristics of amyloid fibers, we isolated amyloid fibers from cells. Successful isolation of amyloid fibers was verified by TEM (Additional file 1: Figure S2A). Amyloid fibers without treatment were run by SDS-page, and no bands could be seen (Additional file 1: Figure S2B), which was in consistent with the characteristics that amyloid fibers could dissolve in SDS. Moreover, we treated the extracted amyloid fibers with protease K, DNase I and RNase. After treated, amyloid fibers could still be detected by TEM (Additional file 1: Figure S2C).
Characteristics of amyloid fibers aggregated by purified proteins and fibrillation influencing factors
To learn the factors influencing amyloid fibrillation, the already known amyloid forming proteins, truncated protein C123 was purified for aggregation. C123 had a long lag time when observed at 37 °C, PH = 7 (Fig. 4), and until 48 h a small amount of amyloid fibers be observed by TEM. However, amyloid fibers could already be detected after culture at 37 °C for 4 h in biofilm formation, which meant that amyloid fibrillation was much rapid in vivo. ThT fluorescence results showed that low PH (PH = 3) obviously accelerated the process of amyloid fibrillation, and high temperature (60 °C) could also increase the process (Fig. 4). Acidic PH and high temperature both reduced the lag times, and the matured amyloid fibers could be observed at 4 h or 5 h after culture.
At the same time, the aggregated amyloid fibers at different phases were observed by TEM, and two aggregated types could be found (Fig. 5a). Type I was the dominating one, firstly few disordered granular structures could be seen, and these structures multiplied gradually. Then large amount of twisted short rod-like structures, about 4–6 nm in width and 20–100 nm in length, were observed. Finally, intertwined rigid amyloid fibers were seen. Type II could also be seen, but with relatively smaller amount. Firstly, small granular particles appeared and then formed into relatively straight short rods, after that several rods would unite together. Gradually, amyloid fibers grew longer by recruiting more particles to one tail of the fiber. In the end, they compacted into thick matured amyloid fibers with various lengths. The aggregated amyloid fibers formed by purified C123 truncated protein were different from amyloid fibers on S. mutans in morphology, and the aggregated fibers were rigid while the amyloid fibers on S. mutans were “soft”. The above two mode aggregated patterns were provided for better interpretation (Fig. 5b).
Amyloid fibers forming complex with eDNA
Firstly, we investigated the position relationship among live bacteria, amyloid fibers and eDNA in the early biofilm through laser scanning confocal microscope (LSCM). Results showed that different amounts of amyloid fibers (red) were found around live bacteria, and a part of eDNA (blue) gathered around live bacteria (Fig. 6a), indicating that amyloid fibers and eDNA might colocalize in biofilm. But the staining might be false positive, thus more evidence was needed. To explore whether the extracted amyloid fibers had eDNA, we tried to extract eDNA from amyloid fibers, which we called (a)eDNA to distinguish from eDNA extracted directly from extracellular matrix. We successfully extracted (a)eDNA. Agarose electrophoresis (AGE) was used to assess the molecular weight differences, and results showed that the molecular weight (a)eDNA was smaller but approximate to genomic DNA (> 10,000 bp, Fig. 6b), much larger than that of eDNA (< 100 bp, Fig. 6c), which indicated that (a)eDNA be protected from degradation after forming complex with amyloid fibers.
After knowing that eDNA formed complex with amyloid fibers, we wondered whether DNase could reduce amyloid fibers and biofilm biomass in vivo. DNase and EGCG were added into BHIs solely or together when culturing biofilm. Results indicated that the amount of amyloid fibers was significantly reduced by adding DNase I alone (Fig. 6d), but no significant differences were found in biofilm biomass when adding DNase I alone (Fig. 6e). However, combined use of EGCG and DNase I was more efficient in inhibiting amyloid fibers and biofilm biomass (Fig. 6d, e).