Artificial biofilms establish the role of matrix interactions in staphylococcal biofilm assembly and disassembly

We demonstrate that the microstructural and mechanical properties of bacterial biofilms can be created through colloidal self-assembly of cells and polymers, and thereby link the complex material properties of biofilms to well understood colloidal and polymeric behaviors. This finding is applied to soften and disassemble staphylococcal biofilms through pH changes. Bacterial biofilms are viscoelastic, structured communities of cells encapsulated in an extracellular polymeric substance (EPS) comprised of polysaccharides, proteins, and DNA. Although the identity and abundance of EPS macromolecules are known, how these matrix materials interact with themselves and bacterial cells to generate biofilm morphology and mechanics is not understood. Here, we find that the colloidal self-assembly of Staphylococcus epidermidis RP62A cells and polysaccharides into viscoelastic biofilms is driven by thermodynamic phase instability of EPS. pH conditions that induce phase instability of chitosan produce artificial S. epidermidis biofilms whose mechanics match natural S. epidermidis biofilms. Furthermore, pH-induced solubilization of the matrix triggers disassembly in both artificial and natural S. epidermidis biofilms. This pH-induced disassembly occurs in biofilms formed by five additional staphylococcal strains, including three clinical isolates. Our findings suggest that colloidal self-assembly of cells and matrix polymers produces biofilm viscoelasticity and that biofilm control strategies can exploit this mechanism.


Screening for biofilm-forming clinical isolates of S. epidermidis.
To identify biofilm forming clinical isolates, we screened the 54 patient samples of the library of Sharma et al. 1 . For this, we grew all 54 isolates in 10 mL tryptic soy broth with 1 wt. % glucose (TSB G ) at 60 RPM and 37°C overnight in 15 mL culture tubes and visually inspected the walls of the container for biofilm growth. We found 6 potential strains (P2, P6, P12, P18, P37, P47) that were further evaluated for biofilm formation by growing a single colony in 400 μL TSB G for 18-hours at 60 RPM and 37°C. CLSM was used to inspect for strong biofilm formation, as determined by visual observation of arrested cells within the 18-hour biofilm. We identified S. epidermidis P18, P37, and P47 as suitable strains.

Effect of probe size and probe surface chemistry on diffusing wave spectroscopy (DWS) measurements
A probe size and surface chemistry study on PIA and EPS solutions harvested from biofilms indicated no local heterogeneity and no probe-polymer interaction in these solutions.
DWS microrheology of the biofilms however exhibited strong probe size dependence. Probes larger than the diameter of S. epidermidis cells ( > 0.5 m ) exhibited nonergodic dynamics, as characterized by a g 2 (t) intercept << 1 2 , because of entrapment within or between biofilm clusters, and were therefore not analyzed further. Probes of size equivalent to that of the bacterial cells -0.5 m diameterexhibited decay in g 2 (t) that was consistent with thermally induced random motion. The g 2 (t) of probes smaller than the cellular diameter (< 0.5 m) decayed to a non-zero plateau; this behavior indicates probe localization because of their entrapment within the biofilm. The biofilm creep compliance, J(t), as a function of probe size is shown in Supplementary Figure 3. We found that the short time response (t < 10 -3 s ) was independent of probe size while the long time regime (t  10 -3 s) was found to be strongly dependent on probe size. Probes of size 0.5 ma dimension equivalent to the size of a S. epidermidis bacterial cellresulted in a biofilm J(t) that matched the mechanical rheometry measurements 3 .
Microscopic visualization of the probes in the biofilm showed that 0.5 m carboxylate and amine probes strongly associated with the biofilms (fraction of aggregated or stuck probes > 70%), while 0.5 m sulfate probes showed much weaker association of < 15%. Using the theory of multiple scattering in binary suspensions 4 and the structure factor formulation for aggregates 5 , we found that the impact of this association caused < 1 % change on the mean free path of multiply scattered light and a < 6 % change in the calculated MSD 6 .

Accounting for initially flocculated planktonic bacteria in artificial biofilms
We accounted for the presence of flocs in the planktonic cultures used for CLSM microrheology in two ways. First, the 1% of trajectories that were most immobilized at each condition probed were discarded from further analysis. Second, the error in the average <Δx 2 (Δt)> due to the flocs within the initial cultures was estimated. To determine this error, we resolved the van Hove self-correlation function of displacement 7 into initially flocculated (slow displacement dynamics) and singlet bacterial (fast displacement dynamics) contributions. We