Staphylococcus epidermidis strain RP62A, a pathogenic strain which forms biofilms, was used for all experiments. Loop inoculates from frozen glycerol stock (–80°C) were grown on tryptic soy agar (TSA, Fisher Scientific) plates for at least 16 h at 37°C. A single colony was selected and cultured in tryptic soy broth with glucose 1% (TSBG) at 37°C under agitation at 200–250 rpm. Exponential phase growth was defined as an optical density of 0.4 ± 0.1 measured at 600 nm (Supplementary Figure 1). A 30 mL sample of the culture at this phase was removed and left undisturbed for 30–60 min at room temperature to allow for settling of flocculated cells prior to use in subsequent studies. For stationary phase, the remaining culture was returned to the incubator at 37°C to continue growth under agitation until an optical density of 1.4 ± 0.05 was obtained. Since rheometry experiments required significantly larger volumes, separate cultures were grown to exponential and stationary phases as defined above.

Staphylococcus epidermidis were cultured to exponential and stationary phase as described above. Cell density was quantified in each sample with a Incyto Hemacytometer (Thermo Fisher Scientific, Waltham, MA, United States) and cells were centrifuged into a pellet at 5,000 rpm for 3 min. Supernatant was discarded, and pelleted cells were diluted to concentration of 2.0 × 10⁹ cells/mL. Two conditions were prepared in triplicate for each growth phase–unlabeled control and labeled fibrinogen. Control and labeled fibrinogen pellets were resuspended in 1X Hank’s Buffered Salt Solution (with Calcium, Magnesium, no phenol red, 1 g/L D-Glucose, Thermo Fisher Scientific, Waltham, MA, United States). Fluorescently labeled fibrinogen (Alexafluor 647 Fibrinogen, 650/668 nm, Thermo Fisher Scientific, Waltham, MA, United States) was added to a final concentration of 0.5 mg/mL and incubated for 15 min. Cells were then analyzed by flow cytometry (Cytoflex, Beckman). Histograms of area fluorescence (excitation 650 nm/emission 668 nm) from 10⁵ events per sample were obtained. Mean fluorescence intensity (MFI) was determined as a surrogate for total fibrinogen concentration on the cell surface using FCS Express (v.7, De Novo Software, Pasadena, CA, United States). Nine experiments were completed in triplicate with the median and interquartile range (IQR) reported. Statistical analyses were performed with Wilcoxon paired signed rank test with significance defined as p < 0.05.

Fibrinogen coating of glass coverslips was performed as previously described (Kuusela et al., 1985) with minor modification. Briefly, fibrinogen from human plasma (EMD Millipore Corporation, Billerica, MA, United States) was reconstituted as directed and stored at –80°C in aliquots. Frozen aliquots were thawed and diluted to concentration of 10 μg/mL using sterile distilled water. Glass coverslips were coated with fibrinogen for 4 h at 20°C, blocked with 1% bovine serum albumin for 1 h, and washed three times with normal saline. This method was previously determined to have ∼80% binding efficiency (Kuusela et al., 1985) and for our geometry is estimated to result in a fibrinogen surface density of 80 molecules/μm².

The stationary phase sample was diluted with normal saline to match optical density of the exponential phase culture then left undisturbed at 20°C for 1 h to remove large flocculates. Exponential and stationary phase cells were fluorescently labeled with Syto 9 (Thermo Fisher Scientific, Waltham, MA, United States). Cells were loaded into a 10 mL syringe on a standard syringe pump (Harvard Apparatus). Fibrinogen coated glass coverslips were assembled into a parallel-plate flow chamber (Warner instruments RC-31 with 250 μm thick, 3.2 mm slot gasket). Cell suspensions were infused at a rate of 125 μL/min (Reynolds number = 1.2, wall shear stress = 0.067 Pa) through the parallel-plate flow chamber mounted on a epifluorescent microscope (Nikon Eclipse 80i). Images were captured every 3 s for 30 min. A total of 10 experiments were completed in duplicate and the mean for each experiment was used.

Custom image analysis was performed using MATLAB (v.R2018b, Natick, MA, United States). Briefly, a new adhesion event was defined as the new appearance of a bacterium which persisted for two or more consecutive images in a single location. The deposition rate was calculated as the slope of adhesion events versus time for each experimental replicate. Residence time was defined as the total time a single bacterium persisted at a single location. Student’s paired t-test was used for hypothesis testing comparing deposition rate, expressed as cells/cm² s, versus growth phase with p < 0.05 defined as significant. For residence time in seconds, a χ² test was used to compare the population of bacteria persisting for greater than 500 s for each growth phase with significance defined as p < 0.01.

Target genes were selected to probe for proteins whose function may be responsible for the observed differences in bacterial binding to fibrinogen. Specifically, we chose genes involved in bacterial adhesion ranging from cell–cell adhesion via aap and icaA, cell-abiotic adhesion via atlE and cell-host protein (e.g., fibrin/fibrinogen, collagen) adhesion via the sdr family. In addition, sspA, a serine protease which putatively degrades fibrinogen (Otto, 2009; Sugimoto et al., 2013), was probed as a potential mediator of bacterial release from and/or subsequent degradation of the fibrin clot. The ica repressor gene (icaR) was probed as a control as this gene has been shown to decrease in the transition from exponential to stationary phase (Yao et al., 2005).

RNA was extracted from S. epidermidis grown to either exponential or stationary phase. One mL of bacterial cultures was centrifuged at 8,000 rpm for 1 min to pellet the cells. After washing with RNase free water, cells were resuspended. Cell lysis was achieved with 5 μL lysostaphin (2 μg/μL), incubation at 37°C for 45 min followed by addition of 75 μL of TRIzol and vigorous pipetting. Samples were vortexed and then centrifuged at 12,000 rpm for 10 min at 4°C. Supernatant was removed and incubated at room temperature for 5 min. Chloroform (15 μL) was added to the sample, vortexed for 15 s, and incubated at room temperature for 3 min. Following this, samples were centrifuged at 12,000 rpm for 15 min at 4°C. The aqueous phase was isolated and transferred to RNeasy Mini Kit spin columns (Qiagen) and RNA was purified per the manufacturer’s instructions. High quality RNA yield was confirmed by UV spectrophotometry (NanoDrop 2000, Thermo Scientific) and RNA integrity was confirm using a Bioanalyzer (Agilent 2100). Samples were stored at –80°C prior to RT-PCR.

QuantiTect Probe RT-PCR (Qiagen) was performed in 50 μL reactions using an Applied Biosystems Step 1+ real time PCR system. The reaction mixture contained 25 μL 2 × Probe RT-PCR Master Mix, 0.5 μL RT Mix, 5 μL of exponential or stationary phase RNA template (5 ng), 5 μL of each gene forward/reverse-primer (F/R-primer, 1 μM) and probe (0.2 μM) and 4.5 μL of RNase-free water. The RT-PCR Master Mix contained HotStarTaq DNA polymerase, ROX reference dye, dNTP mix, and RT-PCR buffer [Tris–HCl, KCl, (NH₄)SO₄, 8 mM MgCl₂, at pH 8.7]. Thermal cycling protocol began with 30 min reverse transcription step at 50°C, followed by an initial activation step of 15 min at 95°C, denaturation of 15 s at 94°C, and combined annealing/extension step at 65°C for 60 s for a total of 45 cycles. Sequences for forward, reverse and probe oligonucleotides are shown in Supplementary Table 1. Fold change in expression (stationary vs. exponential phase) was determined using the ΔΔCT method with the 16s ribosomal subunit as an endogenous control. Student’s t-test was used for hypothesis testing of significance (p < 0.05) of the fold change in gene expression. p-values were corrected for multiple comparisons using Benjamin-Hochberg control of false-discovery rate.

Materials and constitution methods are described in detail in previously (Ma et al., 2017). Briefly, citrate-free thrombin from human plasma and fibrinogen (Fg) from human plasma were purchased from EMD Millipore (Billerica, MA, United States). Fibrinogen solutions (0.5 mg/mL) were made by diluting the stock solution with 1X Hank’s Balanced Salt Solution at 37°C. Clot was initiated by adding thrombin (0.5 U/mL as final concentration). Clot elasticity was characterized by measuring the linear elastic modulus, G′, on a mechanical rheometer (TA Instruments AR-G2, cone and plate fixture with a 2° cone angle and a 40 mm diameter). All tests were performed at a frequency ω = 0.1 Hz under an imposed strain γ = 0.01 at 37°C in a humid atmosphere. Clot microstructures were observed using a Nikon A1Rsi confocal laser scanning microscope with a 100 ×, 1.45 NA, oil immersion objective and images were analyzed to determine mesh size using MATLAB as we have previously described (Ma et al., 2017). The steady state of fibrin structures is defined as the variability in G′ in any 2 min interval being no more than 0.5% for 20 min (Ma et al., 2017). Bacteria concentration for both exponential and stationary phase was 4.0 × 10⁹ cells/mL to match in vivo biofilm concentrations (Lungren et al., 2014). Statistical comparisons were made via ANOVA with post-hoc pair-wise Tukey test with significance level of p < 0.05.

Figure 1A shows an example of a single flow cytometry experiment to quantify fibrinogen affinity. The mean fluorescence of the stationary phase population is greater than that of the exponential phase with a similar distribution. Composite mean fluorescence index of all experiments (N = 9 with triplicate measures) is shown in Figure 1B. On average, the mean fluorescence of stationary S. epidermidis incubated with fluorescently labeled fibrinogen was significantly higher as compared to exponential phase under the same condition (3,642 ± 349 vs. 6,388 ± 919, p = 0.004).

To determine if the increased affinity translated to actual adhesion to a fibrinogen coated surface, where shear induced binding enhancement may contribute, we evaluated deposition rate and residence time in a continuous flow, parallel-plate system. Fluorescently labeled S. epidermidis flowing past immobilized fibrinogen appear as streaks on image capture but are more clearly defined circular shapes as they slow and come to rest on the surface (Figures 2A–E). Persistence of a cell in a single location for more than one consecutive image defined an adhesion event (white box in Figures 2B–D). Absence of a previously identified adhered bacteria (white box in Figure 2E) defined a release event. The total time spent adhered was defined as the residence time. The number of adhesion events in a high power field of S. epidermidis on immobilized fibrinogen showed a steady, linear increase over time for both exponential and stationary phase (Figure 2F); however, the deposition rate (slope of adhesion events vs. time) of stationary phase cells was significantly greater than the deposition rate of cells in exponential phase (5,360 ± 1,776 cells/cm² s vs. 2,212 ± 264, cells/cm² s, p = 0.037) (Figure 2G). Additionally, stationary phase S. epidermidis had longer residence time when compared to exponential phase (Figure 2H). Most notably there was a population of stationary phase cells that had a residence time of greater than 500 s, skewing the distribution to the right as compared to the exponential phase. That is, only 2.7% of exponential cells had a residence time of greater than 500 s while 13.1% of the stationary phase cell had a residence time greater than 500 s (p = 0.004).

To identify potential mediators of growth phase-dependent fibrinogen adhesion, phase dependent changes in expression of specific adhesin genes were determined. The relative fold-change in gene expression for stationary phase compared to exponential phase S. epidermidis are shown in Table 1. There was a modest, 2.2-fold increase in sdrG expression. Other cell adhesion molecules had no significant change in expression under these conditions. We also considered the serine protease, sspA, which has been predicted to specifically degrade fibrinogen (Otto, 2009; Sugimoto et al., 2013). There was a 3.5-fold increase in sspA expression in the stationary versus exponential phase. As predicted, icaR showed a decrease in expression in the transition from exponential to stationary phase.

Next we determined how growth phase-dependent fibrinogen binding affects fibrin clot mechanics and structure. Clot formation during rheometry is observed by rapid increase in the linear elastic modulus (G′) over time (Figure 3A). Based on our previous work (Ma et al., 2017) and reproduced here for purposes of comparison, the G′ of a pure fibrin clot (0.5 mg/mL fibrinogen) without bacterial cells reaches a steady-state at about 40 min with a value of 1.5 ± 0.2 Pa. The addition of stationary phase S. epidermidis retards clot formation with a two-step kinetic profile of G′ (Figure 3A), consistent with our previous work (Ma et al., 2017). The first step represents a transient plateau at a lower elastic modulus than the pure fibrin clot. This plateau is followed by a second rapid increase in G′ to a final steady-state value of 5.7 ± 1.0 Pa. This two-step kinetic profile of clot formation is completely absent for exponential phase S. epidermidis of the same cell concentration. In addition, the steady state for this condition is not reached within 100 min. The growth of the elastic modulus of the exponential phase is significantly retarded relative to the stationary phase. Its kinetics are also somewhat slower than pure fibrin. The final (t = 100 min) modulus of the exponential phase is not significantly lower than that of pure fibrin (1.3 ± 0.4 Pa).

Figures 3B–E compare the steady-state structures of the corresponding conditions with the fibrin elasticity measurements. The pure fibrin network has a homogeneous pore size distribution. As we have previously shown (Ma et al., 2017) and reproduced here for purposes of comparison, the pure fibrin mesh size under these conditions is 5.7 ± 0.8 μm (Figure 3C). This value is consistent with other measurements in the literature (Piechocka et al., 2010). The fibrin network structure formed in the presence of exponential phase S. epidermidis remains homogeneous (Figure 3D) and the mesh size is slightly increased (7.0 ± 0.5 μm) over pure fibrin (p = 0.023). In comparison, the network formed in the presence of stationary phase S. epidermidis is heterogeneous (Figure 3E) with two characteristic mesh sizes (2.6 ± 0.8 and 23.1 ± 6.8 μm), as previously reported (Ma et al., 2017), which are significantly different than either pure fibrin or fibrin with exponential phase cells (p < 10^–5). We also observed that the exponential phase cells are localized to the voids of the fibrin network (Figure 3D) rather than bound to the fibrin fibers as seen for stationary phase cells (Figure 3E).