Physical mechanisms driving the reversible aggregation of Staphylococcus aureus and response to antimicrobials

Formation of non-sessile, auto-aggregated cells of Staphylococcus aureus contributes to surface colonization and biofilm formation, hence play a major role in the early establishment of infection and in tolerance to antimicrobials. Understanding the mechanism of aggregation and the impact of aggregation on the activity of antimicrobials is crucial in achieving a better control of this important pathogen. Previously linked to biological phenomena, physical interactions leading to S. aureus cellular aggregation and its protective features against antimicrobials remain unraveled. Herein, in-vitro experiments coupled with XDLVO simulations reveal that suspensions of S. aureus cells exhibit rapid, reversible aggregation (> 70%) in part controlled by the interplay between cellular hydrophobicity, surface potential and extracellular proteins. Changing pH and salt concentration in the extracellular media modulated the cellular surface potential but not the hydrophobicity which remained consistent despite these variations. A decrease in net cellular negative surface potential achieved by decreasing pH or increasing salt concentrations, caused attractive forces such as the hydrophobic and cell–protein interactions to prevail, favoring immediate aggregation. The aggregation significantly increased the tolerance of S. aureus cells to quaternary ammonium compounds (QAC). The well-dispersed cell population was completely inactivated within 30 s whereas its aggregated counterpart required more than 10 min.

The Gram-positive S. aureus is an opportunistic pathogen able to thrive in a wide range of environments such as on dry inanimate surfaces, in body fluids or on human skin 1 . Traditionally, S. aureus cells have been studied as free-floating planktonic organisms but lately, multicellular communities of S. aureus forming biofilms have attracted more attention due to their increased resistance to antimicrobials [2][3][4] . Recently, autoaggregation has been identified to be among the first steps in building a biofilm 5 . In contrast with biofilms, bacteria cell aggregates are not necessarily surface attached but are free-floating. They are also generally smaller than typical biofilms, and they do not form the mushroom-like towers that have been observed with in vitro biofilm growth 2 . A deeper understanding of the mechanisms underlying the aggregation of S. aureus cells and the impact of aggregation on the activity of antimicrobials will facilitate the development of novel disinfection and therapeutic strategies. Until now, the mechanism of aggregation in bacteria has mostly relied on biological phenomena such as quorum sensing, stress response, predation pressure or homotypic interactions between the bacteria surface and/ or exogenous host proteins and molecules [5][6][7] . A different, yet complementary approach consists in viewing the suspensions of bacterial cells as colloids, and in applying principles of soft condensed matter physics to predict the behavior of living cells 8,9 . S. aureus is a non-motile, 1-micron spherical, negatively charged bacterium that closely resembles colloidal suspensions of particles when cultivated in a liquid medium. Destabilization of colloids and formation of aggregates can be caused by surface-surface attractive forces such as hydrophobic 10 or oppositely charged 11 ; depletion interactions 9,12 and surface potential variations 13 . In this work, we demonstrate that planktonic suspension of S. aureus cells exhibit rapid, reversible aggregation in part driven by the bacteria surface potential and hydrophobicity. In our experiment the bacteria surface potential was governed by the pH and salt concentration of the extracellular medium. Bacteria cell aggregation was emphasized at low net cellular surface potential and by the presence of proteins. The consequence of such aggregation was studied in terms of bacterial tolerance to a common disinfectant, quaternary ammonium compound (QAC) and it was demonstrated www.nature.com/scientificreports/ that a population containing a greater fraction of aggregated cells was more than 20 times harder to kill than the one with fewer aggregates.

Results and discussion
To elucidate if physical forces can cause S. aureus aggregation, hydrophobicity and aggregation behavior of S. aureus suspended in water with low (0.015%) and high (0.5%) concentrations of salt (NaCl) and in Nutrient broth was studied as a function of pH (Fig. 1). Nutrient broth is a proteinaceous medium used in many standardized microbiology tests. A 0.015% (2.6 mM) NaCl solution was selected to match the salt concentration of Nutrient broth and 0.5% NaCl (85.6 mM) was selected because it is used in other standardized media including AOAC Nutrient broth in addition to being a concentration relevant to physiological conditions. In each media, multiple bright-field optical microscopy images of the bacterial suspensions were analyzed at several pH values ("Materials and methods" section). Overall, S. aureus aggregation was observed to be pH-dependent in all three media. The aggregation occurred rapidly, in a matter of 15 s and the overall size of the aggregates noticeably increased as pH decreased (Fig. 1a). The total percent of cells aggregated ("TCA") was evaluated based on the percentage of cells that were found in aggregates of sizes larger than 4 µm (Fig. 1b). An arbitrarily set pH, denoted as pH Agg , was defined as the threshold pH whereat 10 or more percent of the cell population were aggregated. The pH Agg values were measured at approximately 4.8, 8.0 and 9.1 in water at low and high NaCl concentrations, and in Nutrient broth, respectively. As pH dropped below the pH Agg value, a sharp increase in the overall TCA populations was observed in each medium, reaching up to 85% in water at low NaCl concentrations, 75% in water with 0.5% NaCl and 88% in Nutrient broth. The TCA population was further subdivided into two groups based on the cross-areal diameter of individual aggregates: percent of cells in aggregates of sizes between 4 and 8 µm ("N C48 ") and percent of cells in aggregates larger than 8 µm ("N C8 "). At pH < pH Agg , N C48 increases at a moderate www.nature.com/scientificreports/ rate to the reach maximum values of 20-29% whereas N C8 continuously increased, leading to the final values of 62%, 52% and 69% for water at low and high salt concentrations and for Nutrient broth, respectively. These observations could be explained by the formation of new aggregates in combination with the growth of already formed aggregates as pH dropped. Further analyses revealed that, in all three media, aggregates of sizes larger than 8 µm grew both in size and in number at pH below the pH Agg value (see Fig. S1 in the Supplements). In summary, when salt or proteins were present at physiologically relevant concentrations, the aggregation of S. aureus was enhanced. Indeed, the pH Agg values shifted towards greater values along with more than 25% of the cell population existing in aggregates of at least 8 µm in size at larger pH values (pH 5 and 6.5 for 0.5% NaCl and Nutrient broth as compared to pH 4.2 in 0.015% NaCl). In this work, the aggregation of S. aureus occurred in a matter of seconds upon change of the extracellular pH and salt content. Additionally, SEM imaging of cells grown in Nutrient broth at pH ~ 7 ("Materials and methods" section) did not reveal external web-like polymers or protein matrices surrounding the bacterial cells (see Fig. 2a). These observations suggest that the mechanism of aggregation in S. aureus may not derive from changes in cellular metabolism or from de-novo synthesis of adhesive factors (both requiring a longer time for gene expression and protein synthesis) as previously thought by Haaber et al 6 . Since the overall surface potential of bacteria is pH-and salt-dependent 14 , it is possible that the rapid aggregation observed in S. aureus could, in fact, be due to changes in the bacterial surface potential. We measured the surface potential of S. aureus as a function of pH at 0.015% and 0.5% NaCl. Results presented in Fig. 2b show that the absolute surface potential of S. aureus becomes less negative as pH decreases. Indeed, the zeta potential (ζ) of bacteria dispersed in water with 0.015% NaCl increases from − 25 mV (7 < pH < 11) to + 10 mV (pH 3). By decreasing the net surface potential, the electrostatic repulsion between the cells also decreased thereby causing aggregation. The addition of 0.5% of NaCl shifted the entire set of |ζ| values upwards on the zeta potential graph, indicating an overall decrease in the surface potential across the entire pH range studied. This behavior is characteristic of charge screening by salt in colloidal suspensions, as reported by Leckband and Israelachvili 13 and resulted in a greater aggregation in water www.nature.com/scientificreports/ with 0.5% of NaCl as compared to 0.015% NaCl (Fig. 1b). In addition to surface potential, hydrophobic interactions can play an important role in cell-cell aggregation 13 . The hydrophobicity of S. aureus cells was measured 15 and the results presented in Fig. S2 in Supplements showed the S. aureus strain used herein to be hydrophobic (in alignment with the results of Reifsteck et al.) 10 regardless of the extracellular pH or salt content. Simulations of S. aureus aggregation using an extended DLVO model [16][17][18][19] (XDLVO, see Fig. S3 in Supplements) revealed that without this hydrophobic property (modeled by attractive hydrophobic forces 13 ), cells carrying a strong negative potential at pH 5-7 (below pH Agg ), would not aggregate in water with 0.5% of NaCl. Indeed, across the pH range 3-12, the hydrophobic attractive forces are about twice as large as the Van Der Waals attractive forces. In Nutrient broth, the surface potential of bacteria is expected to be similar to the surface potential of bacteria dispersed in water containing a matching salt concentration of 0.015% NaCl. However, as compared to water with 0.015% NaCl, the aggregation of S. aureus was considerably enhanced in Nutrient broth (pH Agg is shifted from 4.8 to 9.1). While Nutrient broth is a complex medium, proteins such as the ones it contains have already been reported to promote cell aggregation via several mechanisms 5,13 including depletion and bacterial cells bridging. In our study, the two latter potential mechanisms could have been enhanced at low pH due to decrease in the net surface potential and aggregation of proteins contained in Nutrient broth (see Fig. S4 in Supplements). The consequence of S. aureus cell aggregation was studied in relation to bacterial tolerance to QACs, which are broad-spectrum antimicrobials 20,21 commonly used in disinfectant and sanitizer formulations. QACs are membrane-active agents comprising a positively-charged head which electrostatically interacts with the negatively charged bacteria membrane, and a hydrophobic tail which can penetrate into the membrane ultimately leading to membrane structural damages and bacteria lysis 20 . Suspensions of cells (6.7 log 10 CFU mL −1 ) in Nutrient broth were adjusted to pH values between 3 and 11. Then, each sample was challenged with either 20 or 2 ppm of Didecyldimethylammonium chloride (DDAC) for 30 s, 1, 2, 5 and 10 min ("Materials and methods" section). The minimum time required to inactivate the entire cell population (time-to-kill) was plotted as a function of pH. Results presented in Fig. 2c show that for both DDAC concentrations, the time-to-kill values decrease with increasing pH with the shortest time-to-kill values observed at pH 9 and 11 (see Tables S1 and S2 in Supplements). Two parts per million (ppm) of DDAC at pH 9 were sufficient to achieve a complete kill within 5 min of exposure time. Increasing both DDAC concentration and pH drastically reduced the time-to-kill. Indeed, we report a complete deactivation of S. aureus within seconds (30 s) of exposure to 20 ppm of DDAC at pH 9, compared to 3 min being required at pH 6.5. As pH is decreased from 7 to 3.2, the prediction of the aggregation for 6.7 log 10 CFU/mL yield aggregation of 35% of the cell population (see Fig. S5 in Supplements). Overall, S. aureus cells are less sensitive to DDAC when the bacterial surface is less negatively charged and the cell population is more aggregated. Interestingly, a tenfold increase in the DDAC concentration reduced the time-to-kill from more than 10 min to 3 min at pH 6.5 whereas a diminution of the time-to-kill was not observed at lower pH values (pH 5 and 3). As previously shown, when pH decreased, the proportion of aggregated cells increased, with larger and more numerous aggregates forming at the lowest pH values. Additionally, in low NaCl conditions such as found in Nutrient broth or water with 0.015% NaCl, the bacterial surface potential measured at pH 7 and pH 9 (Fig. 2b) were comparable (ζ ~ − 22 mV); whereas the aggregation of cells was greater at pH 7 than it was at pH 9. Thus, the superior antimicrobial efficacy observed with DDAC at pH 9 as compared to pH 7 cannot be attributed to the difference in the electrostatic attraction between negatively-charged bacteria and positivelycharged QACs. We therefore, speculate that the increased tolerance to QACs observed at low pH values were likely driven by the physical barrier provided by cellular aggregation.
Bacterial aggregates including sessile biofilms often rely on external shear forces or secretion of enzymes to disperse 2,22 . To determine whether the aggregation in S. aureus is reversible, we imaged aliquots of S. aureus cells at pH adjusted to 3, and then raised to 9. The process was repeated in a different set of identical samples starting from pH 9 and then lowered to pH 3. Results presented in Fig. 3a show the aggregation to be reversible in all three media studied. Indeed, aggregates formed at pH 3 rapidly dispersed when the pH was raised to 9. Similarly, the singlet cells observed at pH 9 quickly aggregated when the pH was lowered to 3. This behavior suggests that in low pH ranges, the net surface potential of bacteria is small enough such that long range electrostatic interactions are overpassed by attractive interactions, thus inducing aggregation. Irreversible aggregation occurs when bacteria get extremely close to each other 13 . The observed reversible aggregation can be explained by the presence of a short range (~ 0.5 nm) repulsive interaction 17 between the bacteria, which precludes any irreversible aggregation due to Van der Waals interactions. We have included such a repulsion in our XDLVO model in water with NaCl (see the Supplements for more details). Standard disinfection tests such as developed by the EPA 23 and the "AOAC use dilution test" 24 require the complete kill of test bacterial population within 10 min of exposure in proteinaceous media and are conducted as a qualitative assay based on turbidity similar to the methodology applied in this study. To meet such demanding standards, industrial and healthcare disinfectants often formulated at neutral pH (considered as an optimal pH for QACs performance) require the use of several thousands of ppm of QACs. Since aggregation of S. aureus is reversible with increasing pH, a change in formulation pH would be a quite simple solution to help inactivate this bacterium more efficiently, with lower usage of biocidal actives.
To investigate whether the surface potential-mediated aggregation of S. aureus is a general and observable trait in other species of bacteria, we imaged Enterococcus hirae (Gram positive, non-motile) and Pseudomonas aeruginosa (Gram negative, motile) suspended in Nutrient broth and in water with 0.015% and 0.5% NaCl, each adjusted to pH 3, 7 and 9. The results, presented in Fig. 3b, indicate that at pH 3 E. hirae cells aggregated whereas P. aeruginosa cells did not in all three media. The aggregation of E. hirae occurred within a few minutes, which was slower than that of S. aureus. The bacterial surface potential was determined as a function of pH for both bacteria dispersed in water with 0.015% and 0.5% NaCl. The results presented in Fig. 3b show that the net surface potential in E. hirae decreased significantly at lower pH values. On the contrary, the surface potential in P. aeruginosa decreases to a much lesser extent within the pH range of 2 to 12, confirming the results reported by Shephard et al 25  www.nature.com/scientificreports/ to a stronger electrostatic repulsion between the bacteria overcoming the attractive forces mediated by proteins. Indeed, as compared to E. hirae, the net surface potential of P. aeruginosa is higher at pH below 4. These results confirm that a significant reduction in the bacterial net surface potential (< 11 mV) is necessary for cellular aggregation to occur in Nutrient broth.
In summary, the current research determined that short-time scale cellular aggregation of S. aureus is dependent on cell surface potential, which is governed by extracellular pH and salt concentration. At a higher pH, the strongly negatively charged cells are fully dispersed, whereas at a lower pH, cells with a lower net surface charge aggregate. The aggregation is amplified by the addition of salt due to cell surface charge screening, or proteins due to bacteria bridging or depletion. Our experiments revealed that the intrinsic hydrophobicity of S. aureus is an important bacteria feature contributing to cellular aggregation. S. aureus cellular hydrophobicity remained consistent despite variations in pH and salt concentrations. Additionally, our XDLVO model estimated that the hydrophobic interactions are about 2 times larger than the Van Der Waals attractive interactions. Upon decrease in net cellular negative potential, the attractive hydrophobic interactions prevail over the electrostatic repulsive interactions, hence leading to immediate aggregation. A similar surface potential driven aggregation behavior observed with E. hirae but not with P. aeruginosa suggests that bacteria with a lower net surface potential are more susceptible to aggregation. The impact of aggregation on the resistance of S. aureus to QACs is significant. The time-to-kill is 20 fold greater at low pH when aggregates are formed as compared to high pH where cells are dispersed. This result is in agreement with previous reports 6 suggesting that bacterial aggregation provides a microenvironment in which innermost cells are protected from antimicrobial agents. Interestingly, S. aureus is known to acidify its extracellular medium as it multiplies in a closed batch system 26 . Moreover, S. aureus is often Figure 3. Cellular aggregation is reversible and specific to bacteria with a lower cell potential. (a) Reversibility of S. aureus aggregation in water with 0.015% and 0.5% NaCl and in Nutrient broth. (b) Aggregation behavior of E. hirae and P. aeruginosa in water with 0.015% and 0.5% NaCl and in Nutrient broth as a function of pH. Each scale bar is 4 µm. Surface potential of P. aeruginosa and E. hirae suspended in water with 0.015% and 0.5% NaCl. The dotted lines are to provide visual guidance. E. hirae aggregate at low pH (lower cell potential) while P. aeruginosa does not. www.nature.com/scientificreports/ found on host skin 7 which displays a slightly acidic pH when intact. A low pH environment might dissuade the growth of many organisms but may favor survival and formation of auto-aggregated populations of S. aureus cells that are self-protected. This protective phenotype combined with maintained mobility afforded by non-sessile aggregation, is likely to enhance survival and successful colonization of S. aureus under various environmental conditions. The reversible, surface potential-dependent nature of aggregation in S. aureus along other similarly charged bacteria could provide guidance to a design of more effective disinfectants and sanitizers. This paper described aggregation behavior in S. aureus which was previously grown in the conditions required by standard disinfection test methods. Additional work is necessary to determine if similar aggregation mechanisms are applicable to other strains and growth conditions. Indeed, parameters such as bacterial strain, growth phase (exponential vs stationary), growth temperature and stock culture storage temperature (fridge vs room temperature) can impact expression profile of membrane and cell wall associated proteins. These changes in protein expressions can, in turn, affect bacterial surface charge and hydrophobicity 10 ultimately impacting the forces driving the aggregation of S. aureus. Future investigations on these environmental variables will provide guidance on how bacterial aggregation impacts disinfection in both laboratory and real-world settings. Aggregation study. In water with 0.015% NaCl. The overnight culture was centrifuged twice at 10,000 rpm for 10 min. The resulting bacterial pellets were re-dispersed in ultrapure water with 0.015% NaCl (BioXtra) at pH 7. The final OD 600nm of the bacterial dispersion was 0.3.

Materials and methods
In water with 0.5% NaCl. The overnight culture was centrifuged twice at 10,000 rpm for 10 min. The resulting bacterial pellets were re-dispersed in ultrapure water containing 0.5% NaCl (BioXtra) at pH 7. The final OD 600nm of the bacterial dispersion was 0.3.
In nutrient broth. The overnight culture (OD 600nm = 0.3) was used without further dilution or wash step to mimic the culture preparation methods recommended by standard disinfection tests.
For all three media types (water with 0.015% and 0.5% NaCl and Nutrient broth), the final bacterial dispersions were aliquoted into 4 mL samples and were adjusted to appropriate pH levels using either hydrochloric acid or sodium hydroxide. A gentle manual shake was performed to homogenize each sample before analysis (taking place only a few minutes after pH adjustment). Optical imaging and zeta potential measurements were carried out using the same overnight culture. All experiments were performed in triplicates using three independent overnight cultures.
Optical microscope imaging. For the hydrophobicity test, the optical microscopy images were acquired on a Leica TCS SP8 inverted confocal microscope in bright field and transmission mode. All the other images were acquired on an Olympus microscope in bright field and transmission mode. Images were taken with a 100 × objective oil (NA 1.4). The camera used was a Sony α Nex-7 color camera.

Bacteria counts and statistical analysis.
For the image analysis, a 4 µm size threshold was set based on the fact that S. aureus cells typically form singlets, duplets, triplets and quadruplets (made of 4 one-micron cells in a clusters) whose sizes are below 4 µm. Bacteria were counted manually for three independent replicate experiments. Approximately 2000 bacterial cells were counted for each replicate experiment and for each pH value, leading to a total of more than 6000 bacterial cells for each data points. The number of bacteria aggregated in all three replicate experiments were averaged and the standard deviation of the average was used for the error bars (error = Stdv/ √ 3 ). The size measurements of aggregates were obtained with ImageJ using the greatest crossarea length of individual aggregates from top views. The software used for the data analysis was MICROSOFT EXCEL.
Zeta potential and pH measurements. The Zeta potential of bacteria was measured on a Zetasizer Nano Serie 200. All bacterial samples were centrifuged and re-suspended in either 0.015% or 0.5% NaCl aque- SEM imaging. The S. aureus cell aggregates formed in Nutrient broth (pH 7) were observed by SEM. A 2 mL aliquot of the overnight culture was centrifuged at 10,000 rpm for 10 min and the resulting pellet was re-suspended in 1 mL of a 0.5% glutaraldehyde PBS solution and held for 30 min at room temperature. The centrifugation was repeated a second time and the bacterial pellet was re-suspended in 1 mL of sterile MilliQ water. An aliquot of 100 µL of bacteria was then drop casted onto a previously isopropanol washed silicon wafer and air dried overnight. The following day, the cells were fixed for 1 h with 100 µL of a 0.1% glutaraldehyde solution followed by an additional 1 h fixation with 100 µL of a 0.5% glutaraldehyde solution. Finally, the specimens were dehydrated by adding ethanol in a graded series (70% for 6 min, 90% for 6 min, and 100% for 6 min). The wafers were left to air dry and coated with platinum with a Cressington sputter coater before SEM imaging using a JEOL 7500 HRSEM.
Time-to-kill study. First, 5 solutions of 90 mL of Nutrient broth were pH adjusted to 3.1, 5.0, 6.5, 9.2 and 11.0. Each solution was then divided into 3 sets of 3 solutions (3 replicates) each containing 10 mL of pHadjusted Nutrient broth. In the first set, no DDAC (MAQUAT 4450-E didecyldimethylammonium chloride, Pilot chemical) was added; this series consists of untreated controls. A two ppm (final concentration) of DDAC solution was pipetted into the second set. A 20 ppm (final concentration) of DDAC solution was pipetted into the third set. Then, an aliquot of 100 µL of the overnight culture is mixed into each tube to reach a final S. aureus concentration of approximately 6.7 log 10 CFU mL −1 . The samples were incubated at room temperature and were sampled at specified time intervals: 30 s, 1, 3, 5 and 10 min. At the end of each exposure time, an aliquot of 100 µL of each bacterial dispersion was added into 9.9 mL of TAT neutralizer broth (General Laboratory Products) and vortexed for five seconds. The vials containing TAT broth were incubated at 35 °C, 50 rpm for 18 h (MaxQ4000 shaker, model 4342, Thermo Scientific). Viability was assessed by the presence or absence of turbidity developed post incubation. The controls were incubated at room temperature for 10 min and were sampled for microbial enumeration (neutralization in TAT broth followed by serial dilution in phosphate buffer and plating on TSA plates). The CFUs were counted after 24 h of incubation at 35 °C. The untreated controls showed no reduction in the cell population (see Table S3 in Supplements for CFUs), suggesting that pH alone did not significantly impact cell viability. TAT broth was previously shown to be an effective media to neutralize DDAC 21 .
Reversibility study. S. aureus dispersions in water with NaCl and in Nutrient broth were prepared as described in the aggregation study section of the "Materials and methods". For each aforementioned medium, two aliquots of 4 mL of S. aureus dispersions (OD 600nm = 0.3, pH ~ 7) were adjusted to pH 3 and pH 9 using HCl and NaOH, respectively. Then, the pH of the first bacterial dispersion was adjusted to pH 9 and the second dispersion was adjusted to pH 3. All samples were homogenized with gentle manual shake and optical microscope imaging was carried out after each pH adjustment. For each medium, the experiment was conducted in three independent experiments with reproducible results.
Bacterial specificity study. Aliquots (4 mL) of overnight cultures of H. hirae and P. aeruginosa were pHadjusted using either HCl or NaOH. A gentle manual shake was performed to homogenize each sample before optical imaging. Two independent experiments were performed for each pH value with reproducible results.

XDLVO simulations.
Our simulation model is based on an extended DLVO 13,16-19 (XDLVO) model. It relies on the seminal model for studying the aggregation of colloids. In the DLVO model, two main interaction potential are considered: the Van der Waals interaction energy 13 (VdW) and the electrostatic repulsion interaction 16 . The XDLVO model incorporated two additional interactions: an attractive hydrophobic interaction 13,17 and repulsive steric interactions 13 . The energy landscape was computed and the energy minimum was extracted. From the value of the minimum of energy, the percentages of bacterial populations either in singlets or in aggregates were calculated 18,19 (see the Supplementary information section for more details).

Data availability
All data are available from the corresponding author upon request.

Code availability
All codes are available from the corresponding author upon request.