The influence of surface chemistry on the kinetics and thermodynamics of bacterial adhesion

This work is concerned with investigating the effect of substrate hydrophobicity and zeta potential on the dynamics and kinetics of the initial stages of bacterial adhesion. For this purpose, bacterial pathogens Staphylococcus aureus and Escherichia coli O157:H7 were inoculated on the substrates coated with thin thiol layers (i.e., 1-octanethiol, 1-decanethiol, 1-octadecanethiol, 16-mercaptohexadecanoic acid, and 2-aminoethanethiol hydrochloride) with varying hydrophobicity and surface potential. The time-resolved adhesion data revealed a transformation from an exponential dependence to a square root dependence on time upon changing the substrate from hydrophobic or hydrophilic with a negative zeta potential value to hydrophilic with a negative zeta potential for both pathogens. The dewetting of extracellular polymeric substances (EPS) produced by E. coli O157:H7 was more noticeable on hydrophobic substrates, compared to that of S. aureus, which is attributed to the more amphiphilic nature of staphylococcal EPS. The interplay between the timescale of EPS dewetting and the inverse of the adhesion rate constant modulated the distribution of E. coli O157:H7 within microcolonies and the resultant microcolonial morphology on hydrophobic substrates. Observed trends in the formation of bacterial monolayers rather than multilayers and microcolonies rather than isolated and evenly spaced bacterial cells could be explained by a colloidal model considering van der Waals and electrostatic double-layer interactions only after introducing the contribution of elastic energy due to adhesion-induced deformations at intercellular and substrate-cell interfaces. The gained knowledge is significant in the context of identifying surfaces with greater risk of bacterial contamination and guiding the development of novel surfaces and coatings with superior bacterial antifouling characteristics.

where Ne is the effective number of impacts on a specific specimen and NA,B is the number of coemitted ions A and B in their coincidental mass spectrum. Using Equations S1 to S4, Ne is obtained where NA and NB are the peak areas of ions A and B, respectively. The fractional coverage of specimens (K) on the surface area is given by: where N0 is the total number of impacts. The effective number of impacts on a specimen (Ne) does not depend on the ionization probabilities and detection efficiencies of ions A and B. Thus, for surface objects that are larger than the emission volume; the fractional coverage can be calculated using the coincidental process 2,3 .
The thiol-related ions SH -, C2HS -, and AuSat m/z 33, 57 and 229, respectively, were chosen to calculate the surface coverage of coating on gold surfaces, K (%) using Equations from S3 to S5. Table S1 shows the surface coverage of linear-chain thiols used in this study. 2. Theory. For many colloidal systems (e.g., bacterial suspension) having repulsive interactions with a substrate, the particle adhesion (e.g., adsorption and deposition) process is first-order, which can be modelled as:

S7
where N is the surface concentration of particles (number of particles/m 2 ), nb is the bulk concentration of particles (number of particles/m 3 ), t is time (s), L(N)=1-N/N0 is the Langmuir blocking function, N0 is the saturation surface concentration, and k is the adsorption rate constant (m/s). The solution of the differential equation describes an exponential dependence between the surface concentration and time: the adsorption constant, k, is strongly dependent upon the energetics between the adhering particles and the substrates, and also flow velocity and geometry when the suspending medium is not quiescent.
For colloidal adhesion processes with diffusion-dominant characteristics, which generally occurs when the total interaction between colloids and the substrate is attractive, the number of adhering colloids has a square root dependence on time, t 4,5 : where D∞ is the constant diffusion coefficient in the suspension.
The interactions between bacteria and a substrate are often modeled by the DLVO theory, which assumes the superimposition of the van der Waals interactions and double-layer electrostatic For some cases, additional corrections, which are important at short ranges, such as acid-base interactions 7 , hydration interactions 8 , and hydrophobic interactions 9 have been included in the total S9 energy considerations. However, the kinetic processes are generally governed by the energy barrier occurring at intermediate distances, which is weakly influenced by short-range interactions. As such, it is not unreasonable to neglect these short-range interactions while modeling the kinetics of initial stages of bacterial adhesion.
For cases where the electrostatic interactions are short-ranged in comparison to the size of bacterium, and electrolytes are monovalent, the double-layer energy between a bacterium and flat substrate with different charge densities can be expressed as follows 10,11 : where R is the radius of bacterium, 1/κ is the Debye length (m), and h is the separation between the bacterium and surface.
For the geometry of a spherical bacterium near a flat substrate, the non-retarded van der Waals interactions can be described in terms of the Hamaker constant, AH 12,13 : (S11) where AH is the Hamaker constant, which typically ranges between 10 -21 and 10 -20 J for organic molecules in water 14,15 . The non-retarded Hamaker constant can be predicted either by the Lifshitz theory using the dielectric constants and refractive indexes of the materials and dispersing media 16 or from the measured surface tension of bacteria and the minimum equilibrium distance between two molecules, which is 0.165 nm 17 .
Bacterial adhesion process involves not only substrate-bacteria interactions but also bacteriabacteria interactions. DLVO interactions between a bacterium and a bacterium can be obtained by the multiplying DLVO interactions between a planar substrate and a bacterium with the Derjaguin factor 18 . However, another critical issue that must be considered here is bacterial deformations due to the domination of attractive van der Waals interactions over repulsive double-layer forces at short distances 19 . In this case, the non-retarded van der Waals interaction energy between a wall and a truncated sphere with a deformation (indentation depth), δ can be expressed as 20,21 : where x is the distance between the most deformed point of the truncated sphere and the wall. In addition, it is essential to consider elastic energy that must be spent to change in the shape of the S10 bacteria, which is roughly equal to 22 : * √ (S13) where E* is the reduced Young's modulus and R the particle radius. Our high-resolution images revealed that the mean contact diameter due to the bacterial adhesion is 232.6 ± 25.4 nm for S. aureus (Fig. S5), corresponding to 18.5 ± 4.0 nm indentation depth via the Cord theorem. This value is comparable with the deformation of S. aureus on glass measured by Busscher and coworkers 22 . The reduced elastic modulus of S. aureus is ~47 kPa 22 .
Overall, our modified DLVO model for the energetics of bacterial approach to a substrate includes three terms: double-layer interactions, van der Waals interactions, and elastic deformation energy. Since when deformations start to occur, the distance between bacteria and surfaces is small; double-layer interactions are much weaker than van der Waals interactions. Hence, the standard  (Table S2). kT (J) = 4.11 × 10 -21 ; v e (s -1 ) = 3.00 × 10 15 ; h (J·s) = 6.63 × 10 -34 S13 4. Screening of antimicrobial activity. The antibiotic characteristics of thiol-coated substrates were investigated to determine whether bacterial adhesion and antimicrobial activity acted in concert or not for the model substrates selected. Bacterial suspensions of S. aureus and E. coli O157:H7 were grown in contact with thiol-coated substrates for 24 h at room temperature.
Bacterial suspensions grown without substrates were treated as negative controls for each pathogen. The numbers of bacteria remaining in suspension following incubation was assessed by pour plate method. Each condition was replicated four times.