Fluorescence Spectroscopy Analysis of the Bacteria–Mineral Interface: Adsorption of Lipopolysaccharides to Silica and Alumina

We present here a quantification of the sorption process and molecular conformation involved in the attachment of bacterial cell wall lipopolysaccharides (LPSs), extracted from Escherichia coli, to silica (SiO2) and alumina (Al2O3) particles. We propose that interfacial forces govern the physicochemical interactions of the bacterial cell wall with minerals in the natural environment, and the molecular conformation of LPS cell wall components depends on both the local charge at the point of binding and hydrogen bonding potential. This has an effect on bacterial adaptation to the host environment through adhesion, growth, function, and ability to form biofilms. Photophysical techniques were used to investigate adsorption of fluorescently labeled LPS onto mineral surfaces as model systems for bacterial attachment. Adsorption of macromolecules in dilute solutions was studied as a function of pH and ionic strength in the presence of alumina and silica via fluorescence, potentiometric, and mass spectrometry techniques. The effect of silica and alumina particles on bacterial growth as a function of pH was also investigated using spectrophotometry. The alumina and silica particles were used to mimic active sites on the surface of clay and soil particles, which serve as a point of attachment of bacteria in natural systems. It was found that LPS had a high adsorption affinity for Al2O3 while adsorbing weakly to SiO2 surfaces. Strong adsorption was observed at low pH for both minerals and varied with both pH and mineral concentration, likely in part due to conformational rearrangement of the LPS macromolecules. Bacterial growth was also enhanced in the presence of the particles at low pH values. This demonstrates that at a molecular level, bacterial cell wall components are able to adapt their conformation, depending on the solution pH, in order to maximize attachment to substrates and guarantee community survival.


Adsorption Measurements
An experiment was carried out to measure the amount of alumina / silica solid adsorbed onto the LPS polymer, as discussed in the manuscript. These adsorption measurements were carried out with a constant concentration of LPS (10 2 wt %)

Calculation of AmNS Polymer Loading
The concentration of AmNS in labelled polymer was determined from UV absorbance measurements. The dissolved polymer gave overlapping absorbance at many wavelengths to the AmNS label however was found to absorb differently at 318 nm (Fig. ESI 3). A serial dilution of the AmNS label in water was carried out (Fig. ESI 4) and the molar extinction coefficient determined from a Beer Lambert plot of the absorbance at 318 nm ( Fig. ESI 5).  Measured viscosity values were expressed in terms of reduced (η red ) and inherent (η inh ) as flows.

Equation S1
Equation S2 Equation S3 Equation S4 Equation S5 Equation S6 Where, t r , t p , and t s represent the relative efflux time, the efflux time of aqueous polymer solution and the measured efflux time of solvent, respectively. And ρ r , ρ p and ρ s represent the relative density, the density of aqueous polymer solution and the density of solvent, respectively. Whereas, η r is the relative viscosity of aqueous polymer solution, η sp is the specific viscosity, and C ρ is the concentration (in mass per volume) of aqueous polymer solution.
Plots of η sp / C ρ and ln η r / C ρ as a function of polymer concentration were constructed as shown in Fig ESI 8

Fluorescence of AmNS
Fluorescence label AmNS shows slight pH responsiveness as it is reduced in acidic solutions as shown in Fig. ESI 10. However there is no further change in fluorescence intensity between pH 7 or 11. However when the AmNS label was used to functionalise the LPS polymer it became increasingly desensitised to the pH of solution with only slight variation in the emission intensity. The fluorescence lifetime (Fig. ESI 12) of AmNS and AmNS-LPS was also recorded however was not reported in the manuscript as the lifetime of the labelled polymer did not vary greatly upon binding to a mineral surface.
AmNS (10 -5 wt %) showed a fluorescence intensity decrease at pH 3, 7 and 11, which was fitted to a single exponential decay with a χ 2 ~ 1. At pH 3, 7 and 11 the lifetime was determined as 6, 10 and 11 ns respectively. Analysis of the anisotropy decay of the probe resulted in an extremely short correlation time (ca. 0.1 ns) which reflects free rotation in solution and was not affected by the pH of the solution. The AmNS probe was also mixed with silica (as reported in the manuscript In Fig. 2) and anisotropies recorded at pH 3, 7 and 11 ( Fig. ESI 12). From this we can concur that any effects seen in the AmNS-LPS anisotropy data come from the macromolecule's changing environment and not from the fluorescence label itself.  The fluorescence lifetime data of the AmNS-LPS (Fig ESI 14) show it increases with pH (pH 5 -10) whilst the anisotropy decreases with pH (the major transition occurring from pH 3 -6). The anisotropy decay data indicates pH increases above pH 3 lead to an expansion of the LPS molecule (leading to the measured decrease in dynamic motion) 1 . Previously published work has suggested the hydrodynamic radii of LPS grows from approx. 400 to 200 nm between pH 3 and 9 2 .

Additional Polymer Adsorption Properties
This manuscript describes the binding of LPS to mineral surfaces however previous research carried out in this area have identified a range of polymer materials capable of binding to mineral surfaces. In a previous publication we described the behaviour of poly(acrylic acid) binding onto a calcite surface 3 . These labelled polymers (prepared via copolymerisation with acenaphthylene as opposed to functionalised with AmNS) were also exposed to silica and alumina particles as described in this publication, and far greater adsorption of the PAA chains was observed than was seen in the LPS ( Fig  ESI 14). However the molecular weight difference between the polymers may mean this is not in itself a strictly valid comparison. In this case it was observed that nearly 100% of the PAA adsorbed onto the alumina at pH 3, whilst only 20% adsorbed onto the silica. Conversely only 20% of the LPS adsorbed onto the alumina and 10% onto the silica.
ICP isotherms were also carried out to confirm this (mirroring data shown in Fig. 6 and 7).  In conclusion this shows Poly(acrylic acid) had a high adsorption affinity for Al 2 O 3 , in contrast to the weak interaction on the SiO 2 surface. Strong adsorption was observed at low pH for both minerals. Fitting was carried out only on decay data (i.e. all scattered light (Anisotropy > 0.1) was discounted for fitting purposes). The details of these fits (A, B, χ 2 representing goodness of fits) are shown in Table ESI 2.    Equation 1 is a single exponential fit used to describe the decays.

Correlation Time Data
Theoretically improved fits could be carried out using a dual exponential fit (a modification of  Equation 1 where T c = A + R 01 *exp(-i/T 1 ) + R 02 *exp(-i/T 2 )) for increased resolution (as shown in Fig  ESI 22) however these do not offer significantly reduced residuals compared to the single exponential fits.

Proposed Structure of AmNS-LPS
The chemical structur of LPS from E. coli O111:B4 has been proposed by Ohno and Morrison 4 , we have adapted this structure to show our proposed point of AmNS binding.

DOSY NMR Analysis of Polymer Samples
Diffusion NMR studies were carried out in D2O to confirm whether attachment of AmNs-LPS fluorescence tag altered the conformation / size / hydrodynamic radii of the LPS macromolecule. Both pre and post labelled materials were dissolved in D2O and analysed via a 64-gradient pulse DOSY experiment utilising a bipolar LED sequence, with a sine shaped gradient pulses and gradient strengths incremented between 0.28 and 5.19 G mm -1 in 64 steps equally spaced in gradient squared. This technique previously disclosed 5 , allows for accurate indication of macromolecular diffusion, and indicates the hydrodynamic radii of the material is not changing in response to the labelling event. Raw data is shown in Fig ESI 24.

Bacteria Experiments with Alumina / Silica Nanoparticles -Additional Data
Bacterial experiments were carried out where the absorbance (measured by absorption at 570 nm) was used to indicate the number of colony forming units per ml of solution. To do this a calibration curve was prepared where accurate solutions of known cfu were measured -as shown in Fig 7 a. The raw data shown there is also included in Table ESI 3.  Fig 7 also shows the Δ Absorption (Δ abs ) in the bacteria solutions after incubation with 0.2 and 2 wt% slica respectively after 12 hours. Δ abs is determined by comparing the absorption of a solution of the same wt % solid inorganic material with and without the bacteria. From our calibration plot (Table  ESI 3) it is apparent that a density of 10 7 cfu leads to a Δ abs of 0.06 -so values greater than 0.1 can clearly be indicative of high levels of bacterial growth. The results are displayed and discussed within the manuscript however for complete transparency the raw absorption values of the silica and alumina solutions are shown in Table ESI4 and 5 respectively. In each case 6 repeat measurements were used and the average and std. deviations determined from this data (with the exception of pH 5 where 3 repeats were used).  From the Δabsorbance values plotted in Figure 7 it is possible to evaluate the CFU of bacteria that are growing in suspension due to the presence of the particles. We also cultured certain suspensions (pH 2 + pH 10) for both silica and alumina -the results were in good agreement with the calibration curve and are presented in Table ESI 6. The data shows that in the presence of particles there is a significant increase in bacterial growth compared to the original culture -and that this is dependent on the pH of the solution.