Water immobilization by glass microspheres affects biological activity

We recently reported that the water holding capacity of myofibrillar protein hydrogels could be increased upon addition of small amounts of microparticles, particularly glass microspheres. Glass microspheres were found to decrease the spin-spin relaxation time (T2) of water protons in the gels, which was interpreted as enhanced water binding by the glass. We were thus interested in determining whether the observed effects on water proton relaxation were a direct consequence of water-glass interactions. Here we show how glass microspheres reduce the mobility of pure water, reflected in large decreases in the T2 of water protons, decreases in the self-diffusion coefficient of water molecules, a lower water activity, and strengthening of O-H bonds. Even though glass is considered an inert material, glass microspheres were shown to inhibit the growth of human embryonic kidney cells, and stimulate or inhibit the growth of leukemia and monocytic lymphoma cells in vitro, depending on dose and time. The germination of alfalfa seeds and the growth of E.coli cells were also inhibited upon exposure to glass microspheres. This work indicates that the properties and behavior of materials, even ones considered inert, can be affected by their size. These observations suggest possible toxicological consequences of exposure to microparticles, but also open us possibilities to affect cellular/organism function via modulation of macromolecular hydration.

the amount of multilayer bound water in the system from the difference between total bound water determined by NMR diffusion experiments and monolayer surface bound water determined from water activity measurements ( Figure S1). Estimates of monolayer surface bound water are relatively straightforward to obtain from fits to moisture sorption isotherm models; however, the more ill-defined multilayer bound water, described as layers of progressively less immobilized water from bound surface monolayer water to bulk free water, is difficult to determine. Here we show a promising approach to determining the amount of this multilayer water, which may help quantify this dynamic component of water binding.
The infrared spectrum of water was also significantly affected by addition of 4μm glass microspheres (Fig. 1C). A statistically significant 19 cm −1 increase (P < 0.05) in the peak wavenumber (K max ) for the O-H stretching vibration was interpreted as a strengthening of the O-H bonds of water based on the relationship between peak wavenumber (K) and the force constant of a bond (f), namely π ≈ f c K (2 ) 2 , where c is the speed of light. On the other hand, the melting point, melting entropy and enthalpy of water, as determined by differential scanning calorimetry, remained unchanged in the presence of 5% (v/v) 4 μm microspheres (not shown).
Since glass microspheres induced decreases in the T 2 of water in myofibrillar hydrogels 2,3 , we decided to try it on systems containing only glass microspheres in deionized water. Figure 2A shows the decrease in the T 2 (measured at 20 MHz NMR field strength) of water protons in a 0.5% xanthan gum solution upon addition of glass microspheres of three different diameters. A large drop in T 2 from ~1500 ms to ~250 ms was observed upon addition of only 1% (v/v) 4μm glass microspheres. The smaller the particles, the greater were the observed effects, in a concentration-dependent manner ( Fig. 2A). The large decrease in the observed T 2 is due to the exchange of the water molecules between two (or more) chemical environments. In systems with chemical exchange, the observed relaxation rate, R 2 (=1/T 2 ), is a sum of the R 2 in the absence of exchange, and an exchange contribution. The exchange component can dominate the observed R 2 values in this situation. Decreases in T 2 measured at 300 MHz field strength were even larger (Table S1).
By plotting all T 2 data as a function of the amount of glass surface area in the system (Fig. 2B), it became obvious that the effect was solely dependent on the available amount of surface area for interaction with water molecules. We interpreted this as water molecules interacting with the glass surface, consistent with infrared spectroscopy results (Fig. 1C). The spin-lattice relaxation time (T 1 ) of water followed similar trends ( Figure S2), but the magnitude of the decrease was smaller. As expected, decreases in T 2 were correlated to decreases in D ( Figure S3A), where a decrease in D is associated with an increase in the amount of bound water. This increase in the amount of bound water was correlated with an increase in the monolayer amount of bound water using water activity measurements ( Figure S3B). T 2 measurements are ideal for the study of surface interactions of water and can provide insight on chemical exchange 7 . Results obtained in this study are consistent with atomic scale molecular dynamics computer simulations and experimental results of water molecules on amorphous silica surfaces 8,9 . These studies revealed the arrangement and interactions of water molecules at the silica interface and their effect of NMR relaxation, water diffusivity and IR spectra.
To further characterize the observed water interaction with glass surfaces, we carried out relaxation dispersion measurements by pulsed NMR. The R 2 values obtained at different inter-pulse spacings (τ CP ) will be constant if there is no chemical exchange, and will vary with τ CP if there is chemical exchange between bound and free water. A plot of the R 2 of water protons for our three systems as a function of 1/τ CP indicated exchange of water protons between the surface of the glass and the bulk ( Figure S4).
T 2 values decreased linearly with increasing temperature for suspensions of glass in 0.5% xanthan gum ( Figures S2 and S5), while the opposite was observed for the T 2 of pure water and T 1 values ( Figures S2 and S6). Again, this effect was solely dependent on the amount of glass surface area present in the suspensions ( Figure S7). Such T 2 thermal behavior has been previously reported for pure water and silica-water systems where observed T 2 values are dominated by exchange contributions 7,10 . The T 1 behavior is also consistent with previous reports 7, 10 .
To eliminate the possibility of a paramagnetic impurity being present in our sample and affecting results, we centrifuged the samples containing the glass microspheres and measured T 2 of the supernatants. The values were not significantly different from those of deionized water (Table S2). We were also concerned about pH-related effects and thus proceeded to acid wash the glass microspheres to determine if the soda-lime nature of the glass influenced pH and indirectly water mobility at the glass surface. Results shown in Table S1 show that acid washing did not affect the measured T 2 of water. We also measured the surface charge of the glass microspheres, which was ~−40 mV in the pH range 2-8, and did not change with acid washing ( Figure S8). Maybe the presence of this surface charge explains the apparent lack of aggregation of these glass microspheres at the low concentrations used, as determined by static light scattering ( Figure S9).
Since water is such an integral part of living biological matter, subtle changes in its dynamics could possibly have significant effects on structure and function of biological systems. We thus proceeded to explore the effects of glass microspheres on the growth of human embryonic kidney cells, leukemia and monocytic lymphoma cells, E.coli, and on the sprouting of alfalfa seeds.
The growth of embryonic kidney cells after 24 hr was severely inhibited by the presence of even only 0.5% (v/v) 4 µm glass microspheres in the growth medium (Fig. 3A). The growth of both the leukemia and lymphoma cells was also inhibited, although less drastically, above 1% (v/v) levels. Surprisingly, though, cell viability was enhanced for the acute promyelocytic leukemia cell line (KG1a) after 24 hr exposure to growth medium containing 0.5% (v/v) glass microspheres (Fig. 3A), but not for cell leukemia (Jurkat T) or monocytic lymphoma (U937) cell lines. This would indicate a possible activation of tumor cell growth for certain cell lines at specific levels of 4 µm glass microsphere exposure. The stimulation effects were, however, transient, and the activation of KG1 was not observed after 48 hr of exposure to the glass microspheres ( Figure S10B). The kidney cells remained fully inhibited regardless of incubation time. The mechanism responsible for the different sensitivity of the cells to glass microsphere addition remains unknown.
The germination of alfalfa seeds was strongly affected by the presence of 4 µm and 50 µm glass microspheres in the water, both in terms of the number of sprouts (Fig. 3B and the number of leaf pairs (Figures S11) (P < 0.001). This effect was concentration dependent, with saturation reached at 0.5% (v/v) addition levels ( Figure S12). Acid washing of the glass microspheres did not affect any of these results either. We were also concerned that glass was binding minerals, making them less accessible for sprout growth. We therefore sprouted seeds in a standard plant tissue culture growth medium as well. The effects were like those in deionized water ( Figure S13). The growth of the prokaryotic organism E. coli was also significantly (P < 0.05) affected after 4 hr exposure during growth in tryptic soy medium in the presence of 4um glass microspheres (Fig. 3C). It is possible that water binding could be affecting the hydration of structural proteins, enzymes, membranes and other supramolecular structures in biological systems, which affects their function 4,11 .
Glass microspheres are widely used in the powder composites industry and as additives in paint and cosmetics for shine. Glass is usually considered an inert substance. However, our results suggest that a reduction in the size of glass in the low micrometer range might have an effect on biological structures and their function, possibly mediated through interactions with water. It is plausible that microparticles derived from many materials (Table S3)  Glass Microsphere Size Distribution Analysis. The particle size distributions of the glass microspheres were measured using a light scattering device (Mastersizer 2000, Malvern Instruments Ltd., UK) equipped with a small volume sample dispersion unit (Hydro 2000SM). Samples of each of glass size were first mixed in a separate beaker of DI water to ensure minimal clumping. This mixture was then added to the DI water in the dispersion unit. Stirring was set at 1600 rpm and a refractive index of 1.52 was used to determine each particle size distribution. Analyses were carried out in triplicate. NMR Spectroscopy. NMR relaxation experiments were collected on a Bruker AVANCE spectrometer with a 1 H operating frequency of 300 MHz, using a 5 mm BBFO probe. NMR diffusion experiments were collected on a Bruker AVANCE spectrometer with a 1 H operating frequency of 600 MHz, using a 5 mm TXO probe. The sample temperature was regulated at 25.5 ± 1 °C unless otherwise specified. In all cases, spectra were baseline corrected with a polynomial function prior to integration of the water peak (I water ). T 1 measurements were acquired using a saturation recovery sequence. Generally, the water resonance was saturated for 10 seconds using a 50 Hz continuous wave decoupling field followed by a relaxation delay period (τ) and a pulse to measure the signal intensity after relaxation. Ten spectra, with relaxation delays (RD) ranging from 0.1 to 8.0 sec, were collected in a random order. T 1 values were calculated via a non-linear-least-squares fit of: where I f and α are fitted parameters representing the water peak area at infinite relaxation delay and a correction factor for imperfect saturation pulses, respectively. T 2 measurements were acquired using a CPMG pulse sequence. The spacing between consecutive 180° pulses was fixed at 2 msec for all experiments in order to mitigate any difference arising from exchange effects. Nine spectra, with relaxation delays ranging from 0.032 sec to 0.512 sec were collected in a random order. T 2 values were calculated via a non-linear-least-squares fit of: water 0 2 where I 0 is a fitted parameter representing the water peak area at zero relaxation delay. T 2 relaxation dispersion measurements (aka CPMG-RD) were acquired using a CPMG pulse sequence in which both the number of 180° pulses and the spacing between them (τ cp ) were jointly varied such that the total relaxation time (BigT) was always 0.192 sec (for 4 micron and 7-10 micron samples) or 0.384 sec (for 30-50 micron samples). Twenty-seven spectra, with τ cp values ranging from 1 msec to 96 msec, were collected in a random order; also collected was one spectrum with zero relaxation delay (I noCP ). For each τ cp value, the corresponding T 2 (τ cp ) was calculated by:  2 where s(0) is the intensity in the absence of encoding/decoding gradients, D is the diffusion coefficient, γ is the gyromagnetic ratio, g is the gradient strength, δ/2 is the duration of each gradient pulse, Δ is the overall diffusion time, and τ is a small correction factor associated with the BPPSTE sequence. The integral of the water resonance (S(g)) was measured for 35 different gradient strengths ranging from 2.6 to 32.3 G/cm. The diffusion coefficient of water was calculated from a non-linear-least-squares fit of these intensities to eq. [4], with experimental parameters held fixed at their respective values of δ = 2 msec, Δ = 50 msec, and τ = 0.22 msec. S(0) was not fixed during the fitting process.
Water Activity. Glass microspheres were dried at 120 °C for 48 hours and cooled in a desiccator prior to testing to ensure that any residual moisture was removed from the surface of the glass. Approximately 3 g of each particle size range were transferred into a disposable sample trays and water activity was measured using an Aqua Lab Dew Point Water Activity Meter 4TEV (Decagon Devices Inc., Pullman, WA). A known amount of DI water was added drop wise to the glass microspheres and the samples were covered and left to equilibrate for 30 min before each measurement at 25 °C. Water activity vs. water content data was fitted to the GAB model using GraphPad Prism 5.0 software (La Jolla, CA, USA) to derive estimates of monolayer bound water. Differential Scanning Calorimetry (DSC). DSC was performed on three replicates of DI water, as well as a 0.05 volume fraction dispersion of the 4 µm glass microspheres in DI water using a DSC 1 instrument (Mettler-Toledo, Mississauga, ON, Canada). The glass bead and water mixtures were mixed immediately prior to testing. 4-6 mg of sample was placed in an aluminum DSC pan, capped and hermetically sealed. Samples were cooled from 25 °C to −25 °C and subsequently heated to 40 °C, at a rate of 5 °C/min. Peaks were analysed using Star Software (Mettler-Toledo) provided with the DSC instrument.