Mesoporous Silica Nanoparticles Inhibit Cellular Respiration

We studied the effect of two types of mesoporous silica nanoparticles, MCM-41 and SBA-15, on mitochondrial O2 consumption (respiration) in HL-60 (myeloid) cells, Jurkat (lymphoid) cells, and isolated mitochondria. SBA-15 inhibited cellular respiration at 25–500 μg/mL; the inhibition was concentration-dependent and time-dependent. The cellular ATP profile paralleled that of respiration. MCM-41 had no noticeable effect on respiration rate. In cells depleted of metabolic fuels, 50 μg/mL SBA-15 delayed the onset of glucose-supported respiration by 12 min and 200 μg/mL SBA-15 by 34 min; MCM-41 also delayed the onset of glucose-supported respiration. Neither SBA-15 nor MCM-41 affected cellular glutathione. Both nanoparticles inhibited respiration of isolated mitochondria and submitochondrial particles. Nanoparticles have been suggested for cellular drug (or gene) delivery and other important biological applications.1–3 Consequently, their potential adverse effects on human health have attracted much attention. The term “nanotoxicology” is now commonly used to describe cellular injuries produced by various nanoparticles.2 For example, treatment of T-cells with multiwalled carbon nanotubes (20–40 nm diameter and 1–5 μm length) caused doseand time-dependent apoptosis.4 Water-soluble fullerene nanoparticles can transfer electrons to O2, producing reactive O2 species (ROS) and lipid membrane peroxidation.5 Silica nanoparticles induce mitochondrial oxidative stress and deplete cellular glutathione (GSH).6 Many of these cytotoxic effects can be visually confirmed by atomic force microscopy.7 Cytotoxins block cellular respiration (mitochondrial O2 consumption) by a variety of mechanisms. In toxin-induced apoptosis,8–10 the intracellular process is mediated by a series of cysteine, aspartate-directed proteases, termed caspases. Caspase activation leads to mitochondrial dysfunction, which includes opening of the permeability transition pores, collapse of the mitochondrial membrane potential, and uncoupling of oxidative phosphorylation.11 Other potential deleterious actions include toxin-induced direct mitochondrial injury, inhibition of metabolism, and blocking mitochondrial fuel

Nanoparticles have been suggested for cellular drug (or gene) delivery and other important biological applications. [1][2][3] Consequently, their potential adverse effects on human health have attracted much attention. The term "nanotoxicology" is now commonly used to describe cellular injuries produced by various nanoparticles. 2 For example, treatment of T-cells with multiwalled carbon nanotubes (20-40 nm diameter and 1-5 µm length) caused dose-and time-dependent apoptosis. 4 Water-soluble fullerene nanoparticles can transfer electrons to O 2 , producing reactive O 2 species (ROS) and lipid membrane peroxidation. 5 Silica nanoparticles induce mitochondrial oxidative stress and deplete cellular glutathione (GSH). 6 Many of these cytotoxic effects can be visually confirmed by atomic force microscopy. 7 Cytotoxins block cellular respiration (mitochondrial O 2 consumption) by a variety of mechanisms. In toxin-induced apoptosis, [8][9][10] the intracellular process is mediated by a series of cysteine, aspartate-directed proteases, termed caspases. Caspase activation leads to mitochondrial dysfunction, which includes opening of the permeability transition pores, collapse of the mitochondrial membrane potential, and uncoupling of oxidative phosphorylation. 11 Other potential deleterious actions include toxin-induced direct mitochondrial injury, inhibition of metabolism, and blocking mitochondrial fuel delivery.
Systematic investigation of the cytotoxicity of nanomaterials is essential prior to their medical application. Considering the special behavior at nanoscale, changes in particle material, size, shape, surface area, aggregation, and conjugation are expected to alter nanoparticles' biological effects. MCM-41 nanospheres have been shown to efficiently enter cells and induce significant toxicity. 12,13 This study investigates the effects of MCM-41 and SBA-15 mesoporous silica nanoparticles on cellular bioenergetics (cellular respiration and ATP content). These nanomaterials are currently considered as potential vehicles for drug and neurotransmitter deliveries. [14][15][16] We show that the mesoporous silica nanomaterials affect mitochondrial function, in a manner dependent on their physical properties.
MCM-41 and SBA-15 mesoporous silica nanoparticles were made by supramolecular self-assembly techniques under base-and acid-catalyzed conditions, respectively. [17][18][19] The transmission electron microscopy (TEM) images ( Figure 1) show MCM-41 materials are rather regular spheres (300-650 nm in diameter), while SBA-15 materials are irregularly shaped with dimensions of hundreds of nanometers. The materials were characterized by powder X-ray diffraction (XRD), which showed sharp low-angle Bragg reflections and highly ordered mesostructures (see Supporting Information, Figure S1). The lattice spacing for MCM-41 was ∼44 Å and that for SBA-15 was ∼106 Å. The nitrogen gas adsorption isotherm (NGAI) profiles for both materials showed type IV isotherms with steep capillary condensation steps (see Supporting Information, Figures S2A and S3A), confirming highly ordered mesoporous particles with large surface areas. From the NGAI measurements, MCM-41 and SBA-15 have BET surface areas of 821 and 506 m 2 /g, respectively, and monodisperse pore size distributions with BJH (Barrett-Joyner-Halenda) pore diameters of 31 and 55 Å, respectively (see Supporting Information, Figures S2B  and S3B). From measured lattice spacings and pore diameters, the wall thicknesses of MCM-41 and SBA-15 were 13 and 51 Å, respectively.
Composition of the materials was checked using 29 Si CP-MAS solid-state NMR spectroscopy, which showed Q3 and Q4 silicon sites centered at about -101 and -115 ppm, respectively. The 13 C CP-MAS solid-state NMR spectroscopy did not show peaks corresponding to residual unhydrolyzed ethoxy groups and surfactant templates. These results are consistent with reported physical properties for such materials. [17][18][19] Oxygen concentration was measured from the decay of phosphorescence from a palladium phosphor. 20 Cells (0.5 × 10 6 /mL) were suspended in media plus 2 µM Pd phosphor and 0.5% fat-free bovine serum albumin with and without 500 µg/mL MCM-41 or SBA-15, and placed in closed 1 mL vessels for O 2 measurement. Results are in Figure 2A. The rate of cellular mitochondrial O 2 consumption (k, in µM O 2 / min) is the negative of the slope in a plot of [O 2 ] vs t. For untreated cells, k ) 0.64 (r 2 ) 0.988); for cells treated with MCM-41, k ) 0.66 (r 2 ) 0.990); for cells treated with SBA-15, k ) 0.32 (r 2 ) 0.961, 49% inhibition). The k-values for cell-free media containing MCM-41 or SBA-15 were 0.04 or -0.01, respectively (similar to instrument drift). Thus, the interaction between nanoparticles and O 2 or Pd phosphor is negligible.
To measure dependence of I on incubation time with SBA-15, every hour samples were withdrawn simultaneously from four conditions, either untreated or treated with 50, 100, or 200 µg/mL SBA-15. Oxygen measurements were then made every minute alternately on all four samples for about an hour, giving the results of Figure 2C (results for 100 µM not shown for clarity). The incubation with the particles thus continued during the measurements. The k-values for all conditions, with statistical errors from linear fitting, are given in Table 1 and Figure 5 in Supporting Information. For 50 µg/mL SBA-15, the inhibitions I at hours 1-6 were 27%, 36%, 9%, 53%, 0%, and 0%. For 100 µg/mL SBA-15, I-values at hours 1-6 were 25%, 25%, 26%, 68%, 0%, and 0%. For 200 µg/mL SBA-15, I-values at hours 1-6 were 64%, 38%, 34%, 51%, 0%, and 46%. Because the statistical errors in k-values are ∼10% (see Supporting Information), there is a large uncertainty in each calculated I-value. Therefore, we do not consider a big fluctuation in I from hour h to hour h + 1 as meaningful, unless it persists at hour h + 2. In another experiment ( Table 2 in Supporting Information), inhibition was again significantly greater for higher concentrations and disappeared after a few hours at lower concentrations. Thus, the inhibition is timedependent as well as dose-dependent, showing recovery with lower concentrations. Cyanide gave almost complete permanent inhibition; after 2 h, the inhibition averaged >94% ( Table 2 in Supporting Information), showing O 2 consumption was due to oxidations in the mitochondrial respiratory chain.   Figure 3 shows the effect of SBA-15, with lines being the best fits to the three-line function: Here, t 1 is the onset of accelerated respiration; t 2 is just after the time of NaCN addition. Continuity requires a + bt 1 ) c + dt 1 , and c + dt 2 ) e, reducing the number of parameters from seven to five. The rate of respiration, k, is equal to -b for t < t 1 and -d for t 1 e t e t 2 .
A similar set of experiments was carried out with MCM-41 (see Supporting Information, Figure S6). In the absence of glucose, no appreciable increase in respiration rate was observed; the best-fit value of t 1 was 26 min. The k-value (µM O 2 /min) for 26 < t < 310 min was 0.18 (r 2 > 0.981) and for t > 310 min 0.03 (83% inhibition by NaCN). In the presence of glucose without MCM-41, t 1 was 22 min and k for 22 < t < 296 min was 0.40 (r 2 > 0.995) and for t > 296 min 0.02 (95% inhibition by NaCN). In the presence of glucose plus 50 µg/mL MCM-41, t 1 was 32 min and k for 32 < t < 301 min was 0.47 (r 2 > 0.997) and for t > 301 min 0.01 (98% inhibition by NaCN). With 200 µg/mL SBA-15, t 1 was 49 min and k was 0.40 (r 2 > 0.996) for 49 < t < 307 min and 0.05 for t > 310 min (88% inhibition by NaCN). In all conditions with glucose, the k-values were 0.42 ( 0.04 (2.3-fold that without glucose) and reduced to almost zero by NaCN. Thus, MCM-41 had no noticeable effect on respiration rate, but the onset of accelerated respiration was   200 µg/mL, 0.24 (r 2 ) 0.954, 46% inhibition). The inhibitions are similar to those measured for Jurkat cells ( Figure  2B). Figure 4B shows how respiration rate varies with incubation time with SBA-15 at two concentrations. After 1 h, percent inhibition was 36.0 ( 14.6 (mean ( standard deviation) at 200 µg/mL and 25.0 ( 10.7 at 50 µg/mL; these results were similar to those for Jurkat cells, Figure 2C. There was no sign of recovery out to 7 h. Cyanide inhibited respiration almost completely after 1 h; the average percent inhibition was 95.9 ( 3.8. Detailed results are in Table 3 of Supporting Information.
The effect of MCM-41 on HL-60 respiration (10 6 cells per mL) is shown in Figure 4C. We next investigated the effect of nanoparticles on cellular ATP level. ATP was measured using the luciferin/luciferase bioluminescence system. 8,9 Jurkat and HL-60 cells (0.5 × 10 6 /mL) were incubated at 37°C with no addition or with 50 µg/mL SBA-15, 50 µg/mL MCM-41, or 10 mM NaCN. Samples were taken from each condition at 0.25, 0.5, 1, 2, and 4 h for ATP determination. Results for Jurkat cells are shown in Figure 5A. Measured ATP concentrations are plotted vs incubation time; each is the average of three to five determinations, with the error bars giving standard deviations. For untreated cells, the best-fit line, shown dashed, had a slope (in nanomoles ATP per 0.5 × 10 6 cells h -1 ) of 0.100 ( 0.080 (essentially zero). The slopes were also zero for cells treated with SBA-15, MCM-41, and NaCN; their values being 0.078 ( 0.134, -0.079 ( 0.090, and 0.033 ( 0.051, respectively. Thus, time-average ATP concentrations were calculated, using six points for untreated cells and five points (for t > 0) for the other conditions. They were (mean ( standard deviation, in nanomoles of ATP per 0.5 × 10 6 cells) 2.36 ( 0.28 for untreated, 2.24 ( 0.32 for SBA-15, 2.53 ( 0.42 for MCM-41, and 1.45 ( 0.44 for NaCN. Thus, neither SBA-15 nor MCM-41 at 50 µg/mL significantly affected cellular ATP over 4 h (other experiments, not shown, extended this to 6 h). Cyanide, on the other hand, decreased cellular ATP by 39% within 1 h. Similar results were noted with HL-60 cells ( Figure 5B), although statistical errors were somewhat larger than those for Jurkat cells. Since the slope was close to zero (0.179 ( 0.063 nanomoles of ATP h -1 ), time averages were calculated, using six points for untreated cells and five points for the other conditions. Average cellular ATP levels (mean ( standard deviation, in nanomoles of ATP per 0.5 × 10 6 cells) were 2.14 ( 0.33 for untreated cells, 2.31 ( 0.77 for SBA-15, 2.23 ( 0.66 for MCM-41, and 1.33 ( 0.38 for NaCN. At 200 µg/mL, SBA-15, but not MCM-41, significantly diminished cellular ATP ( Figure 6A). For untreated HL-60 cells the average ATP level over time (mean ( standard deviation in nanomoles of ATP per 0.5 × 10 6 cells) was 0.725 ( 0.054. For MCM-41 treated cells the ATP level deviated by more than the standard deviation from the level for untreated cells only once (at hour 4), implying a random error, rather than a real decrease. Including all points, the average ATP level for MCM-41 treated cells was 0.661 ( 0.116, equal to that for untreated cells within standard error. For SBA-15 treated cells, the ATP level was 23% lower at hour 1 (p < 0.04), 41% lower at hour 2 (p < 0.02); decreases at hours 3-6 were 24%, 14%, 28%, and -2% (i.e., zero). The average decrease between 1 and 5 h was 27%, with full recovery at hour 6. These results parallel the temporary inhibition of HL-60 respiration by SBA-15 ( Figure 4B).
The above experiment was repeated in the presence of 10 mM NaCN ( Figure 6B). Cyanide decreased cellular ATP in all conditions. The time-average cellular ATP level for t > 0 (mean ( standard deviation in nanomoles of ATP per 0.5 × 10 6 cells) with NaCN alone was 0.434 ( 0.093, with NaCN plus MCM-41 0.377 ( 0.076 (13% decrease, probably not statistically significant), and with NaCN plus SBA-15 0.232 ( 0.102 (47% decrease, which is significant). This shows that ATP hydrolysis (consumption) occurred in all conditions, and even faster with SBA-15 present.
The effects of nanoparticles on cellular GSH were also studied. Jurkat and HL-60 cells (10 6 per mL) were left untreated or treated at 37°C for up to 4 h with 50 µg/mL SBA-15 or MCM-41. Samples were taken every hour for measurement of cellular GSH. The results, shown in Figure  S9 (Supporting Information), suggest that the effect of these nanoparticles on cellular GSH was minimal. In other experiments, cells were incubated at 37°C for 6 h without or with 50 µg/mL SBA-15, 50 µg/mL MCM-41, or 5 mM NEM (which rapidly reacts with GSH, preventing mBBr labeling). Results (mean ( standard deviation for three measurements in each case) are shown in Figure S9C (Jurkat cells) and Figure S9D (HL-60 cells). Although NEM completely blocked mBBr labeling, no significant changes in cellular GSH (p > 0.09) were noted with 50 µg/mL of either SBA-15 or MCM-41. Even at 200 µg/mL, neither nanoparticle had a significant effect on cellular GSH after a 6 h incubation (data not shown).
Finally, we evaluated the effects of nanoparticles on respiration of isolated mitochondria and submitochondrial particles (SMP). Mitochondria or SMP were suspended in 10 mM Tris-Cl (pH 8.2), 250 mM sucrose, 2 µM Pd phosphor, and 0.5% albumin with and without SBA-15 or MCM-41. Concentrations of 50, 100, and 200 µg/mL were used for both nanoparticles. Succinate (10 mM) was added at t ∼30 min to initiate respiration and NaCN (10 mM) was added later on to terminate respiration. The results of one set of experiments are given in Figure 7; when these experiments were repeated, results (k-values) were quite comparable to those in Figure 7. The same was true for the cell respiration experiments.
Cellular respiration includes the delivery of O 2 to mitochondria, the breakdown of reduced fuels (e.g., succinate) with passage of electrons to O 2 (forming H 2 O), and the resulting synthesis of ATP. Toxins frequently target mitochondria and impair oxidative phosphorylation. We have measured cellular respiration and ATP content in the presence of two types of mesoporous silica nanoparticles, SBA-15 and MCM-41 ( Figure 1). Cyanide markedly inhibited respiration and ATP content, establishing that both processes occurred in association with oxidations in the respiratory chain. Thus, reactive oxygen species did not contribute significantly to the observed O 2 consumption. That cellular GSH remained unchanged ( Figure S7 in Supporting Information) in the presence of nanoparticles further confirmed the negligible role of ROS.
Our results show SBA-15 interferes with cellular respiration, whereas MCM-41 has no effect ( Figure 2). This must be connected with the different physical properties of mesoporous silica nanoparticles (e.g., shape, size, and pores, Figures S1-S3). Inhibition is proportional to dosing for concentrations e100 µg/mL but levels off thereafter ( Figure  S4 in Supporting Information). This may suggest that not all mitochondria are equally susceptible to SBA-15 (e.g., mitochondria near the cell membrane may be more vulnerable). The mitochondrial injury also depends nonmonotonically on incubation time with SBA-15 ( Figure 2C). For 50 or 100 µg/mL SBA-15, the impairment of respiration increased with exposure time up to 3 h and then decreased, indicating time-dependent inhibition, with recovery. The time dependence of cellular ATP ( Figure 6A) paralleled that of inhibition of cellular respiration. For 200 µg/mL, the inhibition did not decrease at long times, indicating recovery was impossible for cells exposed to this high concentration.
If the particles are considered to be spheres of diameter 500 nm and density 2.5 g/cm 3 , each particle has a mass of 1.64 × 10 -13 g, and 200 µg/mL corresponds to 1.22 × 10 9 particles/mL. With 10 6 cells per mL, this means 1200 particles per cell. The mean Jurkat cell volume, determined on the Coulter model Z2 (Beckman Coulter, Inc.) was 1.7 × 10 -13 L, which is 2600 times the volume of a nanoparticle. The surface area of a Jurkat cell is calculated, assuming spherical cells with radius 3.4 × 10 -6 m, as 1.5 × 10 -10 m 2 , which is 760 times the cross-sectional area of a nanoparticles. However, it is not known whether the particles are internalized or attached to the cell surface. We did, however, perform a number of experiments to elucidate the mechanism of respiration inhibition.
First, we measured respiration of cells depleted of metabolic fuels (Figure 3). Introduction of glucose led to an increase in respiration rate, but not immediately. The delay (t 1 ) reflects the time required for cellular glucose uptake, generation of reduced metabolic fuels via glycolysis, transport of the fuels into the mitochondria, etc. At a later time (t 2 ) NaCN was injected to shut down respiration. We fitted measured [O 2 ] as a function of t to a three-line function, with the third line horizontal. This involved choosing values of five parameters, including k (rate of respiration), t 1 , and t 2 , to produce the best fit.. With 50 µg/mL SBA-15, t 1 increased by 12 min and k decreased by 44%; the corresponding numbers for 200 µg/mL SBA-15 were 34 min and 56% ( Figure 3A). These results confirm concentrationdependent interference by SBA-15 in glucose-supported respiration. With MCM-41, there was little effect on the k-values ( Figure S6 in Supporting Information). However, there was an increase in the delay time t 1 with increasing MCM-41 concentration, similar to that of SBA-15 ( Figure  3). Thus MCM-41 affects glucose processing but not respiration.
Second, we measured cellular ATP levels in the presence of 50 or 200 µg/mL nanoparticles. There was little change in ATP concentration with time with 50 µg/mL of either nanoparticle, whereas cyanide decreased cellular ATP by 39% within 1 h ( Figure 5). However, 200 µg/mL SBA-15 (but not MCM-41) significantly diminished cellular ATP ( Figure 6A). Compared to untreated cells, the ATP level in SBA-15-treated cells was 23% lower at hour 1, 41% lower at hour 2, and fully recovered at hour 6. This parallels the inhibition profile of SBA-15 on respiration ( Figure 2C). The results with cyanide present of Figure 6B confirm that ATP hydrolysis (consumption) is not significantly altered in the nanoparticle-treated cells.
Third, cellular GSH was measured without and with 50 µg/mL SBA-15 or MCM-41 ( Figure S7A-B in Supporting Information). Measurements were made on samples taken every hour and the results fitted to exponentials. The values of the exponential parameters were about the same for the three conditions in both cell types, suggesting minimal effect of nanoparticles on cellular GSH. Even at a nanoparticle concentration of 200 µg/mL, there was no significant effect on cellular GSH at 6 h ( Figure S7 C-D in Supporting Information). Thus, it is unlikely that significant ROS was generated in the nanoparticle-treated cells.
Finally, we measured the respiration of mitochondria and SMP (prepared by sonication of mitochondria) in the presence of the nanoparticles (Figure 7). For the SMP, the substrate (succinate) is in direct contact with complex II of the respiratory chain, whereas substrate delivery is needed for the mitochondria. (To test this definitively, intact rat liver mitochondria exhibiting a high level of respiratory control are necessary.) Both SBA-15 and MCM-41 were equally toxic to the mitochondria and SMP.
Thus, it appears that the physical properties of nanoparticles determine their effects on cellular bioenergetics. Cyanide-sensitive oxidative phosphorylation is inhibited in cells exposed to SBA-15, which has lower specific surface, larger size, and larger pores than MCM-41. Although its smaller pores may limit entry of essential biomolecules, MCM-41 has only a small effect on cellular respiration. Our results suggest that the minimal effect of MCM-41 is likely due to limited access to cellular mitochondria. The cellular uptake of MCM-41 may be hindered by its high surface area per unit mass.