Ion transport through dimethyl sulfoxide (DMSO) induced transient water pores in cell membranes

Abstract

It is well known that dimethyl sulphoxide (DMSO) increases membrane permeability, which makes it widely used as a vehicle to facilitate drug delivery across biological membranes. However, the mechanism of how DMSO increases membrane permeability has not been well understood. Recently, molecular dynamics simulations have demonstrated that DMSO can induce water pores in biological membranes, but no direct experimental evidence is so far available to prove the simulation result. Using FluxOR Tl⁺ influx assay and intracellular Ca²⁺ imaging technique, we studied the effect of DMSO on Tl⁺ and Ca²⁺ permeation across cell membranes. Upon application of DMSO on CHO-K1 cell line, Tl⁺ influx was transiently increased in a dose-dependent manner. The increase in Tl⁺ permeability induced by DMSO was not changed in the presence of blockers for K⁺ channel and Na⁺-K⁺ ATPase, suggesting that Tl⁺ permeates through transient water pores induced by DMSO to enter into the cell. In addition, Ca²⁺ permeability was significantly increased upon application of DMSO, indicating that the transient water pores induced by DMSO were non-selective pores. Furthermore, similar results could be obtained from RAW264.7 macrophage cell line. Therefore, this study provided experimental evidence to support the prediction that DMSO can induce transient water pores in cell membranes, which in turn facilitates the transport of active substances across membranes.

Keywords::

• Ca2+ permeation
• cell membrane
• dimethyl sulfoxide (DMSO)
• ion permeation
• Tl+ permeation
• water pore

Introduction

Due to its ability to enhance membrane permeability, dimethyl sulfoxide (DMSO) is frequently applied to facilitate the transport of active molecules across the biological membranes (David 1972). It has been shown that transfection of exogenous DNA incubated with DMSO is more efficient than that without any treatment (Li et al. 2006). As a chemical permeability enhancer, DMSO is also used in percutaneous drug delivery (Williams and Barry 2004). Besides, the internalization efficiency of arginine-rich cell penetrating peptides could be significantly enhanced with the help of DMSO (Wang et al. 2010). How does DMSO increase membrane permeability to facilitate the transport of active molecules? The widespread use of DMSO has led to numerous studies and hypotheses about its properties and interactions among the biological organisms. X-ray diffraction measurement of biological membranes has suggested that the thickness of the bilayer membranes decreases with the increase in DMSO concentration (Yu and Quinn 1998). In good agreement with this result, molecular dynamics (MD) simulations have indicated that DMSO induces the membrane thinning and increases the fluidity of the membrane's hydrophobic core (Gurtovenko and Anwar 2007a). MD simulations have also shown that DMSO can cause the membrane to become floppier, which would enhance the membrane permeability and enable the cell membrane to accommodate osmotic and mechanical stresses during cryopreservation (Notman et al. 2006). Moreover, recent MD simulations have demonstrated that DMSO is able to make the hydrophilic headgroups of the lipids shield the hydrophobic tails from the water at the bilayer edge, to yield a transient water pore to allow water molecules readily enter into (Gurtovenko AA, Anwar J. 2007a and 2007b). The formation of the transient water pores dramatically reduces the free energy barriers to permeation of ions through the membrane, so that ions are able to transport across the membrane, avoiding highly unfavorable contacts with the hydrophobic tails in the membrane interior.

Though MD simulations provided detailed information at the atomic resolution on the mechanism of the action of DMSO, it is still lack of direct experimental evidence that supports the ability of DMSO to induce transient water pore in the biological membrane. In the present work, using intracellular ion imaging techniques, we studied the effect of DMSO on Tl⁺ and Ca²⁺ permeation across cell membranes. Our results provided compelling experimental evidence to support the prediction of the MD simulations that DMSO could induce transient water pores in cell membranes.

Materials and methods

Cell culture

CHO-K1 cells were grown in Ham's F-12 nutrient mixture (Invitrogen, Co. Grand Island, NY, USA), while RAW 264.7 cells were cultured in Dulbecco's modified Eagle's medium (DMEM). Both cell lines were supplemented with 10% fetal bovine serum and cultured in a humidified 37°C incubator (5% CO₂). The CHO-K1 cells were passaged twice weekly through exposure to 0.05% trypsin. For the cell imaging experiments, cells were transferred onto glass coverslips pretreated with poly-l-lysine in a culture medium at 37°C incubator (5% CO₂) to improve cell adhesion. After 2–4 days growth, the cells were transferred to a Hanks' balanced salts solution (HBSS, Sigma; in mM): 1.3 CaCl₂, 0.8 MgSO₄, 5.4 KCl, 0.4 KH₂PO₄, 136.9 NaCl, 0.3 Na₂PO₄, 10 D-glucose and 4.2 NaHCO₃. The osmolality of all solutions was 290–310 mmol/kg as measured with a vapour pressure osmometer (Model 5520, Wescor, Ultah, USA)

Cell imaging

For the real-time imaging of Tl⁺ fluorescence, the cells were loaded with a Tl⁺-sensitive fluorophore using the solution provided by the maker (FluxOR Thallium Detection Kit, Invitrogen). The experiments were carried out according to the provided protocols with slight modification. Briefly, the cells were loaded with a loading buffer containing 1% FluxOR reagent, in which the Tl⁺-sensitive dye was included, in the dark for 20–40 min. After loading, the cells were washed out with an assay buffer to remove the extracellular dye. Then, the solution was changed to a stimulus buffer containing 2 mM Tl⁺. The intracellular Tl⁺ fluorescence was measured with a Leica DMI4000B inverted microscope at 2 s intervals for 600–1800 s and analyzed with LAS AF 6000 software (Leica, Germany). DMSO was directly added into the bath solution. The Tl⁺ fluorescent dye has an excitation peak at 490 nm and an emission peak at 525 nm. The effect was evaluated by the calculated FluxOR fluorescence ratio normalized with the basal fluorescence level. All experiments were performed at room temperature (22–24°C). For intracellular Ca²⁺ imaging, the cells were incubated in the culture medium containing 10 μM fluo-3/AM and 0.02% pluronic F-127 (Sigma Chemical Co.) for 30 min. Fluo-3 intensity (480 ± 15 nm excitation, 530 ± 30 nm emission) was monitored every 2 s and was plotted as fluorescent intensity (F_max) with regard to that of the basal calcium level (F₀). The inhibitors for K⁺ channels and Na⁺-K⁺ ATPase were applied to the bath solution after the cells were loaded with the Tl⁺-sensitive fluorophore. Again, all experiments were performed at room temperature (20–25°C).

Data analysis

Data were analyzed statistically by one-way analysis of variance (ANOVA) followed by the Bonferroni test for multiple comparisons. All the data were mean ± SD. The criterion for a significant difference was P < 0.05.

Results

DMSO induced transient Tl⁺ influx in CHO-K1 cells

In order to observe the fluorescent intensity of the individual cells, we did not use a microplate reader to detect Tl⁺ influx as the protocol suggested. Instead, we applied the intracellular ion imaging technique to detect Tl⁺ influx to judge whether DMSO could induce water pores in the cell membrane. Since Tl⁺ is not present in living cells in any considerable amount, there is a strong driving force for the extracellular Tl⁺ to enter into the cells so that even very small amounts of intracellular Tl⁺ can be detected. If DMSO can induce water pores in the cell membrane, it should be efficiently detected by fluorescent measurement of the Tl⁺ influx. The Tl⁺-sensitive dye was loaded into cells, which exhibited very low basal fluorescence in the absence of Tl⁺ (Figure 1A and 1B). Switching to the Tl⁺-containing buffer caused a gradual increase in fluorescence even in cells that were not exposed to DMSO (Figure 1A and 1B), probably due to endocytosis or via some constitutive opening of endogenous ion channels. Approximately 10 s after application of 4% DMSO, a transient increase in intracellular Tl⁺ was observed (Figure 1A and 1B). To obtain reproducible data, all the treatments were conducted after the basal fluorescence level getting stable in Tl⁺-containing buffer (Figure 1C) in the following experiments.

Figure 1. Fluorescence imaging of the DMSO-induced increase in intracellular Tl⁺ in CHO-K1 cells. (A) Representative intracellular Tl⁺ fluorescent images at different time points: 98 s, without Tl⁺; 198 s, after addition of 2 mM Tl⁺; 396 s, immediately after application of 4% DMSO; 456 s, Tl⁺ fluorescent intensity approaching the stable level; 600 s, Tl⁺ fluorescent intensity getting stable. (B) Representative raw data traces for time course of fluorescence changes described in (A). (C) Representative traces for DMSO-induced fluorescence changes. The DMSO was applied after obtaining stable Tl⁺ fluorescent level. Scale bar = 20 μm. F.A.U., fluorescence arbitrary unit. This Figure is reproduced in color in the online version of Molecular Membrane Biology.

In accordance with the result from the MD simulations mentioned above, the effect of DMSO to induce pore formation is concentration-dependent. As shown in Figure 2A and 2B, the fluorescence was transiently increased with the increase in the concentration of DMSO from 0.1–2%. Due to membrane blebbing of the CHO cells induced by high concentration of DMSO (2%, Figure 2A, inset), the fluorescent intensity declined immediately after the transient peak was significantly lower than that of the basal level (Figure 2A). Statistical data showed that the relative fluorescence intensity (F_max/F₀) was increased to 1.21 ± 0.09 for 0.1% DMSO, and 1.29 ± 0.08, 1.51 ± 0.21, 1.99 ± 0.41, 2.05 ± 0.30 and 2.29 ± 0.34 in response to the addition of 0.2%, 0.4%, 0.8%, 2% and 4% DMSO, respectively (Figure 2C). This result indicated that DMSO could induce Tl⁺ influx in the cell at concentrations as low as 0.1%.

Figure 2. Dose-dependent effect of DMSO on intracellular Tl⁺ fluorescence changes in CHO-K1 cells. (A) Representative intracellular Tl⁺ fluorescent images upon sequential DMSO exposures at increasing concentration. 180 s: stable fluorescent level before addition of DMSO; 218 s, 530 s, 840 s and 1242 s: Immediately after application of 0.1%, 0.4%, 0.8% and 2% DMSO, respectively; 500 s, 700 s and 1000 s: stable fluorescent level after DMSO application. Inset: Representative image shows membrane blebbing of the CHO cells induced by 2% DMSO. (B) Representative raw data traces for time course of sequential DMSO exposures at increasing concentration on Tl⁺ influx. (C) Statistic summary on relative Tl⁺ fluorescent intensity (F_max/F₀) upon application of different concentrations of DMSO. The number of independent experiments is indicated on the bar figure, while the number of cells used for statistical analysis is shown in the parenthesis. [DMSO], concentration of DMSO (v/v). (*P < 0.05, **P < 0.001; one-way ANOVA). This Figure is reproduced in color in the online version of Molecular Membrane Biology.

Effect of K⁺ channel and Na⁺-K⁺ ATPase blockers on the DMSO-induced Tl⁺ influx

It has been indicated that Tl⁺ is permeable through potassium channels (Niswender et al. 2008); therefore, Tl⁺ uptake through voltage-sensitive K⁺ channels activated by DMSO in the cells should be considered. Even though CHO-K1 cells are known for very limited expression of endogenous ion channels (Gamper et al. 2005), we still detected the Tl⁺ influx in the presence of a wide-spectrum K⁺ channel blocker tetraethylammonium (TEA) (1 mM) to prevent potential Tl⁺ entry through voltage-gated K⁺ channels. A similar transient increase in Tl⁺ influx was detected in CHO cells upon application of DMSO in the Tl⁺-containing buffer (P = 0.15, Figure 3A), indicating that the endogenous K⁺ channels played no role in this process. Tl⁺ ions are very similar to potassium ions with regard to charge, hydrated radius and mobility in water (Zerahn 1983). To judge if Tl⁺ could enter into the cell by Na⁺-K⁺ ATPase, we detected the effect of DMSO on Tl⁺ influx in the presence of ouabain, a blocker for Na⁺-K⁺ ATPase. There is no significant change between control group and the ouabain-treated group (P = 0.22, Figure 3B), indicating that Na⁺-K⁺ ATPase plays no role in the transient Tl⁺ influx induced by DMSO. Therefore, it is highly likely that the transient increase in Tl⁺ was caused by the influx of Tl⁺ from the DMSO-induced water pores in the plasma membrane.

Figure 3. Effect of K⁺ channel and Na⁺-K⁺ ATPase blockers on the DMSO-induced increase in Tl⁺ transients. (A) Comparison of the effect of DMSO on relative Tl⁺ fluorescent intensity (F_max/F₀) in the absence and presence of K⁺ channel blocker TEA. (B) Comparison of the effect of DMSO on relative Tl⁺ fluorescent intensity (F_max/F₀) in the absence and presence of Na⁺-K⁺ ATPase blocker ouabain. This Figure is reproduced in color in the online version of Molecular Membrane Biology.

DMSO induced Ca²⁺ influx in CHO-K1 cells

Ion selectivity of channels is due to specific interaction between individual ion and the ion selectivity filter in the pore region of the channel (Doyle et al. 1998). A water pore in the plasma membrane should not have selectivity for different ions, i.e., the pore should be permeable to different cations and anions as well. In good agreement with this inference, simulation study has demonstrated that both K⁺ and Cl^- ions could permeate through the DMSO-induced water pore (Gurtovenko AA, Anwar J. 2007b). Under physiological condition, intracellular Ca²⁺ concentration is around sub-micromolar, while extracellular Ca²⁺ concentration is at millimole level. Therefore, if DMSO induces water pores in plasma membrane, the pore-mediated Ca²⁺ influx should also be observed due to the strong driving force for the Ca²⁺. Comparing the three panels in Figure 4A, the intracellular fluorescent intensity of most cells was significantly increased upon DMSO application. The relative intracellular Ca²⁺ fluorescent intensity was increased to 1.64 ± 0.22-fold of the baseline level by 0.5% DMSO (Figure 4B and 4C), giving further evidence that DMSO could induce transient water pores in the cell membrane with no significant ion selectivity. Besides, the DMSO-induced increase in Ca²⁺ influx supports the above result that DMSO-induced Tl⁺ influx is not through K⁺ channels.

Figure 4. Effect of DMSO on Ca²⁺ influx in CHO-K1 cells. (A) Representative intracellular Ca²⁺ fluorescent images before (228 s) and immediately after (232 s) application of DMSO as well as after fluorescence becoming stable (600 s). (B) Representative raw data traces for time course of Ca²⁺ fluorescence changes. (C) Relative Ca²⁺ fluorescent intensity (F_max/F₀) in the absence and presence of 0.5% DMSO. This Figure is reproduced in color in the online version of Molecular Membrane Biology.

Effect of DMSO on Tl⁺ influx in RAW264.7 cells

We hypothesized that the effect of DMSO to induce transient water pores should not be limited to CHO-K1 cells, which is derived from the ovary of Chinese hamster. To demonstrate this point, we tested whether similar results could be obtained from another cell line, RAW264.7, which is a transfectable macrophage cell line with the capacity to form osteoclast-like cells. Similar to the results from CHO-K1 cells, the Tl⁺ fluorescence intensity of RAW264.7 cells was significantly increased upon application of 0.4% and 0.8% DMSO (Figure 5A–C). In contrast to that in CHO cells, DMSO could cause membrane blebbing of the RAW264.7 cells at concentrations as low as 0.8% and consequently lower fluorescent intensity immediately after the transient peak (Figure 5B), suggesting that RAW264.7 cells were more susceptible to DMSO than CHO cells. The relative intracellular Tl⁺ fluorescent intensity was increased to 1.56 ± 0.13 and 2.09 ± 0.21-fold of the baseline level upon application of 0.4% and 0.8% DMSO (Figure 5C), respectively. In spite of the different susceptibility, similar results obtained from the different cell lines with completely different origin and properties might suggest that that transient water pore induced by DMSO in the plasma membrane is a common phenomenon.

Figure 5. Effect of DMSO on Tl⁺ influx in RAW264.7 macrophage cells. (A) Representative intracellular Tl⁺ fluorescent images upon application of 0.4% and 0.8% DMSO. 280 s: stable fluorescent level before addition of DMSO; 344 s: Immediately after application of 0.4% DMSO, 564s: stable fluorescent level after application of 0.4% DMSO, 648 s: immediately after application of 0.8% DMSO, 726 s: Fluorescent level after application of 0.8% DMSO. (B) Representative raw data traces for time course of Tl⁺ influx upon sequential application of 0.4% and 0.8% DMSO. (C) Statistic summary on relative Tl⁺ fluorescent intensity (F_max/F₀) upon application of 0.4% and 0.8% DMSO (*P < 0.05, **P < 0.001; one-way ANOVA). This Figure is reproduced in color in the online version of Molecular Membrane Biology.

Discussion and Conclusion

DMSO is known to enhance the penetration of both hydrophilic and hydrophobic molecules (Williams and Barry 2004). The enhancement of hydrophilic compounds, such as ions, by DMSO is always difficult to explain. There are at least three mechanisms for ion transport across the biological membranes. First, ion transport is primarily mediated by specific proteins, such as ion channels, transporters and pumps. Secondly, ions penetrate the membrane according to the solubility-diffusion theory, which implies that ions partition into the membrane's hydrophobic core and diffuse across the membrane (Bordi et al. 2000). Thirdly, ion is transported via transient water pores in the biological membrane (Jansen and Blume 1995), which helps ions to evade the high-energy Born barrier associated with the solubility-diffusion mechanism. We noticed that the fluorescence levels decreased back to base level after the influx of Tl⁺ ions. A plausible explanation might be that the intracellular Tl⁺ concentration was transiently increased locally, near the sites of the pore, which was decreased as the Tl⁺ ions diffuse in the cytoplasm, resulting in the fluorescence levels decreased back to the base level. Nevertheless, the following pieces of evidence strongly suggest that the ions are transported across the cell membrane through DMSO-induced transient water pores in the plasma membrane:

1. Ion transport through transient water pores is theoretically reasonable. The formation of the transient water pores dramatically reduces the free energy barriers to the permeation of ions through the membrane, so that ions are able to transport across the membrane, avoiding highly unfavorable contacts with hydrocarbon chains in the membrane interior.

2. MD simulations have indicated that the ability of DMSO to enhance membrane permeability is coupled to DMSO-induced water pores in the membranes.

3. MD simulation has directly observed diffusive transmembrane ionic leakage through DMSO-induced transient water pores (Gurtovenko AA, Anwar J. 2007b).

4. Our study showed that DMSO-induced Tl⁺ influx was transient, suggesting that it is unlikely that the ions are transported according to the solubility-diffusion mechanism.

5. DMSO-induced Tl⁺ influx was not affected by blockers of K⁺ channels and Na⁺-K⁺ ATPase, suggesting that the Tl⁺ influx is not through main K⁺ transport proteins in the plasma membrane.

6. DMSO induced transient influx for both Tl⁺ and Ca²⁺, which is in good agreement with the concept that water pores in the plasma membrane should have no selectivity for different ions.

Considering these pieces of evidence, it is difficult to conceive of a simple way other than by a pore-mediated mechanism for the observed DMSO-induced ion influx.

MD simulations indicated that the effect of DMSO is dependent on its concentration. At the concentration in a range of 10–30 mol %, DMSO is able to induce the formation of transient water pores (Gurtovenko AA, Anwar J. 2007a). Our study showed that transient Tl⁺ and Ca²⁺ influxes could be triggered by DMSO at concentrations in a range of 0.1–4% (v/v), approximately 0.02–1.0 mol %. This value is much lower than the prediction of the MD simulations. The time scale for our experiments was in seconds. In contrast, the MD simulations are on the time scale of nanoseconds, which is several orders of magnitude shorter than that of our experiment. It has been shown that different time scales of simulations could profoundly affect the simulation results. For example, MD simulations have shown that on the time scale of 20 ns simulation, 10 mol % DMSO is needed to induce a transient water pore, while on the time scale of 5 ns, 15 mol % DMSO concentration is necessary (Gurtovenko AA, Anwar J. 2007a), implying that a longer time scale of simulations might be necessary to obtain more reliable results. On the other hand, a previous simulation study (Notman et al. 2007) has indicated that at low concentration of DMSO the DMSO density profile shows a low density in the bulk but a significant peak at the water-lipid interface. As the concentration of DMSO is increased, the peak density of DMSO at the interface does not change; while the concentration of DMSO in the bulk solvent increases. This result suggested that DMSO tends to accumulate at the water-lipid interface when its concentration is low. Therefore, at low concentrations, the effect of DMSO to induce pore formation should not be strongly dependent on its concentration. This means that it is not easy to determine the lower limit of DMSO to induce pore formation. As DMSO, at low concentration, needs time to condense onto the bilayer surface, longer periods of MD simulations are necessary to simulate the induction of pore formation. In addition, MD simulation has demonstrated that the formation of transient water pores and the consequent ion leakage can be induced and be driven by transmembrane ionic charge imbalance, an inherent feature in living cells (Gurtovenko and Vattulainen 2007). Besides, cell membranes contain a large number of membrane proteins, while the MD simulation system is on protein-free lipid membrane. Considering these differences, it is reasonable that transient water pores could be induced by DMSO in the cell membrane at concentration as low as 0.1% (v/v).

In fact, the pore-mediated transport has been shown in many biological membrane systems. Several previous studies suggested that the most likely mechanism for ion transport across the membrane is permeation through water-filled pores in the membrane (Deamer and Bramhall 1986). It has been shown that pores can be induced in the biological membranes by electric field (Melikov et al. 2001). Pores have also been observed in MD simulations of phospholipid bilayers as transient structures in the bilayer membranes (Tieleman et al. 2003). In addition, a lipid flip-flop translocation of lipid molecule from one bilayer leaflet to the opposite leaflet has been observed to be assisted by water pores (Kandasamy and Larson 2006). These transient water pores are important for chemical control of the lipid distribution across the cell membranes, as the simulation studies have shown that lipids use pores to flip-flop from one leaflet to the other (Tieleman and Marrink 2006). Thus, DMSO-induced pore formation is not only reasonable but also has important biological functions. Collectively, this study provided experimental evidence to support the prediction of the MD simulation that DMSO could induce transient water pores in the cell membranes to facilitate membrane transport. And this ability of DMSO might explain the significant enhancement in the permeability of membranes to hydrophilic molecules induced by DMSO.

Acknowledgements

This work was partly supported by the National Science Foundation of China (31070741, 81000560 and 81170727). We are grateful to Eye Institute of Xiamen University for providing assistance in the experiment.

Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.