Membrane potential controls adipogenic and osteogenic differentiation of mesenchymal stem cells

Control of stem cell behavior is a crucial aspect of developmental biology and regenerative medicine. While the functional role of electrophysiology in stem cell biology is poorly understood, it has become clear that endogenous ion flows represent a powerful set of signals by means of which cell proliferation, differentiation, and migration can be controlled in regeneration and embryonic morphogenesis.

We examined the membrane potential (Vmem) changes exhibited by human mesenchymal stem cells (hMSCs) undergoing adipogenic (AD) and osteogenic (OS) differentiation, and uncovered a characteristic hyperpolarization of differentiated cells versus undifferentiated cells. Reversal of the progressive polarization via pharmacological modulation of transmembrane potential revealed that depolarization of hMSCs prevents differentiation. In contrast, treatment with hyperpolarizing reagents upregulated osteogenic markers.

Taken together, these data suggest that the endogenous hyperpolarization is a functional determinant of hMSC differentiation and is a tractable control point for modulating stem cell function.

Harnessing the potential of stem cells for applications such as wound healing and tissue regeneration is a tantalizing yet daunting task. During embryonic development and tissue regeneration, two events during which stem cells actively proliferate and differentiate, a wealth of literature suggests that biophysical signaling plays a critical role (reviewed in [1], [2]). For example, during both limb bud development in amphibians and mammals and spontaneous limb regeneration in adult urodeles, limbs establish highly localized endogenous electric fields, which, if disrupted by an exogenous current, results in deformed structures [3]–[9]. There is evidence that application of exogenous electric fields can augment regeneration or even induce a degree of regeneration in normally non-regenerative species [10]–[12].

Recent application of molecular genetics to this field has confirmed the crucial role of bioelectric signals in regenerative processes, identified the ion transporters responsible for generating instructive ion flows, and characterized downstream changes in gene expression. For example, in Xenopus tail regeneration [13], the vacuolar H+-ATPase pumps a strong H+ flux through the cell membrane of regeneration bud cells. This very early process controls downstream changes in cell proliferation, regeneration-specific gene expression, and axonal patterning, and is both necessary and sufficient for regeneration of spinal cord, muscle, and vasculature in the tail.

Underlying the complex processes of tissue development and regeneration are individual cellular events such as proliferation, migration, and differentiation, which themselves may be regulated by biophysical signaling. For example, in a study of cell cycle regulation in fibroblasts, activity of the Na+-H+ exchanger NHE1 caused an increase in intracellular pH, which regulated the timing of the cell cycle G2/M transition and resulted in cell proliferation [14]. In a study of nerve growth cone migration, Rho GTPases mediated growth cone steering in electric fields, linking membrane receptor signaling pathways to spatial regulation of the cytoskeleton [15]–[18]. In a corneal wound healing model, endogenous electric fields regulated both cell migration [19] and the orientation and frequency of cell division [20]. Phosphoinositide 3-kinase (PI(3)K) and Src signaling pathways mediated this electrotactic response. Disruption of the gene for PI(3)Kã resulted in diminished electrotactic migration, while disruption of the gene PTEN (phosphatase and tensin homolog) resulted in enhanced migration. These are the first-known genes to control electric-field-directed cell migration. These studies and others have shown the importance of biophysical signaling and have uncovered the mechanisms by which biophysical signals are translated into familiar signaling pathways.

One exciting application of biophysical signaling is in the control of stem cell behavior. Studies have shown that stem cells exhibit unique electrophysiological profiles in their undifferentiated state [21]–[25]. More interestingly, ionic currents and channels have been found to play important roles during myoblast, cardiomyocyte, and neural stem cell differentiation [21], [25]–[28]. However, the ability of these endogenous electrical signals to act as a functional biophysical control mechanism in stem cell biology is poorly understood. Moreover, it is not known whether stem cells' differentiation process is controlled by the electric fields, localized pH and ion gradients, or transmembrane potential changes resulting from the activity of ion channels and pumps. The aim of this study was to characterize membrane potential (Vmem) changes in human mesenchymal stem cells over the course of differentiation toward two different cell lineages, bone and fat, and to investigate a functional relationship between control of membrane potential and differentiation.

In order to determine whether membrane potential of hMSCs changes as a function of differentiation time, we tracked membrane potential changes during osteogenic (OS) and adipogenic (AD) differentiation with confocal microscopy using the voltage-sensitive fluorescent dye DiSBAC2(3). Since DiSBAC2(3) is an anionic bis-oxonol, it tends to partition into the cell membrane when the intracellular charge is more positive, resulting in a higher fluorescence signal when the cell is depolarized. The fluorescence profile during OS differentiation showed decreased membrane potential (hyperpolarization) compared to undifferentiated cells after 2, 3, and 4 weeks of differentiation, with the 4-week-differentiated samples being the most hyperpolarized (Fig. 1). During AD differentiation, all differentiated AD cells were hyperpolarized compared to undifferentiated cells, but the extent of hyperpolarization varied with differentiation time. AD cells exhibited slightly greater hyperpolarization with 1–2 weeks of differentiation vs. with 3–4 weeks of differentiation. Due to technical difficulties, we were not able to obtain direct intracellular recordings of the differentiating cells beyond the first week of culture. However, we estimated Vmem changes from DiSBAC2(3) fluorescence changes by equating a 1% change in fluorescence to a 1 mV change, as has been shown for the closely-related voltage-sensitive dye DiBAC4(3) [29]. From these voltage changes, Vmem values at weeks 1, 2, 3, and 4 can be calculated with respect to Vmem values at week 0, which were successfully recorded from cells during Days 1–4 of OS or AD differentiation (Fig. 2A). From these estimates, OS and AD cells exhibited a net Vmem hyperpolarization of 56 and 74 mV, respectively, at the end of four weeks of differentiation (Table 1).

From these data, we conclude that differentiated hMSCs are hyperpolarized compared to undifferentiated hMSCs, and that the polarization of the cells is associated with their differentiation toward different cell lineages (OS or AD).

After observing trends in membrane potential changes during hMSC differentiation (a hyperpolarized phenotype in differentiated cells vs. undifferentiated cells), we sought to determine whether this hyperpolarization was functionally required for differentiation. To accomplish this, we disrupted the normal progression of membrane potential changes by depolarizing hMSCs during differentiation. Two independent strategies were employed to depolarize membrane potential to ensure that the observed effects were in fact due to membrane depolarization, rather than due to inducer-specific effects. Thus, hMSCs were cultured and differentiated in the presence of high [K+]out or ouabain to achieve depolarization. Normally, the ratio between extracellular and intracellular [K+] is small (low [K+]out/[K+]in). However, as [K+]out is elevated, the ratio increases as [K+]out approaches [K+]in. As this ratio approaches 1, the magnitude of the Nernst potential for K+ decreases, effectively raising the membrane potential closer to 0 mV (depolarization). Increasing the extracellular potassium concentration is a standard means of depolarizing cells. Similarly, blockade of the the Na+/K+ ATPase pump by the specific inhibitor ouabain was used to inhibit the main source of transmembrane potential. Samples were collected after 2, 7, 14/15 and/or 22 days of differentiation for PCR and other tissue-specific analyses. Treated cells were compared to untreated differentiated cells (AD or OS) or to undifferentiated cells (control).

We first confirmed the activity of our depolarizing treatments using direct electrophysiology. Intracellular recordings were taken from MSCs cultured in OS or AD medium at early time points during differentiation (Days 1–4) (Fig. 2A). On average, OS cells were at a resting potential of −37 mV. Exposure to 10 nM ouabain depolarized the cells to −15 mV, while exposure to 80 mM K+ depolarized the cells to −7 mV. On average, AD cells were at a resting potential of −47 mV. Exposure to 10 nM ouabain depolarized the cells to −22 mV, while exposure to 80 mM K+ depolarized the cells to −9 mV. Confocal imaging of DiSBAC-loaded cells also reported depolarization (higher pixel intensity) for OS and AD cells exposed to ouabain and high K+ (Fig. 2B). We then tested the effects of these treatments on the differentiation trajectory of the hMSCs.

The effect of depolarization on AD differentiation was assessed by quantifying the expression of the AD-related genes peroxisome proliferator activated receptor γ (PPARG) and lipoprotein lipase (LPL). PPARG is a an adipogenic-specific transcription factor in the nuclear hormone receptor superfamily [30], [31]. Upregulation of PPARG expression is induced during mitotic arrest very early in adipose differentiation and stimulates the adipogenic pathway, activating the transcription of other adipose-specific genes [32]–[34]. LPL is one of several highly specialized proteins induced during adipocyte differentiation by PPARG, and plays an important role in lipoprotein and energy metabolism [35], [36]. LPL hydrolyzes triglyceride moieties of lipoproteins, generating free fatty acids necessary for triglyceride synthesis and lipid accumulation [37].

Potassium gluconate was added to AD culture medium to elevate the extracellular concentration of K+. On Days 2, 7, and 14, samples treated with 80 mM K+ (denoted AD-80K) had significantly lower LPL expression compared to untreated AD samples (decrease of 50.7-fold, 219.7-fold, and 43.5-fold, respectively, p<0.007, Fig. 3). Oil Red O staining on Day 7 also resulted in a different staining pattern in K+-treated AD cells: staining was diffuse and fairly uniform throughout the sample, revealing tiny lipid droplet formation mostly at the periphery of the cells (Fig. 4A,B), compared to the more extensive lipid droplet formation in AD cells (Fig. 4E,F). Oil Red O absorbance levels were also reduced 8.7-fold (p<0.002, Fig. 4I). On Day 22, LPL levels of depolarized samples continued to remain lower than untreated AD samples (83.9-fold, p≪0.001). These results demonstrate that exposure to 80 mM extracellular K+, which depolarizes hMSCs, can suppress AD differentiation for approximately three weeks.

To ensure that the observed effects were not due specifically to the high extracellular K+ levels but rather to the reduction of transmembrane potential, we next depolarized cells by a different mechanism. Since activity of the Na+/K+ ATPase pump is the main source of Vmem gradient in most cells, effects of the Na+/K+ ATPase inhibitor ouabain were assessed during AD differentiation. On Day 2, cells treated with 10 nM ouabain (denoted AD-ouab) had higher PPARG levels (1.9-fold, p<0.001) but similar LPL levels compared to untreated AD cells (Fig. 3). By Days 7, 14, and 22, AD-ouab cells had lower PPARG levels (1.5-, 2.5-, and 2.4-fold, respectively, p<0.05) and lower LPL levels (25.4-, 7.2-, and 6.2-fold, respectively, p<0.008) compared to untreated AD cells. On all days, however, LPL expression in AD-ouab cells, while lower than that of untreated AD cells, was greater than that of AD-80K cells (34.0-, 7.0-, 5.2-, and 13.5-fold greater for Days 2, 7, 14, and 22, respectively, p<0.0005, Fig. 3). Oil Red O absorbance on Day 7 also showed a 3.3-fold decrease in ouabain-treated AD cells compared to untreated AD cells (p<0.007, Fig. 4I). Unlike K+-treated cells, AD-ouab cells did not stain uniformly (Fig. 4C,D). Only a small population of cells stained for lipid droplets, but those that were positively stained showed more extensive droplet accumulation compared to K+-treated cells. The greater staining intensity of this subpopulation may explain why the Oil Red O quantification in AD-ouab cells was slightly higher than that of K+-treated cells, even though the K+-treated cells had a larger population of positively stained cells (Fig. 4I). Apart from this subpopulation, however, the majority of AD-ouab cells did not show any lipid droplet formation.

Therefore, treatment with 10 nM ouabain also suppresses AD differentiation, although not to the same degree as treatment with 80 mM K+. Some cells were partially able to overcome the effects of ouabain depolarization at later time points, as seen by their ability to increase LPL expression and form lipid droplets.

To better understand the temporal effects of depolarization on the progression of AD differentiation, K+ and ouabain treatments were administered early during culture (Days 1–2 or Days 1–4), then washed out and replaced with AD medium. Cells were cultured until Day 7, and gene expression was evaluated to determine whether the shorter exposure time was sufficient to change expression patterns (Fig. 5). Exposure to 10 nM ouabain for two days did not significantly change PPARG or LPL expression (Fig. 5A), but a four-day exposure resulted in a 2.5-fold decrease in PPARG and 3.7-fold decrease in LPL expression (Fig. 5B, p<0.02) compared to untreated AD. In contrast, exposure to 80 mM K+ for two days decreased gene expression levels (2.1- and 8.5-fold for PPARG and LPL, respectively, Fig. 5A, p<0.02) compared to untreated AD cells. Four-day exposure to K+ further decreased LPL levels dramatically by 113.0-fold (Fig. 5B, p<0.009). When the K+ and ouabain treatments were removed after four days, electrophysiology measurements showed that the Vmem of the cells recovered to similar levels as untreated AD cells by Day 5 (Fig. 6). The ability of the depolarization treatments to suppress AD-related gene expression after washout despite recovery of Vmem suggests that the instructive Vmem signal acts early during the differentiation process and is thus unaffected by the subsequent recovery of normal Vmem levels. It is important to note that the Vmem recordings of untreated AD cells in Fig. 6 cannot be compared to the Vmem values extrapolated from the voltage dye studies in Table 1 because of differences in culture conditions.

The intracellular recordings were acquired from hMSCs that had been cultured and differentiated for a total of five days, whereas the voltage dye studies were conducted after a total of four weeks of culture, with differentiation being initiated at various time points within those four weeks.

After characterizing the AD response to depolarization, we sought to determine whether the observed suppression of differentiation is restricted to AD lineages only, or whether control of Vmem is an important parameter for other differentiation trajectories as well. To address this question, we induced depolarization during OS differentiation using the same approach as with AD differentiation. Since both OS and AD differentiation normally exhibit hyperpolarization, our hypothesis was that just as in AD differentiation, the normal progression of OS differentiation would be inhibited by externally induced depolarization.

Extent of OS differentiation was determined by quantifying the gene expression levels of alkaline phosphatase (ALP) and bone sialoprotein (BSP), the activity of ALP, and the calcium content of the depolarized cells. ALP is a membrane-bound glycoprotein that plays a role in early osteogenesis and also initiates matrix mineralization [38]. BSP is a secreted glycoprotein mainly localized to the extracellular matrix of bone tissue [39]. It is associated temporally and spatially with calcification events during later stages of osteogenesis and acts as a nucleator for hydroxyapatite crystal formation, thus facilitating bone mineralization [40], [41]. The resulting tissue has a high concentration of calcium ions, as the inorganic mineral phase of bone consists mainly of calcium hydroxyapatite [42].

Treatment with both 40–80 mM K+ and 10 nM ouabain suppressed OS differentiation by Day 7. Ouabain treatment decreased ALP by 2.1-fold and BSP by 34.9-fold (p<0.002, Fig. 7A,B). Treatment with 40, 60, and 80 mM K+ decreased ALP by 1.4-, 1.3-, and 4.6-fold, respectively, and BSP by 2.7-, 10.3-, and 69.8-fold, respectively (p<0.002), compared to untreated OS samples (Fig. 7A,B). Alkaline phosphatase (ALP) activity on Day 15 (Fig. 8A) and calcium content on Day 21 (Fig. 8B) were also measured as indicators of bone matrix formation and mineralization, respectively. Ouabain-treated cells had 3.9-fold decreased ALP activity compared to untreated OS cells (Fig. 8A, p<0.001). Similarly, K+-treated OS cells exhibited decreased ALP activity in a dose-dependent manner (1.8, 2.1, 3.7, 7.1-fold decreased activity for increasing K+ concentrations, Fig. 8A, p<0.008). Calcium content on Day 21 was lower for all depolarized samples compared to untreated OS cells, with ouabain treatment yielding the least calcium content (12.8-fold decrease, Fig. 8B, p<0.0009) and K+ treatment yielding non-dose-dependent decreases in calcium content (6.1-, 7.1-, 6.1-, and 3.6-fold decreases with increasing K+ concentrations, Fig. 8B, p<0.002).

The duration of K+ exposure or ouabain exposure was shortened to determine whether depolarization-induced changes differed with treatment duration. Both treatments (80 mM K+ and 10 nM ouabain) decreased OS-related gene expression after an initial two-day exposure (p<0.04, Fig. 9A, left). In addition, as with AD differentiation, a four-day exposure to the treatments resulted in larger decreases in both ALP expression (3.8- and 9.2-fold for ouabain and K+, respectively, p<0.0001) and BSP expression by Day 7 (5.7-fold decrease for ouabain, p<0.0004, and undetectable levels for K+) (Fig. 9B, right). Although the suppression of OS-related gene expression persisted after washout of K+ or ouabain treatments, Vmem recovered to similar values as that of untreated OS cells (Fig. 6). Altogether, these data show that two days of early depolarization can inhibit differentiation as revealed by reduced expression of OS markers, and four days of early depolarization makes these changes even more apparent. These changes in phenotype persist despite recovery of normal Vmem after washout of early depolarization treatments.

The above depolarization studies show that depolarization prevents the normal differentiation of hMSC cells in two lineages. To rule out non-specific effects of depolarization and determine whether control of transmembrane potential is an instructive, bi-directional determinant of differentiation trajectory, we next asked whether hyperpolarization affected differentiation. If Vmem is a true determinant of differentiation potential, as has been suggested [43], hyperpolarization ought to have the reverse effect of depolarization and thus induce differentiation. OS-differentiating cells were exposed to KATP channel openers pinacidil and diazoxide (1, 10, 100 µM), reagents known to hyperpolarize various cell types [44]–[47]. Intracellular recordings confirmed that hMSCs exposed to 10 µM pinacidil or diazoxide within the first four days of OS differentiation indeed hyperpolarized by 9 mV with pinacidil treatment (p<0.04) and by 10 mV with diazoxide treatment (p<0.03, Fig. 10A). Gene expression of hyperpolarized OS cells was evaluated relative to untreated OS cells after 7 days. Exposure to 1 µM and 10 µM pinacidil resulted in a 1.8-fold and 2.2-fold increase in BSP expression, respectively (Fig. 10B, p<0.04), but no significant change in ALP expression. Exposure to 10 µM and 100 µM diazoxide caused upregulation of ALP expression by 2.9-fold and 2.1-fold, respectively, and upregulation of BSP expression by 4.3-fold and 3.1-fold, respectively (Fig. 10C, p<0.05). Thus, hyperpolarization of differentiating OS cells by pinacidil and diazoxide exposure elevates bone marker expression. These results suggest that depolarization is not merely permissive for differentiation, but rather that Vmem is a signal whose value functionally determines the differentiation propensity of hMSCs.

In this study, the characteristic membrane potential profiles for hMSCs during OS and AD differentiation were obtained using the voltage-sensitive fluorescent dye DiSBAC2(3). Fluorescence readings of differentiating cells decreased in intensity as a function of differentiation time, which corresponds to more negative membrane potential, or hyperpolarization. Hyperpolarization occurred during both OS and AD differentiation: based on the measured voltage-dependent fluorescence changes, we estimated that Vmem hyperpolarized by a net change of 56 and 74 mV after 4 weeks of OS and AD differentiation, respectively. These trends agree with several comparative analyses in the literature that examine Vmem and ion flux in various cell types and suggest a relationship between depolarization and control of differentiation and proliferation [1], [43], [48]–[50]. According to these reports, proliferative and relatively immature cells (such as those of the developing embryo) exhibit strongly depolarized membrane potentials, while terminally differentiated quiescent cells exhibit strongly hyperpolarized membrane potentials. Our observation that the process of differentiation was accompanied by a corresponding hyperpolarization in membrane potential is consistent with these data; as the stem cells are moving from their undifferentiated state to a committed phenotype, their membrane potentials shift from values that are characteristic of developing cells towards levels observed in terminally-differentiated, committed, somatic cells.

Vmem changes and other electrophysiological properties are traditionally studied in excitable cells such as neurons and myoblasts, rather than the non-excitable cells studied here. Interestingly, precursors of excitable cell types exhibit similar Vmem hyperpolarization during development and commitment compared to what we have seen in OS- and AD-differentiated hMSCs. For example, maturation of neuroblastoma cells derived from neural crest cells can be characterized by an ordered expression of currents: a human eag-related K+ channel current and a delayed rectifier K+ channel current are detected at early stages and are and replaced at later stages by a tetrodotoxin-sensitive Na+ channel current and an inward rectifier K+ channel current [21], [51], [52]. The changes in current cause the cells to undergo a net Vmem hyperpolarization during maturation. Similarly, the differentiation and fusion of myoblasts to form myotubes is accompanied by Vmem hyperpolarization from −10 to −70 mV [53], [54]. Kir2.1 channel activity, which is responsible for this hyperpolarization, initiates myogenesis by triggering the expression of early transcription factors such as myogenin and myocyte enhancer factor (MEF2) [27]. Therefore, hyperpolarization plays an important role in differentiation and maturation of both excitable and non-excitable cell types. In light of recent work exploring the differentiation potential of hMSCs beyond the OS, AD, and chondrogenic lineages, it would be interesting to compare Vmem changes in neural or myogenic hMSC differentiation to the hyperpolarization reported in neural precursors, myoblasts, and OS- and AD-differentiated hMSCs. Furthermore, if hyperpolarization does indeed occur during stem cell differentiation into neural precursors and myoblasts, Vmem modulation could be an attractive way to improve differentiation into these excitable cell types.

To determine whether hyperpolarization is functionally necessary for OS and AD differentiation, we disrupted the membrane potential of differentiating hMSCs and studied the subsequent changes in tissue-specific phenotypes. Membrane potential disruption (depolarization) was achieved by two classic methods: addition of high concentrations of extracellular K+ (high [K+]out), or addition of the Na+/K+ ATPase blocker ouabain. We used two independent methods to ensure that the effect was due specifically to reduction of transmembrane potential, and not side-effects of the individual treatments.

Addition of 80 mM [K+] to the differentiation medium resulted in marked suppression of the AD markers PPARG and LPL on Day 7 compared to untreated cells. This suppression of AD gene expression continued through Day 22. Lower concentrations of extracellular K+ did not induce the same changes in AD gene expression (data not shown). The different behavior of cells in low vs. high [K+]out in an AD-inducing environment suggests that there is an optimal [K+]out at which the resulting depolarization effects a change in AD phenotype, rather than a range of [K+]out over which a graded response can be observed. During OS differentiation however, Day 7 expression of the mature bone marker BSP varied in a graded, does-dependent manner, with 20 mM [K+]out having no significant effect and 80 mM [K+]out having the greatest effect. The differences in AD and OS cell response to different high [K+]out concentrations indicate that progression toward these two lineages changes the cells' sensitivities to different Vmem levels and their tolerance levels with respect to depolarization. Whereas AD cells have a higher tolerance for Vmem changes and show no response below 80 mM [K+]out, OS cells have a lower tolerance and respond proportionately to changes of 40 mM or more. Thus, not only is hMSC differentiation disrupted by depolarization, but more specifically, commitment to specific lineages (AD or OS) can be characterized by unique sensitivities to Vmem changes.

Addition of the Na+/K+ ATPase blocker ouabain (10 nM) also caused suppression of differentiation markers. Ouabain treatment significantly suppressed both AD and OS gene expression, although the fold-changes are smaller than to those seen with high [K+]out treatment. During OS differentiation, ALP activity and calcium content were also significantly reduced. In contrast to the effects of two treatments that depolarize cells by different mechanisms, bone-related gene expression was upregulated in response to hyperpolarizing K+ channel openers. Because hyperpolarization treatments caused an opposite effect to the depolarization treatments (depolarization suppresses differentiation, while hyperpolarization increases differentiation), our data support a view of membrane potential as an instructive determinant of differentiation state.

Washout experiments showed that early but short exposure to high [K+]out was sufficient to cause similar decreases in tissue-related gene expression as constant exposure. Exposure for the first two days or the first four days of culture (29% and 57% of culture time, respectively) resulted in decreased fat- and bone-related gene expression by Day 7. As with high [K+]out treatment, the effects of ouabain on gene expression can be seen with early but short exposure. An initial exposure of 2 days was sufficient to reduce OS gene expression, and an initial exposure of 4 days was sufficient to reduce AD gene expression. Interestingly, suppression of the differentiated phenotype was observed despite the recovery of Vmem to normal levels after washout of depolarization treatment. This suggests that the signal mediated by Vmem hyperpolarization is transduced to downstream controls of cell behavior during early stages of differentiation.

It is important to note that the differentiation markers assessed in the above studies are bone- and fat-specific, and do not simply reflect general cell viability and health. In response to Vmem depolarization, cells exhibited mostly decreased OS and AD marker expression compared to the normal trajectory of OS and AD differentiation. However, when compared to undifferentiated hMSCs, the depolarized cells consistently showed elevated OS or AD marker expression. The fact that the cells did initiate lineage-specific gene and metabolite production indicates that this cell response cannot be simply due to stress. Rather, the depolarized cells started the differentiation process but were restricted from reaching a fully differentiated phenotype, due to instructive signals provided by the changes in Vmem. Furthermore, gene expression of some markers was increased in depolarized cells even compared to untreated differentiated cells. For example, PPARG expression of ouabain-treated AD cells was almost 2-fold greater than that of untreated AD cells on Day 2 (Fig. 4), implying an increase in transcriptional activity not characteristic of stressed cells. Finally, the discovery that OS differentiation can not only be suppressed by depolarization, but can also be augmented with hyperpolarization, strongly suggests that the Vmem exerts specific control over differentiation events.

Taken together, these depolarization studies suggest that control of the endogenous Vmem is necessary for normal differentiation of hMSCs into bone and fat. Depolarization by addition of high [K+]out or ouabain results in suppression of the differentiated phenotype. Furthermore, depolarization appears to exert its effect early in the differentiation process. Importantly, this effect is not due to toxicity or non-specific stress response because specific marker gene expression was induced by depolarization and hyperpolarization. The ability to either increase or decrease the differentiation propensity by controlled modulation of Vmem suggests strongly that, as proposed previously [43], [48], [55], [56], transmembrane potential level is an instructive signal controlling important aspects of cellular plasticity.