Pulsed DC electric fields couple to natural NAD(P)H oscillations in HT-1080 fibrosarcoma cells

NAD(P)H is well known as a major cellular source of reducing potential, important for many catabolic and biosynthetic reactions. However, NAD(P)H may also serve as a substrate for covalent protein modification or as a precursor of biologically active compounds, including cyclic ADP-ribose (cADPR) and nicotinic acid adenine dinuclotide phosphate (NAADP⁺). cADPR and NAADP⁺ participate in cytosolic calcium release from intracellular stores, and the analysis of these functions has led to the emerging recognition of the importance of NAD(P) with respect to intracellular signal transduction (Cancela et al., 2000; Guse, 2000; Ziegler, 2000). We have recently reported that the cytoslic concentration of NAD(P)H in neutrophils and macrophages fluctuates in a periodic manner, and that not unlike Ca²⁺ oscillations, the amplitude and frequency of these oscillations appear to encode information, and are associated with signal transduction (Petty, 2000; Rosenspire et al., 2000; Kindzelskii et al., 1997). In particular, we have found that in spherical cells the NAD(P)H frequency is of the order of 3-4 minutes, but that upon polarization and adherence, a characteristic oscillation of about 20 seconds is superimposed upon the longer wave length oscillation. Both the amplitude and frequency of this shorter wavelength signal can be externally regulated by various cytokines, as well as the chemotactic signaling factor N-formyl-met-leu-phe (fMLP), and the tumor promoter and protein kinase C (PKC) activator, phorbol-12-myristate-13-acetate (PMA) (Adachi et al., 1999; Kindzelskii et al., 1997). Thus, it appears that in neutrophils and macrophages, cytokine and other intracellular signaling cascades that are initiated by plasma membrane receptors involve FM and/or AM modulation of NAD(P)H oscillations (Adachi et al., 1999; Petty, 2000; Rosenspire et al., 2000).

Surprisingly, we have also found that these (20 second) NAD(P)H oscillations are sensitive to electric fields, in that they will resonate with externally applied low power pulsed DC, (Kindzelskii and Petty, 2000) or AC fields (Rosenspire et al., 2000). Studies of the biological effects of extremely low frequency (ELF) electromagnetic fields is a field generally fraught with controversy and inconsistent results (Berg, 1999; McCann et al., 1998). Nevertheless, our findings on NAD(P)H resonance provide us with new insight, and potentially the key to understanding biological consequences of low power electromagnetic fields on cells. However, our previous studies have focused only on neutrophils and macrophages. We thought it important to demonstrate that ELF electric field resonance with NAD(P)H is a general phenomena, and one not restricted to cells of hematopoietic origin and, in particular, neutrophils and macrophages. Accordingly, this study focuses on the effect of pulsed DC fields on HT-1080 cells, a tumor cell line derived from a human fibrosarcoma of kidney origin (Rasheed et al., 1974). Although there are some differences, we have found that, as is the case for neutrophils and macrophages, NAD(P)H resonates with the electric fields, directly altering cellular physiology and function.

Reagents were obtained from Sigma Chemical Co. (St Louis, MO, USA), unless otherwise noted. HT-1080 tumor cells were obtained from the American Type Cell Culture Collection (Rockville, MD, USA). Cells were cultured in Eagles MEM with non-essential amino acids, and 10% fetal bovine serum in a humidified 10% CO₂ atmosphere.

Two Pt electrodes (0.25 mm in diameter) separated by 1.1 cm were attached to a glass microscope slide, and the space between the electrodes filled with the cell suspension. A coverslip was placed over the electrodes so as to exclude air. The entire assembly was then sealed with silicone grease, forming a chamber with external dimensions of 0.254×11.0×25.4 mm, with parallel electrodes running along the 25.4 mm dimension. The slide was placed coverslip down on an axiovert inverted epifluorescent microscope (Carl Zeiss, New York, NY, USA), equipped with a Zeiss temperature-controlled stage set to 37°C. An electrical stimulator (Grass Medical Instruments, Quincy, MA, USA) was used to apply DC pulses across the cell chamber.

Microscopy and fluorescence quantification was accomplished as previously described (Kindzelskii et al., 1996). Briefly, cells were examined using an automated axiovert inverted fluorescent microscope (Carl Zeiss) with mercury illumination (HBO100 burners), a quartz epifluorescent condenser, and quartz objectives. Cells were individually illuminated, and single cell NAD(P)H autofluoresence characterized using 365DF20 and 430DF20 excitation/emission filters, and a 405 long-pass dichroic mirror. Fluorescence levels were quantified using a Hamamatsu (Bridgewater, NJ, USA) cooled photomultiplier tube held in a Products for Research (Danvers, MA, USA) housing attached to the microscope and coupled to an amplifier (Photochemical Research Associates, London, Ontario, Canada). In some experiments, fluorescence was quantified using a D104 fluorescence microscope detection system interfaced with a computer running Felix software (Photon Technology International, Monmouth Junction, NJ, USA).

Cells were suspended in 2% gelatin incorporated with hydroethidine at 45°C, prior to being loaded into the cell chamber, and mounted on the microscope stage as described above. Reactive oxygen metabolites produced by the cells diffused into and were trapped within the gelatin matrix, where they oxidized the hydroethidine to ethidium. Ethidium was detected by epifluorescence microscopy using a narrow bandpass discriminating filter set (Omega Optical, Brattleboro, VT, USA) with excitation at 540/20 nm, and emission at 590/30 nm, in combination with a long pass dichroic mirror of 560 nm. Fluorescence intensity was quantified as described above.

The SCGE assay was accomplished as previously described (Kindzelskii and Petty, 2000). Briefly, cells were directly incorporated into low melting point agarose prior to being loaded into the cell chamber and mounted on the microscope stage, as described above. After the agarose solidified, cells were viewed under the microscope and, out of the several cells in the visual field, one polarized cell was selected. The frequency and phase of its NAD(P)H oscillation was determined by fluorescence microscopy as outlined above, and DC voltage pulses applied across the Pt electrodes. The pulses were phase matched, so that each pulse would coincide with a NAD(P)H minimum. Metabolic resonance was thus established with the applied field, and maintained for timed periods. The target cell as well as control cells not in resonance were then removed from the scope and processed using the SCGE assay for DNA damage as previously described (Kindzelskii and Petty, 1999).

Detection of 3′ OH termini of DNA was accomplished as previously described (Kindzelskii and Petty, 1999). Briefly, DNA strand breaks were detected using the Apo-Direct kit (PharMingen, San Diego, CA, USA), where terminal deoxynucleotidyl transferase is used to add fluorescein-dUTP to free 3′-OH sites of DNA strand breaks. After induction of metabolic resonance, cells were processed and labeled as suggested by the manufacturer. The same cells were then relocated and photographed using a fluorescein filter set.

Upon passage, when HT-1080 cells are initially plated at low density (50-100 cells/cm²), cells assume either a spherical (resting) or polarized (migratory) morphology (Fig. 1A,B). Initially ∼50% of the cells are migratory, and assume the polarized morphology. However the percentage of migratory cells decreases over time, until at confluence, while retaining a ‘spread’ morphology, almost all cells are non-migratory (Fig. 1C). We have previously shown that changes in NAD(P)H can be conveniently followed over time on the single cell level by monitoring NAD(P)H autofluorescence in an epifluorescence microscope interfaced with a photomultiplier-tube-based photon counting system (Kindzelskii et al., 1997; Kindzelskii and Petty, 2000). When NAD(P)H was assessed in spherical (resting) HT-1080 cells, the concentration varied periodically as a sinusoidal oscillation with a period of roughly 3 minutes (Fig. 2a). In addition to the 3 minute NAD(P)H oscillation, migratory cells exhibited a more rapid oscillation with a period of roughly 25 seconds that was superimposed upon the 3 minute oscillation (Fig. 2b). NAD(P)H oscillations ceased in both resting and migratory cells that had been killed by chemical fixation (Fig. 2c). Confluent cells exhibited only the 3 minute oscillation (data not shown).

We have previously shown that pulsed DC electric fields resonate with NAD(P)H oscillations in polarized neutrophils (Kindzelskii and Petty, 2000). Therefore, having established the presence of NAD(P)H metabolic oscillations in HT-1080 cells, we sought to test the effects of pulsed DC electric fields on migratory HT-1080 cells. In the absence of an applied field, NAD(P)H oscillations are observed (Fig. 3a), as described in Fig. 2. However, when a pulsed DC electric field (0.2 V/m), is applied at the troughs of the NAD(P)H oscillation (Fig. 3b, arrows), the oscillation amplitude increases. As shown in Table 1, under these conditions the amplitude of the NAD(P)H oscillation increases 216% after the fourth pulse. Thus, NAD(P)H resonates with the applied field. When the field application is discontinued, the amplitude ‘rings-out’ over time, as it returns to its initial level after about 3-4 oscillations (Fig. 3c).

In contrast to the preceding results, when the field is applied (180° out of phase) at the crests of NAD(P)H intensity, the amplitude of the NAD(P)H oscillation is depressed with each pulse, until ‘rapid’ NAD(P)H oscillations cease altogether (Fig. 3d). At this point, although cells retain a polarized (migratory) morphology, NAD(P)H oscillates in only the 3 minute period mode, characteristic of resting cells.

In Fig. 3 we established that DC fields with strengths of 0.2 V/m resonated with NAD(P)H in HT-1080 cells. To assess the effect of different field intensities on resonance, cells were pulsed with DC fields with varying strengths of 5.0-9.6×10⁻⁶ V/m. We find that fields of ≥10⁻⁵ V/m induce metabolic resonance, so that the field cut-off for resonance is at least 10⁻⁵ V/m. Furthermore, once a critical field intensity has been reached, and after resonance is established, the maximum amplitude of the NAD(P)H oscillation is independent of the intensity of the inducing field (Fig. 4).

We have previously demonstrated that in neutrophils, production of oxygen radicals is cyclically phased to NAD(P)H oscillations, and that under conditions whereby NAD(P)H is in resonance with externally applied electric fields, production of oxygen radicals increases (Kindzelskii et al., 1998; Kindzelskii and Petty, 2000; Rosenspire et al., 2000). However neutrophils are specialized cells whose antimicrobial function depends to some extent upon production of oxygen radicals. Therefore, we determined whether external (NAD(P)H phase matched) electric fields also increased production of oxygen radicals in HT-1080 cells, which are not normally associated with anti-microbial activity.

Hydroethidine is oxidized to ethidium by superoxide ions. Other workers have shown that this reaction is specific for superoxide (Carter et al., 1994; Rothe and Valet, 1990). Ethidium, due to the delocalized π electron clouds, is far more fluorescent at longer wavelengths than hydroethidine. Thus, hydroethidine can be used as a label to detect the deposition/trafficking of superoxide. Accordingly, HT-1080 cells were suspended in a hydroethidine impregnated gelatin matrix, and then observed by fluorescence microscopy and quantitative fluorometry in order to monitor production of oxygen radicals in the presence or absence of external electric fields. At field strengths of 0-10⁻⁵ V/m, below the threshold for field induction of metabolic resonance, no significant accumulation of ethidium was found, indicating no measurable superoxide formation. However, during metabolic resonance at field strengths above 10⁻⁵ V/m, a step-wise accumulation of superoxide anions was observed (Fig. 4).

Ethidium tends to label DNA, suggesting that electric field induction of radical formation might occur near sensitive sites and mediate genotoxic effects. We have already shown that in neutrophils increased production of oxygen radicals by resonant electric fields leads to DNA damage (Kindzelskii and Petty, 2000; Rosenspire et al., 2000). Therefore we tested the ability of phase-matched pulsed DC fields to mediate DNA damage in HT-1080 cells. Accordingly, cells were incorporated into low melting point agarose, and then exposed for periods of time to pulsed DC fields that were in resonance with NAD(P)H. DNA damage was then assessed by the use of a standard comet assay (Kindzelskii and Petty, 1999). As Fig. 5 and Table 2 indicate, cells exposed to electric fields for as little as 30 minutes showed comet tails, indicative of DNA damage. Untreated cells or cells exposed to phase-mismatched fields served as negative controls, and did not exhibit comet tail formation (Table 2).

We have recently reported upon a modification of the standard comet assay incorporating fluorescein-dUTP labeling and occultation microscopy, which offers sensitivities roughly an order of magnitude beyond the standard comet assay for the detection of damaged DNA (Kindzelskii and Petty, 1999). Using this improved assay, we have detected DNA damage after cells have been exposed to resonating electric fields for as little as 5 minutes (Fig. 6).

We have previously demonstrated that pericellular proteolysis in neutrophils is an oscillatory phenomena, and that DC electric fields that were phase matched to, and resonate with, cytosolic NAD(P)H oscillations, increased pericellular proteolysis (Kindzelskii et al., 1998; Kindzelskii and Petty, 2000). Aside from being an important feature of neutrophil function, pericellular proteolysis is an aspect of tumor cell behavior that is linked to metastatic potential. To assess the effect of electric fields on tumor cell pericellular proteolysis, HT-1080 cells were suspended in a gelatin matrix containing Bodipy-BSA. A single migratory HT-1080 cell was identified by differential interference contrast (DIC) microscopy, and then the microscope condenser adjusted so that only the selected cell would be illuminated. The phase and frequency of its NAD(P)H oscillation was then determined by epifluorescence microscopy, as outlined above. At this point, DC pulses (0.2 V/m) were delivered in phase with, and at the NAD(P)H minima, conditions that we have shown lead to NAD(P)H resonance with the applied field. We then measured Bodipy fluorescence in the immediate vicinity of the selected cell. Bodipy-BSA becomes fluorescent upon proteolysis of the BSA moiety, so that quantitative (Bodipy) fluorescent microscopy of cells embedded within a Bodipy-BSA doped matrix is a straightforward technique to measure proteolytic activity levels around single cells (Kindzelskii et al., 1998).

Figure 7A shows that in the absence of an applied electric field, the fluorescent Bodipy signal increases in a step-wise pattern, with the step frequency equal to the endogenous NAD(P)H frequency. Thus, migratory HT-1080 cells are characterized by endogenous pericellular proteolytic activity that is oscillatory in nature, and linked to oscillations of NAD(P)H. Proteolytic activity appears to be characterized by ‘quantal’ bursts. However, when pulsed DC electric fields are applied (Fig. 7B, arrows) under resonant conditions that increase the NAD(P)H amplitude, proteolytic activity is greatly enhanced. Accumulation of fluorescent Bodipy remains step-wise and retains the endogenous frequency, but each step height is significantly enhanced. It thus appears that under (NAD(P)H) resonant conditions, each quantal unit of proteolytic activity is increased, but the burst frequency remains unaffected.

In this report we have shown that HT-1080 cells exhibit natural oscillations of NAD(P)H. As is the case for neutrophils, we find a complex oscillatory behavior pattern. There is a predominant oscillation of approximately 3 minutes, with a superimposition of a higher frequency oscillation with a period of about 20 seconds. Again, as in neutrophils, the appearance of the shorter period oscillation is strictly correlated with cell migratory behavior, and it is the shorter period oscillation that seems to be associated with the control of cell function. We have previously found that disruption of normal 20 second NAD(P)H oscillations in pyoderma gangrenosumm patients is associated with abnormalities in neutrophil trafficking (Adachi et al., 1998), whereas disruption of the NAD(P)H oscillations in normal neutrophils by non-toxic concentrations of mercuric chloride is associated with inhibition of neutrophil phagocytotic activity (Worth et al., 2001). In migratory neutrophils (Kindzelskii et al., 1998), as well as in HT-1080 cells, pericellular proteolytic activity rises and falls with the same frequency as the shorter period NAD(P)H oscillation, and ceases altogether when the oscillation collapses.

Furthermore, the shorter period oscillation can couple to external electric fields. Just as in neutrophils (Kindzelskii and Petty, 2000), pulsed DC electric fields synchronized with the NAD(P)H minimum, establish a resonance with the NAD(P)H oscillation, in that the amplitude of the NAD(P)H oscillation is increased. This increased oscillation amplitude is functionally significant. First, it signals an increase in pericellular proteolytic activity. Second, in neutrophils we have found that increased NAD(P)H amplitudes also signal an increase in the production of reactive oxygen metabolites (ROMs), leading to DNA damage (Kindzelskii and Petty, 2000). This holds for HT-1080 cells as well. HT-1080 cells in metabolic resonance with externally applied DC electric fields also increase ROM production, and show signs of damaged DNA.

The fact that eukaryotic cells can respond to DC electric fields has been noted before. For instance, during wound healing it is well established that epithelial cells use galvanotaxic signals to coordinate cell movement (Farboud et al., 2000; Nishimura et al., 1996). Although the molecular basis of epithelial galvanotaxis is just beginning to be elucidated (Fang et al., 1999), even at this point it seems that the phenomena that we are describing in this paper do not appear to be simply related to galvanotaxis. First, galvanotaxic signals depend upon steady state DC fields. We have found that NAD(P)H oscillations are completely unaffected by steady state DC fields (results not shown). For DC fields to be effective in regulating NAD(P)H oscillations, they must be synchronously pulsed with the NAD(P)H minima. Secondly, the cut-off critical field strength for directed galvanotaxic movement of epithelial cells is around 5 V/m (Nishimura et al., 1996), and this compares favorably with galvanotaxic cut-offs for other cells such as avian neural crest cells (Gruler and Nuccitelli, 1991). The DC critical field cutoff that we have measured with respect to modulation of NAD(P)H is orders of magnitude smaller (∼10⁻⁵ V/m).

HT-1080 cells, unlike neutrophils (Kindzelskii and Petty, 2000), do not appear to abnormally elongate during resonance conditions. However, the existence of NAD(P)H oscillations, their linkage with the control of cell function, and their ability to resonate with external electric fields appears to be a general phenomena in polarized migratory eukaryotic cells. Metabolic resonance and associated downstream effects are not restricted to neutrophils and macrophages, or other cells of hematopoietic lineage. Furthermore, as the threshold for resonance is extremely low (∼10⁻⁵ V/m), environmental electric fields could conceivably influence cell physiology through this mechanism. For instance, properly phased electric fields could signal increased pericellular proteolytic activity, and subsequently increase metastatic activity in migratory tumor cells. The ability of resonating electric fields to increase DNA damage also suggests that under conditions of metabolic resonance, electric fields could be a primary etiologic agent responsible for the appearance of tumor phenotypes arising among normal cell populations.

In view of the controversies regarding biological effects of weak ELF electromagnetic fields, it has been suggested (Berg, 1999), that investigators reporting ELF electrical effects on cells should address three questions concerning their system.

(1) What is the physiological causality for specific ‘electrical windows’ and their positive or negative efficacy? (2) What are the biochemical targets for either magnetic or electric fields or both? (3) What is the influence of electrical and (or) thermal noise on field efficiency? In this study, it seems clear that the ‘electrical window’, or frequency (1 pulse per 20 seconds) necessary for an electrical interaction, is not arbitrary, but arises directly as a consequence of the pre-existing underlying metabolic oscillation of the same frequency. Physiologically, the applied electrical fields couple to the metabolic oscillation. However, at this point we must admit that we do not know the nature of the biochemical target for the applied fields, nor for that matter do we know the biochemical nature of the ‘pacemaker’ responsible for the underlying 20 second oscillation. It may be that the identification of the electrical target(s) will have to wait for the elucidation of the biochemistry of the pacemaker. With respect to Berg’s third criteria, we are currently investigating the effects of thermal and electrical noise on this system.

This research was supported in part by the Fetzer Institute and the J. P. McCarthy Foundation, and by grant CA74120 from the NIH.