Inhibition of Cellular Respiration by Doxorubicin

Doxorubicin executes apoptosis, a process known to produce leakage of cytochrome c and opening of the mitochondrial permeability transition pores. To define the loss of mitochondrial function by apoptosis, we monitored cellular respiration during continuous exposure to doxorubicin. A phosphorescence analyzer capable of stable measurements over at least 5 h was used to measure [O(2)]. In solutions containing glucose and cells, [O(2)] declined linearly with time, showing that the kinetics of oxygen consumption was zero order. Complete inhibition of oxygen consumption by cyanide indicated that oxidations occurred in the respiratory chain. A decline in the rate of respiration was evident in Jurkat and HL-60 cells exposed to doxorubicin. The decline was abrupt, occurring after about 2 h of incubation. The inhibition was concentration-dependent and was completely blocked by the pan-caspase inhibitor benzyloxycarbonyl-Val-Ala-Asp-fluoromethyl ketone. Respiration in resistant HL-60/MX2 cells, characterized by an altered topoisomerase II activity, was not inhibited by doxorubicin. A decline in cellular ATP was measured in Jurkat cells after 2-4 h of incubation with 20 microM doxorubicin, paralleling the decline in respiration rate. Thus, cells incubated with doxorubicin exhibit caspase-mediated inhibition of oxidative phosphorylation.


Introduction
Doxorubicin, an anthracycline antibiotic, is a widely used anticancer drug (1). This agent intercalates with DNA and produces DNA breaks by stimulating topoisomerase II-cleavable complex formation (2,3). Doxorubicin also targets the mitochondria, impairing cellular respiration (4)(5)(6)(7). In the cell, the quinone moiety of doxorubicin is reduced to semiquinone radicals, generating reactive oxygen species, which directly damage cell organelles (6). Oxidative damage produced by the drug is partially mediated by the doxorubicin-Fe(III) complex (8). The outcome of these events is cell death, primarily by apoptosis (9,10).
Apoptosis is executed by a series of cysteine proteases, termed caspases. Caspase activation leads to mitochondrial dysfunction (11). The mitochondrial perturbation includes opening of the permeability transition pores (PTP) 1 (12). The PTPs, formed at contact sites between the inner and the outer mitochondrial membranes, are composed primarily of cyclophilin D (inhibited by cyclosporine A), the adenine nucleotide translocator (inhibited by bongkrekic acid), and the voltage-dependent anion channel. The PTPs permit passage of protons (which collapses the mitochondrial membrane potential, ∆ψ, and leads to uncoupling of oxidative phosphorylation) and low molecular weight apoptogenic proteins (e.g., cytochrome c, which de-creases mitochondrial oxygen consumption). It has been suggested that the mitochondrial perturbations are transient (11).
Although many of the processes involved in drug-induced apoptosis have been identified, a full understanding of apoptosis requires knowledge of the temporal relationships between them. This necessitates direct measurement of the time course of individual processes, such as respiration. Here, we measure the effect of doxorubicin on cellular mitochondrial oxygen consumption during doxorubicin exposure. The rate of respiration is unchanged for about 2 h, after which it decreases abruptly. We also measure cellular ATP levels, changes in which parallel changes in respiration. The results show that cyanide-sensitive oxidative phosphorylation is inhibited in cells undergoing apoptosis in response to doxorubicin.
Solutions. A 2 mM solution of the Pd phosphor was prepared by dissolving the powder at 2.5 mg/mL in dH 2 O and was stored at -20°C. Aqueous solutions of ATP (0.4 mM) were prepared in 10 mM Tris-HEPES (pH 7.5) and stored at -70°C. The final concentration was determined by absorbance at 259 nm using a molar extinction coefficient of 15400 (14). A working solution of ATP (4 µM) was prepared fresh in a solution containing 0.1 M Tris-HEPES (pH 7.5), 5 mM MgCl 2 , and 0.1% fat-free bovine serum albumin. A lyophilized powder containing luciferin (0.2 mg; molecular weight, 280) and luciferase (22000 units) was freshly dissolved in 1.25 mL of PBS, protected from light, and placed on ice. The final concentration of luciferin (570 µM) was determined from its absorbance at 327 nm, using the molar extinction coefficient of 18000 (14). NaCN solution was prepared at 2.0 M and brought to pH ∼7.5 with 12 N HCl. The zVAD-fmk solution was made by dissolving 1.0 mg in 1.0 mL of DMSO (final concentration, ∼2.14 mM) and was stored at -20°C.
Cells. HL-60, the resistant clone HL-60/MX2, and human T-cell lymphoma (Jurkat) cells were maintained in suspension cultures as described (24). The resistance of HL-60/MX2 exhibits an altered topoisomerase II catalytic activity and reduced levels of topoisomerase II R and proteins (15). Cell count and viability were determined by light microscopy, using a hemacytometer under standard trypan blue staining conditions. Incubation with Drugs. Cells were suspended at 10 6 cells/mL in media (containing 6.0 mM Na 2 HPO 4 and 10 mM glucose; pH ∼7.4), 10% fetal bovine serum, 2.0 µM Pd phosphor, and 1% fatfree serum bovine albumin and placed at 37°C. The drugs (e.g., doxorubicin, zVAD-fmk, and NaCN) were then added. For each condition, 1.0 mL (final volume after all additions, typically <20 µL) of the cell suspension was placed in 1.0 mL glass vials (8 mm clear vials, Krackler Scientific, Albany, NY). The vials were sealed with a crimp top aluminum seal (using a Wheaton hand crimper; Fisher Scientific) and placed in the instrument for oxygen measurements at 37°C. Mixing was accomplished with the aid of parylenecoated stirring bars (1.67 mm × 2.01 mm × 4.8 mm; V&P Scientific, Inc., San Diego, CA).
Oxygen Consumption. Cellular respiration was measured at 37°C (16)(17)(18)(19). The substrate was glucose. The rate of respiration was determined as the negative slope of the curve of [O 2 ] vs t (zeroorder rate constant, k, in µM O 2 min -1 per 10 6 cells). [O 2 ] in the suspension was determined, as a function of time, using the phosphorescence of Pd(II) meso-tetra(4-sulfonatophenyl)tetrabenzoporphyrin. The phosphorescence decay of the probe was characterized by a single exponential, with the reciprocal of the phosphorescence decay time (τ) being linear in [O 2 ], according to Here, τ is the lifetime in the presence of oxygen; τ o is the lifetime in the absence of oxygen; and k q is the second-order oxygen-quenching rate constant. The drift of the Pd phosphor solution without cells was e0.18 µM O 2 min -1 . zVADfmk alone had no effect on the value of k. Samples were exposed to light flashes (10/s) from a pulsed lightemitting diode array with peak output at 625 nm (OTL630A-5-10-66-E, Opto Technology, Inc., Wheeling, IL). Emitted phosphorescent light was detected by a Hamamatsu photomultiplier tube (#928) after first passing it through a wide-band interference filter centered at 800 nm. The amplified phosphorescence decay was digitized at a rate of 1 MHz by a 20 MHz A/D converter (Computer Boards, Inc.). Two hundred fifty samples were collected from each decay curve, and the data from 10 consecutive decay curves were averaged for calculation of τ. The instrument was calibrated using ascorbate and ascorbate oxidase as described below (16).
Phosphorescence measurements on cell suspensions were always carried out simultaneously on 4-6 samples from the same cell culture in order to minimize errors due to variation in culture preparation. Meaningful phosphorescence measurements could not be made until about 30 min after the addition of doxorubicin. This time was required for processing samples, including filling the glass vials, eliminating air bubbles, cleaning and warming the vials to 37°C, placing them in the instrument, and starting the program. Measurements were then made sequentially on the different samples. The missing points for early times did not affect comparison of the rates of respiration among the different samples of cells from the same culture. Usually, each of the 4-6 samples represented a different condition. The measurements were done simultaneously on these multiple samples from the same culture preparation. Because of variability between culture preparations, it is not meaningful to compare k values obtained with different preparations. Only k values for different conditions using the same preparation are comparable. When duplicate measurements for the same condition and the same preparation were performed, the coefficient of variation in k was less than 10%.
In some measurements, increasing the doxorubicin concentration appeared to be associated with increased values of [O 2 ]. This is believed to be an artifact of the measurement method, associated with the red color of doxorubicin (since the excitation wavelength for the Pd phosphor is 625 nm). The red light scattered by doxorubicin affects the measured phosphorescence decay curve, which is fit to an exponential Ae -t/τ , leading to a lowered value of τ (see equation below) and a higher calculated [O 2 ]. Absorbance measurements in the presence of cells showed that the doxorubicin concentration did not change materially during experiments lasting up to 5 h. Thus, the artifact associated with the red color does not affect the time dependence of [O 2 ].
ATP Content. Acid extracts were prepared by adding 200 µL of 10% perchloric acid to pellets containing 10 6 cells. The mixture was sonicated on ice for 30 s, and the supernatant was collected by centrifugation and neutralized by adding 200 µL of 2.0 M KOH. The sample was incubated on ice for 15 min, and precipitated KClO 4 was removed by centrifugation. The ATP content in the resulting supernatant was determined immediately.
The luciferin-luciferase bioluminescence system was used to determine cellular ATP (14,20). Luminescence was measured at 37°C using a luminometer (Chrono-Log Corp.) connected to a Chrono-log AGGRO/LINK interface. The data were exported into Microsoft Excel and analyzed as described below. The reaction mixture contained, in a final volume of 0.4 mL, 0.1 M Tris-HEPES (pH 7.6), 5 mM MgCl 2 , 0.1% fat-free bovine serum albumin, and ATP (40-240 pmol) or cellular acid extracts (5-10 µL). The reaction was started by rapidly injecting 10 µL of luciferin/luciferase mixture (5 nmol of luciferin and 176 units of luciferase) from a 50 µL Hamilton syringe into 0.4 mL of rapidly stirred assay mixture.
Light emission was measured every half second for 600 s, and the resulting intensity vs t curve was fit to an exponential, Ae -kt . As shown below, the [ATP] can be obtained from the value of the exponential parameter k. However, [ATP] is proportional to k only for concentrations below ∼0.15 µM; thus, it was necessary to dilute some cellular extracts before the measurements. Furthermore, variations in the luciferin/luciferase used required that each batch be calibrated against standard solutions before use in ATP measurements.

Results
Phosphorescence Calibration. Figure 1A is an example of a titration of dissolved O 2 with ascorbic acid in the presence of ascorbate oxidase. Two milliliters of Pd phosphor solution (containing media, 10% fetal bovine serum, 2.0 µM Pd phosphor, 1% fat-free albumin, and 1.25 units of ascorbate oxidase, final pH ∼7.3) was titrated at 37°C by addition of 10 µL aliquots of 10 mM ascorbate. [O 2 ] was measured electro-chemically, as percent of saturation S. This was converted to molar concentration C according to where 0.21 × (1 atm) is the partial pressure of oxygen in the atmosphere, 55.5 M is the molarity of water, and K H is the Henry's Law constant for dissolved oxygen in water at 37°C, 5.211 × 10 9 Pa. The calculated values of C are plotted in Figure  1A.
[O 2 ] decreased linearly (r 2 > 0.991) with ascorbate added. The slope of the line (-0.668) may be compared with the theoretical stoichiometry (mol of O 2 consumed per mol of ascorbate added) of 0.5. The discrepancy is due to the fact that the total molarity of the solution and the Henry's Law constant used in the equation above are appropriate for pure water, but the values appropriate for our solutions are not known. Multiplication of the above equation by (0.5/0.668) converted percent saturation S, as measured electrochemically, to [O 2 ] according to C ) S(1.696 µM).
Phosphorescence lifetimes (τ) were measured in a series of ascorbate/ascorbate oxidase solutions (16) Effect of Doxorubicin on Cellular Respiration. Jurkat cell respiration in the presence of 5 and 10 µM doxorubicin is shown in Figure 2A. This inhibitory effect of doxorubicin on cellular respiration was completely blocked by the presence of 20 µM of the caspase inhibitor zVAD-fmk ( Figure 2B). Three repetitions of these experiments consistently showed very similar results. For the untreated cells in Figure 2B, k ) 2.55. For the treated cells, k ) 1.49 for t > 120 min (an approximate 42% decrease from untreated cells). With 20 µM zVAD-fmk, the plot of [O 2 ] vs t was linear for all times, with k ) 2.36, essentially the same as for untreated cells. Thus, the data show that caspase activities mediate the effect of doxorubicin on mitochondria and that about 2 h is required to execute mitochondrial dysfunction.
The inhibitory effect of NaCN is shown in Figure 2C.  Figure 3A shows the analysis of oxygen measurements on 10 6 cells/mL, either untreated or treated with 20 or 40 µM doxorubicin (as in Figure 2). Here, six-point segments (total time 35 min, since measurements were made every 7 min) were fitted to lines; the magnitude of the slope (k) with statistical error is shown in Figure 3B-D, plotted against the start time of the segment. Figure 3B is for no drug, Figure 3C is for 20 µM doxorubicin, and Figure 3D is for 40 µM doxorubicin. In the first case, k was constant, but in the presence of 20 or 40 µM drug, there was a sudden drop in k at about 150 min.
Another way to demonstrate the sudden drop in respiration and also to compare k values for treated and untreated cells for long treatment times involved the following experiment. Two 10 mL cultures of Jurkat cells were prepared; each contained 10 6 cells/mL, one with addition of 20 µM doxorubicin and one without the addition. Both samples were incubated at 37°C, open to the air. Every hour, starting at t ) 0, a 1 mL sample of each was taken and placed in sealed vials for phosphorescence measurement of [O 2 ] in the usual way, alternating between the two. One measurement was made per minute. Because more than 10 min was required to load and equilibrate samples, measurements could be made only ∼15 min after taking a sample. The results are shown in Figure 4A (open circles for no drug and filled squares for 20 µM drug). The set of [O 2 ] values for each sample was fitted to a line to obtain k. The k values with statistical errors are plotted in Figure 4B. For untreated cells, k was essentially constant, although it increased The fact that inhibition of respiration by doxorubicin was not observed in HL-60/MX2 cells, which lack topoisomerase II activity, indicated that topoisomerase II was required for doxorubicin-induced execution of apoptosis and promotion of mitochondrial dysfunction. It also appears that doxorubicin enhanced HL-60/MX2 respiration ∼1.8-fold. This drug effect was fully inhibited by cyanide or rotenone (not shown). Thus, the enhancement was not mediated by redox cycling of doxorubicin by mitochondria (6).
Luminescence Calibration. Cellular ATP was determined from the luminescence vs time curve in the presence of luciferin and luciferase. The observed luminescence as a function of time should follow Re -γt + , with R proportional to the initial [ATP]. Attempts to fit our experimental data to this threeparameter form were unsuccessful because the scatter in measured intensities was a large fraction of the variation in light intensity over 600 s, making the determination of three parameters unreliable. (Typical intensity plots are shown in the top panel of Figure 6.) Instead, we fit our data to the two-parameter exponential form Ae -kt , as shown in the top panel of Figure 6, for which the exponential fits are 41.27 e -0.00035t and 35.43 e -0.00042t . It is obvious that, if R is much smaller than (very small [ATP]), k will approach 0, whereas if R is much larger than (very large [ATP]), k will approach γ, the rate constant for the ATP-luciferin reaction. For small [ATP], the value of k must be proportional to R/ , i.e., to the initial [ATP]; but for larger [ATP], the value of k will approach a limiting value.
This behavior is shown in the bottom panel of Figure 6, which gives the results (k values) of fitting exponentials to luminescence vs time data sets obtained for a series of solutions of  , it was required to dilute the sample before measurement to make [ATP] less than 0.2 µM. It should be noted that the calibration against standard ATP solutions has to be performed for each luciferin solution, because of variations from one preparation to another, and the experimental measurements should be done soon after calibration, because the solution degrades in a few days. Cellular ATP. Jurkat cells were suspended at 10 6 cells per mL media with 10% fetal bovine serum and 1% albumin and incubated at 37°C in sealed containers for up to 3 h. The results of measurement of ATP are shown in Figure 7, as 100 times the ratio of measured ATP at time t to measured ATP at time 0. Three conditions were used as follows: no addition of drug (circles), with addition of 20 µM doxorubicin (squares), and with addition of 20 mM NaCN (triangles). At 1 h intervals after addition of doxorubicin or cyanide, three 1 mL samples of each cell suspension were taken for analysis of ATP. With no addition, cellular ATP remained constant. The addition of doxorubicin had no effect on cellular ATP at 1 h, but at 3 h, ATP decreased to 36% of its initial value. The addition of cyanide produced a sharp decrease in cellular ATP at 1 h and a further decrease to 20% of its initial value at 3 h. For all three conditions, there was a small increase in cellular ATP between 2 and 3 h; further investigation will be necessary to determine whether this increase is significant.
A second series of experiments was designed to show the effect of zVAD. Jurkat cells were suspended at 10 6 cells per mL media with 10% fetal bovine serum and 1% albumin and incubated at 37°C in containers open to the air for up to 5 h (Figure 8). Three conditions were used as follows: untreated (circles), addition of 20 µM doxorubicin (dark squares) and addition of 20 µM doxorubicin plus 20 µM zVAD (light squares). At time 0 and at five succeeding 1 h time intervals, three samples were collected from each of the three conditions and analyzed for ATP as described above. Each result shown in Figure 8     added in addition to 20 µM doxorubicin, cellular ATP did not decrease with time. Indeed, cellular ATP increased slightly: The slope was 0.59 ( 0.24 nmol/10 6 cells/h, equal (within statistical error) to the slope for untreated cells and to the slope for doxorubicin-treated cells for t e 3 h. It is clear that zVAD nullifies the doxorubicin-induced attenuation of cellular ATP as well as the doxorubicin-induced inhibition of respiration.

Discussion
Anthracyclines are known to target the mitochondria (4-7). However, incubations of beef heart submitochondrial particles with doxorubicin produced no noticeable effect on oxygen consumption (17). Thus, the effect of this drug on cellular respiration is likely indirect, perhaps mediated by induction of apoptosis. We utilized zVAD-fmk (a pan-caspase family inhibitor) (21) in order to investigate whether caspases mediate doxorubicin's effect on cellular respiration ( Figures 2B and 5). The results show that doxorubicin induces apoptosis (activation of caspases), which impairs respiration. Thus, the decrease in oxygen consumption is a consequence of apoptosis. The fact that the combination of cyanide and doxorubicin leads to greater oxygen consumption than with cyanide alone ( Figure 2C) suggests that doxorubicin may slightly stimulate oxygen consumption, as shown previously (4,5).
It has been reported that doxorubicin increased the permeability of the inner mitochondrial membrane by opening the mitochondrial PTP. In adenocarcinoma cells, doxorubicin (17 µM for 60 min) decreased the value of ∆ψ by about 30%, an effect that was blocked by zVAD-fmk (7). Such an effect on ∆ψ would be expected to uncouple phosphorylation from oxidation. In addition, doxorubicin increases the permeability of the outer mitochondrial membrane to cytochrome c, an effect that is expected to decrease mitochondrial respiration.
The inhibitory effect of doxorubicin on respiration, observed in our experiments on Jurkat and HL-60 cells, was evident within 2-3 h of incubation with the drug (Figures 2A,B and  5A). This inhibitory effect was blocked by zVAD-fmk ( Figures  2B and 5A) and was not observed in HL-60/MX2 cells ( Figure  5B), which lacked topoisomerase II activity. Thus, it is clear from these observations that doxorubicin-induced inhibition of respiration requires caspase and topoisomerase II activities.
The rate of respiration in cells exposed to the drug remains close to the rate in untreated cells for about 2.5 h and then decreases, remaining relatively constant thereafter. (For cells exposed to the drug in the presence of air, the decrease in k occurs sooner, at about 1.5 h, as seen in Figure 4. It is known that drugs such as doxorubicin are more effective in inducing apoptosis when oxygen is present, possibly because reactive oxygen species play a role.) As will be shown in a future publication, other apoptosis-inducing drugs (e.g., actinomycin D) behave differently: They decrease respiration in Jurkat cells gradually, starting at the earliest times for which measurements can be made. The data in Figure 7 show that ATP levels do not decrease for at least 1 h of the treatment, and a significant decrease has occurred by 2 h, slightly earlier than for respiration.
Observations relevant to the role of mitochondrial ∆ψ were presented in a series of papers by Green and co-workers (22)(23)(24) who examined the apoptotic response of HL-60 and other cells following short exposures to actinomycin D, etoposide, or staurosporine. It was found that apoptosis was initiated without significant changes in ∆ψ and the changes in ∆ψ occurred only later in the cell death pathway. The authors concluded that collapse of ∆ψ, resulting from formation of the PTP, was not a critical feature of apoptosis (22). Studies of single cells treated with etoposide or actinomycin D showed that, if caspases were not activated, the mitochondria maintained critical functions, such as the generation of ATP, even after release of cytochrome c (23). In contrast, caspase activation disrupts complexes I-II of the mitochondrial electron transport chain, resulting in diminished ∆ψ and generation of reactive oxygen species (24). The results in Figures 2 and 3 and Figures 7 and 8 agree with these reports and show that caspase activation impairs oxidative phosphorylation. The decrease in cellular ATP (Figures 7 and 8) also emphasizes the importance of mitochondrial death during apoptosis (11).
Mizutani et al. (25) found that incubation of HL-60 cells with doxorubicin for up to 8 h led to H 2 O 2 -mediated oxidative damage of DNA, which in turn led to an indirect generation of H 2 O 2 via activation of NAD(P)H oxidase. The end result was an increase in ∆ψ and activation of caspase-3. Doroshow (4) reported that high concentrations of doxorubicin (135 µM) stimulated oxygen consumption when added to isolated cardiac mitochondria. The stimulation was more readily demonstrable in the presence of KCN and rotenone. The overall increase in oxygen consumption was attributed to the production of superoxide anion (5). The enhancement of O 2 consumption was immediate, required NADH, and was more pronounced in the presence of rotenone or cyanide. It was concluded that mitochondrial NADH dehydrogenase reduced doxorubicin to its semiquinone, with subsequent transfer of electrons to O 2 (5-7).
The observation that oxygen consumption in the present experiments was completely inhibited by NaCN ( Figures 2C  and 5A) strongly supports the conclusion that the observed O 2 consumption occurred in the mitochondrial respiratory chain. If reactive oxygen species were produced, their amounts did not significantly add to the observed respiration. It is also unlikely that small amounts of reactive oxygen intermediates were responsible for the observed inhibition of respiration and lowering of ATP levels since the effects of doxorubicin were prevented when zVAD-fmk was included in incubation mixtures. Thus, it is clear that, in our experiments, doxorubicin treatment executed apoptosis within about 2 h of drug exposure, decreasing mitochondrial oxidative phosphorylation ( Figures  2-4 and 8). Furthermore, the decline in cellular ATP contributed to the mechanism of cell death by apoptosis (Figures 7 and 8). It is worth noting that the doxorubicin-induced cellular ATP depletion started earlier and was more pronounced in the sealed vials (Figure 7) than the open containers ( Figure 8).
In summary, the results presented show that cells exposed to doxorubicin exhibit partial, dose-dependent inhibition of oxidative phosphorylation. The inhibition occurs after a few hours of exposure to the drug, is mediated by caspases, and requires topoisomerase II activity.