Metabolic Signatures Associated with a NAD Synthesis Inhibitor–Induced Tumor Apoptosis Identified by ¹H-Decoupled-³¹P Magnetic Resonance Spectroscopy

Apoptosis, or programmed cell death, plays a critical role in tissue development and homeostasis throughout normal life (1, 2). In cancer, apoptosis is involved in both tumorigenesis (3, 4) as well as the orchestration of death signals after conventional radiation, chemotherapy, and experimental antiangiogenic therapy, immunotherapy, and gene therapy (5–10). Radiation and many of the conventional chemotherapeutic agents kill tumor cells via the apoptotic pathway by damaging cellular DNA (5, 6), which however often leads to genomic instability and the generation of tumor resistance to subsequent therapies (3, 11–14). More recently, a novel antitumor agent, FK866, has been reported to induce apoptosis of tumor cells without any DNA damaging effects in human leukemia, hepatocarcinoma cells in vitro and various types of tumor xenografts in vivo by reducing the steady state intracellular NAD⁺ levels (15, 16). The depletion of NAD⁺ by FK866 is mediated through highly specific noncompetitive inhibition of nicotinamide phosphoribosyltransferase, a key mitochondrial enzyme involved in the regulation of NAD⁺ biosynthesis from the natural precursor nicotinamide (16). Recently, this antitumor agent has been introduced in clinical trials (17, 18).

Noninvasive in vivo identification of metabolic markers associated with tumor apoptosis could be useful for monitoring treatment response in patients and may allow early prediction of successful therapy. Magnetic resonance spectroscopy (MRS), a noninvasive and nondestructive technique, can provide information about the changes in cellular metabolism, pH, and energetics in vivo during the course of apoptotic cell death. Changes in lipid signals, such as the ratio of methylene/methyl peak areas (CH₂/CH₃) detected by proton MRS (¹H MRS), has been suggested to be an indicator of the altered lipid structure and fluidity that takes place during apoptosis (19). However, the mechanisms underlying these signal changes are still not completely understood, and in theory, other endogenous, nonmalignant cell types, and necrosis could also contribute to these observed changes in lipid signals (19). On the other hand, previous studies using phosphorus MRS (³¹P MRS) have reported alterations in the levels of cellular phosphomonoesters (composed primarily of phosphoethanolamine and phosphocholine), cytidine diphosphocholine, glycolytic intermediates, and nucleoside triphosphates (NTP) in the event of tumor apoptosis (20–25). Besides being diverse in their findings, a majority of these studies have been carried out in vitro with cultured tumor cells, where the number of cells synchronously undergoing apoptosis will be higher. However, the ³¹P MR spectra obtained from tumors in vivo would contain signal contribution from only a small portion of apoptotic cells due to the heterogeneous response of tumor cells to treatment and rapid phagocytosis of cells that have undergone apoptosis. In addition, the achievable in vivo spectral resolution is lower. Therefore, in most cases, it is difficult to identify reliable metabolic markers of tumor apoptosis in vivo by ³¹P MRS.

The present study was aimed to investigate the metabolic effects of the NAD synthesis inhibitor, FK866, in mouse mammary carcinoma in vivo using ¹H-decoupled ³¹P MRS. To discern further the changes in metabolic profiles of tumors observed in vivo, high-resolution in vitro ¹H-decoupled ³¹P MRS studies were carried out with perchloric acid (PCA) extracts of mammary carcinoma tumors excised after similar treatments. In addition, the effects of FK866 on mammary carcinoma tumor growth and radiation sensitivity were studied. It was found that FK866 induced delay in tumor growth and enhancement in tumor radiosensitivity accompanied by an increase in ³¹P MR signals in the phosphomonoester region and a decrease in NAD⁺ levels, pH, and energetic status of mammary carcinoma in vivo in a dose-dependent manner. The histologic and cytofluorometric observations of tumor apoptosis correlate with in vitro ³¹P MRS detected changes in the pathways of tumor glycolysis, guanylate synthesis, pyridine nucleotide pools, and phospholipid metabolism.

Tumor growth and measurements. The mouse mammary carcinoma was prepared by aseptic removal of the tumor from tumor-bearing animals and processed as previously reported (26, 27). Briefly, a single cell suspension was prepared from a solid tumor by teasing and abrasion against a stainless steel mesh immersed in iced MEM supplemented with Earle's Balanced Salt solution containing 1% heparin. Cell suspensions were further separated by aspiration through an 18-gauge needle and the final suspension was continually agitated with a magnetic stirrer before inoculation for cellular injection uniformity. A tumor inoculum of 10⁵ tumor cells in 30 μL was injected s.c. into the dorsum of the foot of male C3H/He mice (Jackson Laboratories, Bar Harbor, ME). This site for inoculation of the tumor was chosen to minimize spectral contamination from adjacent (muscle/fat) tissues. Tumor volumes were estimated from the formula V = π [d₁d₂d₃ / 6], where d₁, d₂, and d₃ are three orthogonal diameters.

Drug treatment. The drug was diluted to working concentrations with normal saline and 60% propylene glycol. Two different schedules with different dosages of the drug were investigated. Schedule I involved eight total injections (i.p) at 25 mg/kg per injection with 12-hour interval between injections. Schedule II consisted of a total of four injections (i.p) 40 mg/kg per injection with 12-hour interval between injections. The higher dose schedule II was well tolerated by the mice (2 of 39 mortalities). The two animal deaths observed may have been a result of the trauma associated with multiple i.p injections as opposed to the true drug toxicity. The experimental procedures used for this study was reviewed and approved by the Institutional Animal Care and Use Committee.

Tumor irradiation. Briefly, animals were anesthetized with an i.p. injection of ketamine (Fort Dodge Animal Health, Fort Dodge, IA; 100 mg/kg) and xylazine (Lloyd Laboratories, Shenandoah, IA; 20 mg/kg). Single dose radiation (17 Gy) was localized to the tumor region, as previously described (31). Irradiation was done using a Philips MG 324 X-irradiation unit (Mahwah, NJ) operating at 320 kV and 10 mA with 0.5-mm copper filtration. The delivered dose rate at 50-cm source-to-surface distance with a field size of 10 × 10 cm was 150 cGy/min.

In vivo ³¹P magnetic resonance spectroscopy.In vivo³¹P spectra from the mammary carcinoma were acquired using a Bruker-CSI 4.7T spectrometer with a 33-cm horizontal bore magnet operating at 81.03 MHz for ³¹P. The acquisition variables were as follows: spectral width of 10 kHz, 45° pulse tip angle, relaxation delay of 1.8 seconds, 1,024 signal averages with 2,000 data points. Two homemade four-turn radiofrequency solenoid coils, one with 10-mm and the other with 15-mm inner diameter, were used to collect the data. Spectra of small tumors (<350 mm³) were acquired using the small coil and spectra of larger tumors were acquired using the larger coil. The 45-degree pulse width for the small coil was 4.1 microseconds and for the larger one was 7.0 microseconds. Other spectral variables were the same as listed above. The mouse was restrained in the animal holder (60-mL syringe barrel with air holes) and the leg carrying mammary carcinoma tumor positioned inside the coil was immersed in a water bath at 37°C, to minimize the magnetic susceptibility effects. Magnetic field shimming over the whole tumor volume was done before the ³¹P spectral acquisition. ¹H line widths of the tumors were 12 to 20 Hz. For schedule I, tumor ³¹P NMR spectra were acquired before the first injection (baseline) and 2 hours after the sixth and eighth injections of FK866. With schedule II, the spectra were acquired before the first and 3 hours after the fourth injection of FK866.

In vitro ³¹P magnetic resonance spectroscopy of tumor extracts. PCA extracts of excised tumors were prepared as described elsewhere (27). Two or three tumors were in general combined to make a single extract to obtain adequate signal-to-noise within reasonable acquisition times. In brief, tissue was ground to a powder in a liquid nitrogen–cooled mortar. Approximately an equal volume of ice-cold 10% PCA was added to the powdered tissue. This mixture was ground further until frozen tissue and PCA were well mixed. The mixture was then transferred to a polypropylene centrifuge tube, thawed, and the precipitate was removed by centrifugation for 5 minutes. The supernatant was removed and neutralized with 1 N potassium carbonate. Following adjustment to pH 10.0, the potassium perchlorate precipitate was removed by centrifugation. The supernatant was then mixed, and regularly shaken over a period of an hour, with Chelex-100 (Sigma, St. Louis, MO) to remove any paramagnetic ions. The Chelex-100 was subsequently removed by centrifugation. The supernatant was frozen and lyophilized overnight. The lyophilized extract was reconstituted in 1 mL deuterated water and the final pH was adjusted to 7.3 before MRS analysis.

High-resolution ³¹P MRS of tumor extracts were acquired on a Bruker AMX 400 MHz operating at 162 MHz for ³¹P using a 5-mm probe. A spectral width of 30 kHz, 32,000 data points, 45-degree flip angle, recycle time of 5.1 seconds, and an average of 1,500 scans were used. Waltz-16 inverse-gated scheme was employed for ¹H decoupling. The data acquisitions were done at 25°C. Spectra were referenced to glycerophosphocholine at 0.0 ppm. Dimethyl methyl phosphonate (1 mmol/L) was used as an internal concentration standard.

Magnetic resonance spectroscopy data processing and analysis. Signal-averaged free induction decays from in vivo ³¹P MRS were processed by applying 5 Hz of exponential line broadening before Fourier transformation. Time domain spectral fitting using advanced method for accurate robust and efficient spectral fitting technique (28) and peak area calculations were carried out using MRUI analysis package running under MATLAB (The Math Works, Inc., Natick, MA). Tumor pH in vivo was calculated from the chemical shift of inorganic phosphate (Pi) resonance relative to phosphocreatine. Free induction decays from in vitro acquisitions were Fourier transformed with 1 Hz exponential line broadening and peak area integrations were carried out using Bruker's XWIN-NMR software.

Histologic confirmation of apoptosis. Histochemical staining of cells undergoing apoptosis/necrosis was done by both H&E and annexin. Briefly, specimens of tumor from different treatment cohorts (mammary carcinoma control, or FK866 treated) that were harvested from animals at 24, 48, and 72 hours after the last injection, were either fixed in 4% buffered formalin and embedded in paraffin blocks for 5-μm sections and H&E staining, or the tumor was removed, minced into small fragments before being dissociated by physical scraping over stainless steel microsieve and repipetted over 75-μm mesh for collection in iced PBS. All glass slide mounted tissues were evaluated by light microscopy at 100× to 400× objectives using a Nikon Diaphot (Tokyo, Japan) Inverted Phase Contrast Microscope equipped with a Spot Imaging system and fluorescence (Diagnostic Instruments, Inc., Sterling, Heights, MI). Representative areas from the tumor were selected for qualification of viability, presence of mitotic figures, pyknosis/necrosis, and programmed cell death.

Cytofluorometric analysis of membrane potential, cell cycle distributions, and apoptosis. Tumors were surgically excised as previously mentioned and carefully disaggregated into a single cell suspension by teasing and physical abrasion of small tumor fragments over stainless steel wire mesh sieve immersed in an iced cocktail containing 5% trypsin/collagenase in PBS. Those cells were further agitated by repipetting over 75-μm mesh before 2× washing and centrifugation at 800 rpm for 10 minutes. Cells free of aggregate clumping were evaluated by hemocytometer or by fluorescence-activated cell sorting and used exclusively for analysis of either JC-1 staining (for evaluation of mitochondrial membrane potential), Annexin V staining (for apoptotic assay), or propidium iodide (PI) staining (for cell cycle distribution). JC-1 (Molecular Probes, Eugene OR) was dissolved in N,N-dimethyl formamide (Sigma Chemical, St. Louis, MO) and stored at 4°C in the dark before use. Samples were vortexed before 20 minutes of JC-1 incubation (1-2 μg/mL), washed in PBS, and centrifuged 800 rpm for 7 minutes before resuspension and analysis (525-590 nm). Annexin V-FITC and PI was done according to the standard staining protocol (BD PharMingen, San Jose, CA) within 1 hour after vortexing the cells. Briefly, FITC-Annexin V (1 mg/mL in binding buffer) was incubated 10 minutes in the dark at 25°C with subsequent PI to a final concentration of 5 mg/mL. All samples were analyzed by FACScan (Becton Dickinson, Franklin Lakes, NJ) flow cytometry at no less than 10⁴ cells using the Lysis II software package.

Statistical analysis. Test of significance was done using two-tailed, paired t test. P < 0.05 was considered statistically significant.

FK866 inhibits tumor growth and increases the radiation sensitivity of mammary carcinoma tumors. The two different dose schedules of FK866 treatment used in the present study: (I) 25 mg/kg × 8 and (II) 40 mg/kg × 4, with an interval of 12 hours between injections (i.p.), induced a significant delay in mammary carcinoma tumor growth. The untreated tumors (n = 10) had a tumor volume doubling time (absolute time to reach 2× the initial treatment volume, a mean of individual responses) of 2.4 ± 0.2 days (mean ± SE), whereas the 25 mg/kg × 8 (n = 7) and 40 mg/kg × 4 (n = 17) treatment groups had doubling times of 4.3 ± 0.3 days (P < 0.002) and 5.2 ± 0.5 days (P < 0.0002), respectively (Table 1). FK866-induced enhancement in mammary carcinoma tumor radiation sensitivity is shown in Fig. 1. For these studies, we have chosen dose schedule II (higher dose) as it showed a more pronounced tumor growth delay. Tumor volume doubling times for animals treated with 40 mg/kg × 4 doses of FK866 followed (3 hours after the fourth dose) by 17 Gy radiation were compared with sham control animals treated with four injections of equal volume saline (12 hours apart) and 17 Gy radiation 3 hours after the last injection. The tumor volume doubling time for saline + radiation group was 3.0 ± 0.3 versus 7.8 ± 0.3 days (mean ± SE) for FK866 + radiation. Because the doubling time of the untreated mammary carcinoma tumors is 2.4 days, it seems that the effect of sequential treatment of the combined modalities may be greater than additive.

FK866 alters the metabolism of mammary carcinoma in vivo and induces tumor acidity. The serial in vivo ³¹P MR spectra as shown in Fig. 2A show the effects of FK866 treatment by schedule I on mammary carcinoma tumor metabolism. The metabolite profiles (Fig. 2B) show a significant decline in NAD/β-NTP after treatment compared with the baseline values, 46% at 62 hours (P < 0.003) and 62% at 86 hours (P < 0.0005) after first dose. Similarly, the decrease in NAD/Pi was significant, 47% at 62 hours (P < 0.007) and 54% at 86 hours (P < 0.004) after the initial dose. Tumor energy status, as reflected through β-NTP/Pi and phosphocreatine/Pi showed no significant changes after FK866 treatment. The β-NTP/phosphomonoester was significantly reduced, 28% at 62 hours (P < 0.03) and 21% at 86 hours (P < 0.05) after first treatment. The pattern of the Pi peak at 62 hours after first dose was different (an enhanced splitting) compared with baseline; however, a pattern similar to the baseline was observed at the later time point (i.e., 82 hours). These fluctuations in Pi signal may indicate the pH heterogeneity of tumors due to the heterogenous nature of tumor perfusion over time as opposed to any drug induced effects. Serial studies of the untreated animals were not conducted in the present work. However, from our lab's extensive prior experience with ³¹P MRS studies on this (mammary carcinoma) tumor model (29, 30), there were no significant changes in the overall spectral pattern, Pi peak, and energy status observed during similar observation periods.

The effects of FK866 treatment (schedule II) on mammary carcinoma tumor metabolism are shown in Fig. 2C. The post-treatment spectrum obtained 39 hours after initial dosing showed a steep increase in signals from the phosphomonoester region and concurrent changes in pH and Pi pool distributions (Table 2). The split in Pi resonance indicates the possible existence of two distinct pH compartments. Whereas the factors contributing to these signals were not ascertained in the present study, the up field peak (Pi_major) and low field peak (Pi_minor) may represent intracellular and extracellular pools, respectively. In general, the intracellular compartments of tumors exhibit a neutral or slightly alkaline pH (31), but in the present study, the drug induced changes in the metabolic flux may lead to a decrease in intracellular pH levels. Compared with the pretreatment pH (7.28 ± 0.02), 75% of the total Pi pool (Pi_major) was observed to shift toward acidic (by 0.50 ± 0.11 units) and the remaining 25% (Pi_minor) is in the alkaline range (a shift of 0.22 ± 0.08 units) at 39 hours after first dose of FK866. From the metabolite profile (Fig. 2D), a significant decline in phosphocreatine/Pi (40%, P < 0.004), β-NTP/Pi (44%, P < 0.01), NAD/β-NTP (65%, P < 0.02), and NAD/Pi (77%, P < 0.001) levels were observed 39 hours after first dose of FK866. A relatively large drop observed in β-NTP/phosphomonoester levels (88%, P < 0.0001) compared with that of schedule I (20-28%) was mainly due to the large build up of signals from the phosphomonoester region in addition to the concurrent decrease in β-NTP levels. The broad unresolved signals from the phosphomonoester region could not be delineated further due to the limited achievable in vivo spectral resolution. Hence, high-resolution ³¹P MRS studies with PCA extracts of excised mammary carcinoma tumors were conducted as described in the following section.

FK866 interferes with the pathways of glycolysis, guanylate synthesis, phospholipids metabolism, and pyridine nucleotide pools. PCA extracts were prepared from cohorts of control and mammary carcinoma treated with FK866 according to schedule II (40 mg per kg per injection × 4), excised at 3 hours after the fourth injection, and analyzed in vitro using high-resolution ³¹P MRS. The spectra obtained from the control and 39 hours after first dose of FK866 are shown in Fig. 3. All the signal assignments were confirmed either by “spiking” extracts with original candidate compounds and/or comparison with published chemical shift values and titration curves (32, 33). The membrane lipid precursors phosphocholine (3.65 or 3.71 ppm), phosphoethanolamine (4.24 ppm), and the degradation products glycerophosphocholine (0 ppm) and GPE (0.96 ppm) were observed in both control and the post-treatment spectra in addition to the signals AMP at 4.16 ppm, UMP at 4.12 ppm (which could be the result of extraction artifacts due to hydrolysis of NTP), and the phosphoethanolamine carbamide at 4.35 ppm. Neutralization of extracts with bicarbonate has been reported to result in the formation of phosphoethanolamine-carbamate from phosphoethanolamine (34, 35). In the present study, spiking the tumor extracts with the commercially available phosphoethanolamine resulted in two peaks, a minor peak at 4.24 ppm and a major peak at 4.35 ppm, which we have subsequently assigned as phosphoethanolamine and phosphoethanolamine-carbamate, respectively. The post-treatment spectra displayed seven additional new peaks. Accumulation of the glycolysis intermediates fructose 1, 6-biphosphate (F-1, 6-P) at 4.33 (6-P) and 4.45(1-P) ppm, glycerol-3-phosphate (G-3-P) at 4.51 ppm, glucose-6-phosphate (G-6-P) at 4.64, and DHAP at 5.14 ppm were noted, which indicates that FK866 affect the pathways of mammary carcinoma tumor glycolysis. The IMP signal appearing at 4.14 ppm after treatment points toward possible interference of guanylate synthesis by FK866. Similarly, the NADP⁺ (2′P) resonance identified to accumulate at 3.98 ppm post-treatment suggests alterations in the pyridine nucleotide pools. In addition, the effects of FK866 on mammary carcinoma phospholipid metabolism were noticed through the post-treatment decrease in phosphocholine as well as an increase in glycerophosphocholine levels. The post-treatment changes in the spectral profile qualitatively explain the observed spectral changes, particularly the signal build-up in the phosphomonoester region, and possible mechanisms of action of the drug in vivo.

FK866-treated tumors exhibit loss of mitotic figures and apoptotic morphology. Untreated mammary tumors were uniform in their cell content with modest endothelialization and sparse areas of necrotic cells with an intermittent apoptotic population of <5%. Tumors were predominately homogeneous with mitotic activity that was well visualized (Fig. 4A). Tumors from animals treated with FK866 alone were studied at both 24 and 48 hours post-treatment and were consistent in the expression of cellular damage (Fig. 4B). These tumors showed discrete areas of coagulative necrosis along the endothelial vessels, elevated levels of apoptotic nuclei (>15%), and intermittent areas of diffuse necrosis when compared with tumors from untreated mammary carcinoma controls that contained only nominal necrotic foci. Loss of mitotic figures and increased necrotic cells were the principal observations at both 24 and 48 hours, although necrotic and pyknotic staining was also observed at a later time (72 hours) point (data not shown). Tumors sectioned at 24 and 48 hours after FK866 were void of any mitotic activity. Carcinoma cells were morphologically vacuolated at 48 hours, demonstrating cytoplasmic contraction and loss of membrane junctions and general cellular dispersion. Blood cell infiltration (hyperemia) and fine endothelial damage was consistent with the dying tissue at 48 hours. Apoptotic bodies were pronounced in the tissues sampled from FK866 treated tumors, although the predominant damage seemed both necrosis and pyknosis (Fig. 4B), which was more apparent in the 48-hour histologic samples observed.

FK866 induces depolarization of mitochondrial membrane potential, increased Annexin V binding, and alters cell cycle distributions of mammary carcinoma tumor cells. Untreated control cells stained with the fluorescent dye JC-1 exhibited mitochondrial staining that emitted an orange/red fluorescence. Cells treated with FK866 exhibited lower intensity staining that was observed within 24 hours after the last dose cycle. The intensity of the fluorescence loss as measured by the depolarization (shift to green fluorescence) and the reduction in forward scatter is graphically depicted (Fig. 5A) and was consistent in all treatment groups as an indicator of the time-dependent changes of membrane potential induced by FK866. The representative fluorescence-activated cell sorting scatter plots from different time points are shown in Fig. 5B. Maximum membrane potential reduction was observed at 24 and 48 hours (post FK866) followed by a rebound at 72 hours. The significant loss of membrane potential when compared with control values (0 hour) is representative of (∼75%) mitochondrial-induced drug inhibition with a maximal effect at 24 hours followed by recovery. This time dependence was similar to elevated levels of apoptosis noted both by Annexin V staining and histology.

Annexin V-FITC and PI effects are summarized and graphically depicted in Fig. 6A and B and Fig. 6C and D, respectively. The simultaneous staining of tumor cells with FITC-Annexin V (green fluorescence) and the nonvital dye propidium iodide (red fluorescence) allows the discrimination of intact cells (FITC⁻PI⁻), early apoptotic (FITC⁺PI⁻), and late apoptotic or necrotic (FITC⁺PI⁺) cells. Annexin V–positive, untreated mammary carcinoma tumors (control) averaged 21% apoptotic values (Fig. 6A) and were higher on average when compared with measurements of cells in culture by this laboratory. This in part may be more reflective of mammary carcinoma that have higher intrinsic hypoxic values (∼50%; ref. 36) than other reported murine carcinomas, or a result of the technique associated with the cell disruption/dissociation (see Materials and Methods). Changes in phosphatidylserine asymmetry, which was analyzed by measuring Annexin V binding to the cell membrane, knowing that any procedure that affects the integrity of the plasma membrane and phosphatidylserine can result in cells positive for Annexin V. Hence, we used Hoechst staining (data not shown) for secondary apoptotic assessment of cell suspensions and H&E morphologic verification of intact tumor samples. When the Annexin V–positive untreated control tumors were compared with FK866-treated cohorts, greater number of positive stained cells were observed at the 48 and 72 hours interval post-treatment. Similarly, transitional phase block in mammary carcinoma tumor cell cycle progression was coincident with G₂-M accumulation at 24, 48, and 72 hours after FK866 treatment. Untreated mammary carcinoma average 22% and 31% of cells in G₂-M at 24 and 48 hours, respectively. However, FK866-treated tumors show apparent shifts in cell cycle effects that show the appearance of transitional phase blocks at G₂-M resulting in increased levels to 50% at 24 to 72 hours and the subsequent reduction in cycling G₁ cells within the 24- to 72-hour period after FK866 treatment. Maximum effects on mammary carcinoma inhibition of cell cycle progression induced by FK866 were observed up to 3 days after the last dose. This change in cell cycle distribution is consistent with the observed cytocidal effects (apoptosis/necrosis; Fig. 4) and the tumor growth delays of 5.2 ± 0.5 days (Table 1) after FK866 treatment.

The data presented shows that the NAD⁺ synthesis inhibitor, FK866, retards mammary carcinoma tumor growth at two different dose schedules. In addition, when used at the higher dose schedule, FK866 was found to increase mammary carcinoma tumor radiation sensitivity when compared with radiation alone. We attribute these effects largely to FK866-induced loss of bioenergetic status and apoptotic cell death in mammary carcinoma tumors. The pronounced apoptosis observed through the histologic sections of FK866-treated tumors and increased Annexin V staining of tumor cells were supported by post-treatment depolarization of the mitochondrial membrane potential and the partial arrest of mammary carcinoma tumor cells at G₂-M phase of the cell cycle resulting in acute loss of mitotic figures and subsequent cellular death.

At the tumor metabolic level, the longitudinal effects observed with FK866 were also dose dependent. At the lower dose (schedule I), there was a significant decline in NAD/β-NTP, NAD/Pi, and β-NTP/phosphomonoester after treatment, although there were no significant change in mammary carcinoma tumor energy profile. However, at higher doses (40 mg/kg), in addition to considerable decline in NAD⁺ levels, a significant reduction in tumor energetic status and pH and a large signal buildup in the phosphomonoester region of the MRS profile (Fig. 2C) were noted within 24 hours post-treatment. This difference in treatment induced metabolic changes between two schedules suggest that dosing schedule I may be below a threshold dose necessary to exert a significant biochemical effect. The elevated resonance at the phosphomonoester region observed with schedule II contains signal contributions from several different metabolites, which reveals a wide array of effects including, modulations of the cellular glucose, guanylate, and phospholipid metabolisms and altered cellular pyridine nucleotide pools. The accumulation of glycolytic intermediates fructose-1,6-P, glucose-6-P, glucose-3-P, and DHAP observed from the PCA tumor extracts (measured longitudinally to coincide with the in vivo measurements) after FK866 treatment, indicated alterations along the glycolytic pathway, possibly due to the inhibition of NAD⁺-dependent glyceraldehyde phosphate dehydrogenase activity. This could explain the reductions observed in tumor energy status after FK866 treatment. The present results are in agreement with the modulations in glycolysis observed from cultured HL60 human promyelocytic leukemia cells undergoing apoptosis after camptothecin treatment (23) and L1210 murine lymphocytic leukemia cells undergoing apoptosis after nitrogen mustard mechlorethamine treatment (24). In addition, the appearance of IMP signals post- treatment indicate that FK866 induced reduction in cellular NAD⁺ might also lead to inhibition of another important NAD⁺-dependent pathway, the rate-limiting IMP dehydrogenase reaction step in de novo synthesis of guanylates, GTP and dGTP. Whereas speculative, each element could conceivably contribute to the GTP reduction, which could in turn result in depletion of cellular guanylate levels. Although guanylate pools can be replenished through the guanine salvage pathway, the level of circulating guanine is low in dividing cells and this route is probably insufficient to satisfy the demand for guanylate moieties in tumor cells (37). Thus, the de novo guanylate-synthetic pathway may be crucial for the biochemical recovery of tumor cells. Malignant transformation and progression of tumor cells are codependent to an increase in IMP dehydrogenase (type II), GMP synthetase activities, and the elevated concentrations of GTP and dGTP (38). Similarly, dGTP is an essential precursor of DNA synthesis comparable to GTP for RNA, although it is more implicated in the promotion of polymerization of microtubules. A decrease in GTP concentrations might also interfere with the biosynthesis of CTP and ATP (38). Hence, any reduction in intracellular guanylate pools resulting from FK866-induced inhibition of IMP dehydrogenase activity could ultimately contribute to the G₂-M accumulations observed when compared with untreated controls (Fig. 6C), subsequent transitional phase block, the loss of mitotic figures (proliferation), and the inhibition of mammary carcinoma tumor growth after 24 to 72 hours.

As previously mentioned, the mammary carcinoma tumor has a considerable hypoxic (∼50%) population in situ (36). The physiologic demands of both acute and chronic hypoxia are equally considerable as they relate to the up-regulation and down-regulation of gene expression that has been implicated in subsequent proliferation, metabolism, angiogenesis, and apoptosis (39). Although speculative, the coordinated effects of reduced tumor blood flow (perfusion depletion) and tumor acidification in the presence of the NAD depletion by FK866 may also be implicated in the cytotoxicity observed in this tumor. Cellular pH homeostasis is tightly linked to the control of apoptosis (40–44). Intracellular acidification has been reported to accompany the early events of apoptosis and facilitate activation of caspases (caspase 3 and caspase 9) following cytochrome c release (40). In cell culture studies with a T-lymphocyte cell line deprived of the growth factor interleukin-2 (42, 43), neutrophils dying in the absence of survival factors (44), HL60 and ML-1 cells treated with etoposide (45) and Jurkat cells induced to enter apoptosis by UV radiation, cycloheximide, or anti-Fas antibody (46), all exhibited a low pH_i. Similarly, in a more recent study, decreased intracellular pH was reported to be associated with camptothecin-induced apoptosis in leukemic cells (47). Thus, acidification has potential as a surrogate marker for the irreversible prelude to cell death. Moreover, in the present experiments, the FK866-induced increase in glycerophosphocholine levels that were accompanied with a decrease in phosphocholine may also be linked with the acidification of mammary carcinoma observed. Under acidosis and inhibition of glycolytic ATP production, cultured mammalian cells have been shown to compensate their energy demand through phospholipid catabolism (i.e., activation of phosphatidylcholine breakdown resulting in an accumulation of glycerophosphocholine; ref. 48). Inactivation of the glycerophosphocholine-diesterase enzymes at a reduced (acidic) pH would prevent choline recycling and has also been suggested to cause a decrease in phosphocholine levels (48). Thus, the conditions of FK866 induced mammary carcinoma tumor acidosis and glycolytic inhibition coupled with significant reduction in the tumor energy status would support the hypotheses of the energy consuming apoptotic pathway induction (49–51). Maintenance of cellular energy would in part be derived from oxidation of fatty acids (i.e., breakdown of phosphatidylcholine), which could be reflected as the increased glycerophosphocholine signals that we observed from the ³¹P MR spectrum of mammary carcinoma extracts.

In addition, the appearance of NADP⁺ (2′P) signal at 3.98 ppm that was observed at 39 hours after initial dose of FK866 treatment might implicate the alterations in cellular redox balance of the pyridine nucleotide pools. The redox state of NAD(P) has been suggested to regulate the mitochondrial permeability transition (52), the decisive step in apoptosis. This regulation of mitochondrial permeability transition is mediated by the NAD(P) redox status–dependent changes in mitochondrial membrane potential (Δψ_m), as controlled by membrane potential-sensitive NADP transhydrogenase (53). Thus, the NADP⁺ signal detected with the post-treatment spectra correlates with the loss of mammary carcinoma tumor mitochondrial membrane potential (Fig. 5) observed at 24 to 48 hours after FK866 treatment. This would be similar to the oxidation/depletion of pyridine nucleotides in Jurkat cells observed during Fas- and ceramide-induced apoptosis, which has been reported to have close temporal and functional correlation with loss of Δψ_m, intracellular acidification, and caspase-3 activation (54). Thus, a massive depletion/oxidation of NAD(P)H is suggested to be a major component of the apoptotic pathway and may have the same predictive value for cell death as has the dissipation of Δψ_m (54).

Recently, FK866 has been introduced into clinical trials as a novel NAD biosynthesis inhibitor in solid tumors (17, 18). Four patients had stable disease for varying periods of time. The limiting toxicity was thrombocytopenia. Thus, the drug by itself was not that active which is similar to our preclinical findings. The data presented here suggests that further studies of this drug should be in combination with other cytotoxic agents such as radiation. ³¹P NMR studies could be used to provide biochemical evidence of drug activity and could be used to optimize the interval between FK866 and cytotoxic agents.

In conclusion, our results suggest that high-resolution in vitro ³¹P MRS studies on tumor extracts in parallel with in vivo ³¹P MR of solid tumors provides comprehensive information about the tumor metabolic changes associated with cytotoxic chemotherapy induced apoptosis. A wide range of metabolic changes observed in the present study shows that the FK866-induced apoptosis in mammary carcinoma is associated with (i) a reduction in cellular NAD⁺ levels, (ii) inhibition of glycolysis and guanylate synthesis along with tumor acidification, (iii) changes in pyridine nucleotide redox pools, and lastly (iv) the activation of phospholipid catabolism in response to the energy demands of the apoptotic cascade. To our knowledge, the present work is the first to report both in vivo and in vitro ³¹P MRS metabolic changes from the same tumor model under identical conditions of treatment with a highly specific inhibitor of NAD synthesis, FK866. These noninvasively obtained measurements could provide a surrogate marker for detecting biologically active doses of FK866, which could be used for subsequent trials in combination with other cytotoxic agents. In addition, as the utility of ¹H-decoupled-³¹P MRS is currently being explored to identify prognostic markers in clinical settings (55), our present results could be useful in the noninvasive evaluation of other novel proapoptotic agents under clinical trials (56).