Metabonomic Evaluation of Metabolic Dysregulation in Rats Induced

Nov 15, 2007 - ... be addressed: Dr. Donald G. Robertson, 1118 Cutler Circle, Saline, MI 48176. ... William R. Foster , Donald G. Robertson , Bruce D...
0 downloads 0 Views 287KB Size
Download PDF
Chem. Res. Toxicol. 2007, 20, 1871–1877

1871

Metabonomic Evaluation of Metabolic Dysregulation in Rats Induced by PF 376304, a Novel Inhibitor of Phosphoinositide 3-Kinase Donald G. Robertson,*,† Kaushik Datta,‡ Dale Wells,† Laura Egnash,† Lora Robosky,† Matt Manning,† Cynthia Rohde,† and Michael D. Reily† Metabonomics EValuation Group and Drug Safety Research and DeVelopment, Pfizer Global Research and DeVelopment, Ann Arbor, Michigan 48105 ReceiVed June 5, 2007

Phosphoinositide 3-kinase (PI3K) is an enzyme fundamental to the regulation of various metabolic processes. Metabonomic studies were undertaken in order to gain mechanistic insight into significant, yet unexplained, toxicity issues associated with PF 376304, a nonspecific PI3K inhibitor under development for anti-inflammatory indications. Two experiments were conducted in which rats were given daily doses of up to 1000 mg of PF 376304/kg for as long as 7 days. Mortality rapidly ensued (within 72 h) at doses of g300 mg/kg. Doses of g100 mg/kg were associated with a profound but transient glucosuria. Despite the magnitude of this effect, within 72 h urinary glucose excretion in surviving animals returned to control levels even with continued dosing. Other metabolic effects associated with drug treatment included increased urinary β-hydroxybutyrate and creatine and decreased citrate. A time-course study revealed elevated serum glucose within 1 h, followed by increases in serum insulin and decreases in serum triglycerides. Serum corticosterone was also significantly elevated within 1 h of treatment. All metabolic effects were largely reversed within 24 h of administration of the third daily dose and remained that way through day 7. The likely explanation for the onset of effects involves the role of PI3K in regulation of glucose at multiple points, but the reversal of the effects in the presence of continued exposure to the drug has not been explained. Finally, the data demonstrate the power of metabonomics technology in mechanistic toxicology investigations. Introduction Metabonomics has proven to be a powerful mechanistic tool across the spectrum of biological sciences (1–3). In particular, the technology has shown great promise within pharmaceutical discovery and development (4–6). It stands to reason that any tool that can aid in evaluation of fundamental metabolic disruption at the level of the whole organism would be beneficial for evaluating the efficacy and toxicity of putative therapeutic agents, especially those that are designed to interact at basic levels of signal transduction. Most kinases fall within that category, and phosphoinositide 3-kinase (PI3K) in particular has been shown to play a key role in such basic cellular processes as proliferation, differentiation, survival, chemotaxis, trafficking, and glucose homeostasis (7, 8). In glucose homeostasis, PI3K functions as a signaling molecule downstream of the insulin receptor (9). After insulin binds to the insulin receptor, the intrinsic tyrosine kinase activity of the receptor is activated, and the receptor undergoes autophosphorylation. This recruits adapter proteins, such as the insulin receptor substrate (IRS) proteins, to the insulin receptor, and these proteins in turn act to recruit other downstream signaling molecules and kinases, such as PI3K, to the cell surface. Once brought to the cell membrane, PI3K ultimately proceeds to activate serine/threonine protein kinase, protein * To whom correspondence should be addressed: Dr. Donald G. Robertson, 1118 Cutler Circle, Saline, MI 48176. Phone: (734) 429-0767. E-mail: [email protected]. † Metabonomics Evaluation Group. ‡ Drug Safety Research and Development.

kinase B (AKT), and atypical protein kinase C (aPKC) (10). In muscle and adipose tissue, AKT has been shown to play a pivotal role in regulation of glycogen synthase activity via glycogen synthase kinase-3 (GSK-3), activation and translocation of the insulin-regulated glucose transporter 4 (GLUT4) to the cell surface, and regulation of fatty-acid synthesis via ATP citrate lyase (ATP-CL). In the liver, AKT inhibits gluconeogenesis by blocking transcription of gluconeogenic enzymes, such as phosphoenolpyruvate carboxykinase (PEPCK) and glucose-6-phosphatase (G6Pase). Meanwhile, aPKC acts in the liver to regulate lipid synthesis via sterol-regulatory-element binding protein-1c (SREBP-1c) expression. It also plays a role in translocation of the GLUT4 transporter to the plasma membrane in both muscle and adipose tissue. PI3Ks have been divided into three distinct classes (I, II, and III) (8, 11). Three Class IA isoforms of the p110 subunit (R, β, and δ) have been identified in mammals and show different expression profiles. While the p110R and p110β isoforms are expressed in all tissues, the p110δ isoform is mainly found in leukocytes (12, 13). Also, each isoform appears to have a slightly different cellular role. The p110R isoform is commonly mutated or has its expression amplified in human cancers (14). Recent reports also indicate that it plays a role in insulin signaling (15, 16). The p110β isoform has been shown to be involved in insulin and lysophosphatidic acid signaling as well as in cell motility (17, 18). The p110δ isoform mainly functions in the immune system (19, 20). Therefore, it is not surprising that the PI3K pathway is being evaluated for therapeutic potential in the areas of metabolic syndrome, oncology, and inflammation (21–23).

10.1021/tx7002036 CCC: $37.00  2007 American Chemical Society Published on Web 11/15/2007

1872 Chem. Res. Toxicol., Vol. 20, No. 12, 2007

Figure 1. Structure of PF 376304 [4-cycloheptyloxy-2-morpholin-4yl-pyrimidine-5-carboxylic acid (2H-tetrazol-5-yl)-amide].

PF 376304 (Figure 1) was developed as a nonspecific inhibitor of the Class I isoforms for potential anti-inflammatory indications. The compound was found to be active in rat inflammation models at oral doses of e10 mg/kg. However, during its early development, significant mortality was noted in rats with this compound and several of its analogues at doses only slightly above the upper end of the therapeutic range (30–100 mg/kg). In an effort to understand the etiology of the mortality, two studies utilizing metabonomics technology were conducted in order to better understand the metabolic consequences of administration of this compound.

Materials and Methods Compound and Administration. PF 376304 was synthesized in-house and had NMR, IR, and MS data consistent with its structure. Chemical analysis indicated the synthetic material was 98.03% active material, and dosing was adjusted by a factor of 1.02 to compensate. The compound was prepared as a suspension in an aqueous vehicle containing 0.5% methyl cellulose/5% polyethylene glycol (PEG 200) and administered by oral gavage on a milligram per kilogram of body weight basis using a dosing volume of 10 mL/kg. Animals. Sprague-Dawley [Crl:CD(SD)BR] rats between 6 and 8 weeks of age were obtained for Experiments 1 and 2 from the Portage, MI, and Raleigh, NC, facilities of Charles River Laboratories, respectively. At study initiation, male rats weighed from 194 to 228 g, and female rats weighed from 164 to 182 g. Routine Animal Husbandry. Animals were housed in an AALAC-accredited facility, and all in vivo studies were reviewed and approved by the IACUC. When not in metabolism cages, animals were housed in individual cages in rooms having temperature (70–78 °F) and humidity (30–70% RH) control and a 12 h light cycle. Animals were fed Purina Laboratory Diet Certified Rodent Chow 5002 and watered ad libitum. Urine Collection. When urine was required, animals were placed into individual plastic metabolism cages (Harvard Apparatus, Holliston, MA), where they remained for the duration of urine collection. Food and water were available ad libitum. Animals were allowed to acclimate to metabolism cages for at least 24 h prior to compound administration. At the specified intervals, urine was collected into tubes containing 1 mL of 1% sodium azide maintained at 0 °C. Urine was centrifuged to remove particulates and frozen for later NMR analyses. One-Dimensional 1H NMR Spectroscopy. Samples for NMR analysis were prepared by mixing 500 µL of urine with 250 µL of buffer in 96-well plates. The buffer was added to provide some normalization of the urinary pH. After mixing, plates were centrifuged to sediment insolubles. The buffer was 0.2 M sodium phosphate buffer in 80/20 H2O/D2O (pH 7.4) containing 1.0 mM TSP (sodium 2,2′,3,3′-deutero-3-trimethylsilylpropionate, an internal NMR reference standard) and 3 mM sodium azide. 1H NMR data were acquired using a Varian Inova NMR spectrometer operating at 600.36 MHz for 1H and equipped with an 1H-{15N, 13C} flow probe (120 µL active volume) and a Varian automated sample transport (VAST) accessory. Sample (450 µL) was injected into the probe with no push solvent. Two cell rinses with isotonic phosphate buffer were completed between samples. The push buffer was prepared by mixing 1 part 0.2 M sodium phosphate buffer

Robertson et al. (pH 7.4) with 2 parts water. One-dimensional 1H NMR spectra were acquired at 27 °C using a one-dimensional NOESY pulse sequence including water presaturation and a mixing time of 100 ms. A total of 64 scans were collected with 64K data points, an acquisition time of 2.73 s, an interpulse delay of 1 s, and a sweep width of 12 kHz. NMR Data Analysis. Spectra were processed and analyzed using in-house Metabonomi software (International Patent Application WO2004038602A1). Spectra were normalized to the total integrated spectral area minus the region containing water and urea from 6.0 to 4.5 ppm. For quantification, the NMR signal arising from the glucose anomeric proton at 5.23 ppm was integrated using Metabonomi for all NMR spectra. To obtain absolute concentrations, the resulting integral was normalized to the internal standard TSP, which was present in all samples at 0.33 mM, and multiplied by 2.78 to account for the 36/64 R/β anomeric ratio for glucose. Concentrations calculated in this manner are approximate and subject to error caused by incomplete relaxation during the relatively short interpulse delay used in these experiments. Experiment 1 Design. In the initial experiment, PF 376304 was administered to three rats per sex at doses of 0 (vehicle), 100, 300, and 1000 mg/kg daily for 7 days. Clinical observations were recorded daily, and routine clinical chemistry and hematology parameters were measured (in overnight fasted samples) at study termination (day 8). Measured chemistry parameters included serum alanine aminotransferase (ALT), aspartate aminotransferase (AST), alkaline phosphatase (AP), albumin, bilirubin, cholesterol, creatinine, γ-glutamyl transferase (GGT), glucose, protein, urea nitrogen, Ca2+, Cl-, K+, and Na+. Hematology parameters included red blood cell (RBC) count, white blood cell (WBC) count, WBC differential, reticulocyte count, platelet count, mean platelet volume, hemoglobin (Hgb), hematocrit (Hct), mean corpuscular volume (MCV), mean corpuscular hemoglobin (MCH), and mean corpuscular hemoglobin concentration (MCHC). Plasma PF 376304 levels were determined for all groups on day 1 and the 100 mg/kg group on day 7, approximately 1, 4, 8, and 24 h postdose. Urine for metabonomic analysis was collected in a single 24 h collection prior to the first dose (pretest) and in two collections at intervals of 0-8 and 8-24 h after the first and seventh doses. Surviving rats were euthanized 24 h following the seventh dose (day 8), and microscopic evaluation of heart, liver, lung, kidneys, and thymus was conducted on all animals. Experiment 2 Design. The second experiment was conducted to follow up on the major findings from Experiment 1. PF 376304 was administered to six male rats per group at doses of 0 (vehicle), 30, 100, and 300 mg/kg daily for 7 days. Clinical observations were made daily. Blood samples (nonfasted) were collected from the first three rats per group 1, 2, and 24 h following the first dose, from the second three rats per group 2, 8, and 48 h following the first dose, and from all animals at study termination (day 8) for determination of serum Na+, Cl-, glucose, triglycerides, insulin, and corticosterone. Urine for metabonomic analysis was intended for collection for 24 h prior to the first dose (pretest) and every 4 h up to 48 h after the first dose, with 24 h samples collected thereafter. However, because of equipment malfunctions, the 8–24 h urine samples from day 1 were combined into one sample per animal, as were the 4–8 h samples and 12–24 h samples on day 2. Subsequent 24 h samples were collected as indicated. Animals were euthanized and discarded after collection of terminal urine and blood samples (day 8). Toxicokinetics. In Experiment 1, plasma drug levels were determined using an LC/MS/MS method having a lower limit of quantitation (LOQ) of 4.88 ng/mL. Concentrations below the LOQ were considered to be zero. Steady state was assumed for day 7, and the 24 h plasma concentrations were used as the t ) 0 concentrations for calculation of toxicokinetic parameters. Watson version 6.4 software (Watson Software Systems, Irvine, CA) was used to calculate the following toxicokinetic parameters: AUC0–24, the area under the plasma concentration–time curve from t ) 0 to t ) 24 h; Cmax, the maximum plasma concentration; and tmax, the time at which Cmax was observed.

Metabonomic EValuation of a NoVel PI3K Inhibitor

Chem. Res. Toxicol., Vol. 20, No. 12, 2007 1873

Table 1. Toxicokinetic Results from Experiment 1 dose (mg/kg)

study day

100

1 7

300

1

1000

1

sex

Cmax (µg/mL)

tmax (h)

AUC0–24 (µg h mL–1)

male female combined male female combined male female combined male female combined

31.0 38.9 35.0 43.5 78.7 57.5 164 164 164 180 177 179

6.7 6.7 6.7 3.0 2.0 2.6 24 24 24 24 18.7 21.3

491 619 555 667 1280 912 2210 2350 2280 2860 2180 2520

Statistics. Statistical analyses were conducted on body weight, weight change, food consumption, and quantitative clinical laboratory data. Treatment comparisons were performed on ranktransformed data using a dose-trend test sequentially applied at the two-tailed 1% and 5% significance levels within one-factor analysis of variance (ANOVA). Dunnett’s test replaced the sequential trend test if the overall linear trend test was not significant at the 5% level and a quadratic trend was significant at the 1% level. All parameters were analyzed separately for each sex and time period.

Results Experiment 1. Clinical Signs. No clinical signs were noted in controls or in males at 100 mg/kg. A single 100 mg/kg female was sacrificed in moribund condition on day 6 with signs of hypoactivity, hypothermia, weight loss, and a red-stained muzzle. A second 100 mg/kg female had a red-stained muzzle noted only on day 8 (at termination). No other clinical signs were noted in the control or 100 mg/kg groups. All animals in the 300 and 1000 mg/kg groups began displaying significant signs of toxicity within the first 48 h after dosing, with all animals either found dead or sacrificed in moribund condition within 24 h of the second dose. In all surviving 100 mg/kg animals, weight gain suppression or loss (8% in females) was noted, along with decreased average food consumption (23% in females). Clinical Chemistry. As a result of the early termination of the 300 and 1000 mg/kg dose groups, clinical pathology measurements were determined only in the control and 100 mg/ kg dose groups. While changes in several parameters were flagged as being statistically significant in the 100 mg/kg group relative to the control group, no measurements fell outside laboratory reference ranges. Therefore, none of the measurements were considered to be toxicologically relevant. Pathology. Other than agonal changes (lung or stomach hemorrhage and thymus lymphocytolysis) noted in several animals found dead or sacrificed moribund, no clearly drugrelated findings were noted in examined tissues. Toxicokinetics. Values of the toxicokinetic parameters from Experiment 1 are summarized in Table 1. On day 1, both AUC0–24 and Cmax increased roughly proportionally to dose between the 100 and 300 mg/kg doses but were essentially flat between 300 and 1000 mg/kg. Early termination prevented subsequent measurements in the 300 and 1000 mg/kg dose groups; however, the delayed tmax on day 1 for both the 300 and 1000 mg/kg groups relative to the 100 mg/kg group suggests that exposure could have increased in those groups with subsequent doses. Metabonomics. The most prominent change in urinary spectral profiles was a remarkable increase in urine glucose. Sample 1H NMR 24 h urine spectral profiles are presented in Figure 2. The most overt effect of drug treatment was a

remarkable increase in urine glucose excretion (in excess of 2 orders of magnitude in some cases). A time course of the changes for urinary excretion of glucose is presented in Figure 3. Due to the polyuria noted in the second study, data for glucose are reported as glucose excretion rate per hour for both studies. Beyond the profound glucosuria, the most striking effect was the rapid return to normal urinary glucose excretion in surviving 100 mg/kg rats. By the time the seventh dose was administered (the 144–152 h sample, which did not include a urine sample from the animal that died), urine glucose excretion in the 100 mg/kg group was comparable to the control (Figure 3). In addition to the changes in glucose, other observed drug-related metabolic changes included decreased urinary citrate 8 and 24 h after the first dose (data not shown) and increased urinary β-hydroxybutyrate (BHB) and creatine, particularly in rats at the 300 and 1000 mg/kg doses (Figure 2). Experiment 2. Clinical Signs. A single 30 mg/kg animal exhibited a red-stained muzzle on days 3–6 of dosing. No other clinical signs were noted in the control group or the 30 or 100 mg/kg treatment groups. All 300 mg/kg animals were either found dead or sacrificed in moribund condition within 24 h of the second dose. Clinical signs noted in most animals prior to death or sacrifice included hypothermia, dyspnea, reduced feces, hypoactivity, lacrimation, and red-stained muzzle. Control and 30 mg/kg animals gained on average 40 ( 11 and 35 ( 8 g, respectively, over the course of the study, while the 100 mg/kg group exhibited significant weight gain suppression, gaining only 19 ( 8 g over the same time period. Due to the clinical condition of the animals, terminal body weights were not determined for the 300 mg/kg group. Clinical Chemistry. The serum glucose, triglyceride, insulin, and corticosterone data are summarized in Figures 4–7. A doserelated increase in serum glucose was evident within 1 h of dosing. Serum glucose was significantly elevated at all time points for the 300 mg/kg group, reaching a peak serum concentration roughly 2 times the vehicle control from 4 to 8 h after dosing with a gradual diminution by the 24 h time point (Figure 4). The 100 mg/kg dose group also had significantly elevated serum glucose from 1 to 8 h postdose, but to a much lesser extent than the 300 mg/kg group. Serum glucose in the 30 mg/kg dose group was significantly elevated only at the 2 h time point. No effect on serum glucose was evident in the 0, 30, or 100 mg/kg dose groups subsequent to the 8 h time point. Serum insulin levels were much more variable than glucose levels but generally followed the same trend observed with serum glucose (Figure 5). Insulin increases appeared to follow the glucose increases, with the first significant increase evident at 2 h. Insulin remained elevated in the 300 mg/kg group at 24 h (just failing to meet statistical significance) but returned to control levels for the remaining treatment groups at 24 h and subsequent time points. Serum triglycerides were inversely related to serum glucose and insulin, with dose-related decreases evident from 4 to 24 h after dosing (Figure 6). Serum corticosterone was elevated relative to the control in all drugtreatment groups within the first 4 h after dosing (Figure 7), with similar increases evident in both the 100 and 300 mg/kg treatment groups. Corticosterone remained elevated at the 8 and 24 h time points in the 300 mg/kg group but was comparable to the control from 8 h on in the 30 and 100 mg/kg groups. While statistically significant effects on serum Na+ and Clwere occasionally noted, all group mean values fell within the normal laboratory ranges and therefore were not considered toxicologically relevant.

1874 Chem. Res. Toxicol., Vol. 20, No. 12, 2007

Robertson et al.

Figure 2. Unnormalized 600 MHz 1H NMR spectra of urine samples collected predose (bottom trace) and 24 h postdose (top trace) from a rat treated with 300 mg of PF 376304/kg. The prominent increases in glucose-related resonances as well as in resonances for creatine and β-hydroxybutyrate (inset) are noteworthy.

Figure 3. Changes in urine glucose excretion induced by PF 376304 in Experiment 1. Values are reported as mean ( SD.

Figure 5. Changes in mean serum insulin induced by PF 376304 in Experiment 2. Values are reported as mean ( SD.

Figure 4. Changes in mean serum glucose induced by PF 376304 in Experiment 2. Values are reported as mean ( SD.

Figure 6. Changes in mean serum triglycerides induced by PF 376304 in Experiment 2. Values are reported as mean ( SD.

Metabonomics. Metabonomic analyses of Experiment 2, which had a more refined time course, confirmed the marked glucosuria evident in the first experiment (Figure 8). In the 100

mg/kg group (which survived till termination), glucose excretion rates appeared to drop sharply after the first dose, were slightly elevated again after the second dose, but were comparable to control from 72 h throughout the remainder of the study.

Metabonomic EValuation of a NoVel PI3K Inhibitor

Figure 7. Changes in mean serum corticosterone induced by PF 376304 in Experiment 2. Values are reported as mean ( SD.

Figure 8. Changes in urine glucose excretion induced by PF 376304 in Experiment 2. The inset shows mean 24-h urine volumes. Values are reported as mean ( SD.

Additionally, a dose of 30 mg/kg was established as a no-effect dose with regard to the glucose dysregulation. Corresponding to the increased urine glucose was a sharp increase in urine volume during the first 24 h after dosing in the 100 and 300 mg/kg dose groups (Figure 8 inset). While urine volume dropped for both groups by 48 h, average urine volume remained elevated (though not significantly) at subsequent time points in the 100 mg/kg group. While other findings were similar to those of Experiment 1, it was evident from the time course in this study that the increases in urine β-hydroxybutyrate occurred subsequent to initiation of the glucosuria; maximal elevations occurred after administration of the second dose (36–48 h after the initial dose) and were largely restricted to the 300 mg/kg dose group (Figure 9).

Discussion Despite concerted efforts by pharmaceutical companies to develop inhibitors of various PI3K isoforms (16, 24), there is a dearth of information on the in vivo sequelae of PI3K inhibition, particularly after repeated doses. Wortmannin and LY294002 are prototypical nonspecific inhibitors of Class I PI3Ks, and hundreds of papers have been published documenting their in vitro effects (25, 26) and, to a lesser extent, their effects in acute in vivo studies (27, 28). Bradley et al. (27) published some of the in vivo consequences of acute (2 h) PI3K inhibition in rats

Chem. Res. Toxicol., Vol. 20, No. 12, 2007 1875

Figure 9. Stacked 600 MHz 1H NMR spectra (2–6 ppm) of timed urine samples collected from a single rat given 300 mg of PF 376304/kg. Glucose and β-hydroxybutyrate resonances are labeled.

using wortmannin. Although they observed elevations in plasma insulin, the elevations were not preceded by increases in blood glucose within the 2 h observation period. We also observed insulin elevations, but they were preceded with slight but significant increases in serum glucose, providing a clear distinction between our results and theirs. The effects of PF 376304 on metabolism were in some respects what might have been expected from blockade of some insulin receptor activities, GLUT4 translocation, and blockade of the inhibitory effects of AKT on gluconeogenesis. The first observable effects after administration of PF 376304 appeared to be elevation of serum glucose, possibly due to the effects of the drug on GLUT4, followed quickly by a concomitant elevation in insulin. The fact that both insulin and glucose remained elevated suggests a dysregulation of the insulin receptor, consistent with PI3K inhibition. Supporting the identification of PI3K inhibition as a mechanism are the elevations in β-hydroxybutyrate, which are suggestive of increased gluconeogenesis. Increased β-hydroxybutyrate may have been due to increased acetoacetate production from acetylCoA (due to fatty acid oxidation), with the latter failing to gain entrance into the TCA cycle because of utilization of oxaloacetate in gluconeogenesis, as frequently occurs in diabetes. Oxaloacetate overlaps with several other peaks in the NMR spectrum, so its fate in the urine of treated animals could not be determined. The only caveats to this speculation are the delayed timing of the β-hydroxybutyrate increases, which significantly trailed the increases in glucose (Figure 9), and the fact that β-hydroxybutyrate increases cannot be considered diagnostic because this substance is commonly elevated in response to many toxicants (5). However, the decreases in serum triglycerides observed later in the process (g4 h) may also fit into this paradigm, as cells turn to fat for an increasing proportion of energy utilization, generating greater amounts of acetyl-CoA and thus exacerbating the ketonuria. While glucose was obviously significantly disrupted in both experiments, we were unable to define a cause of death of the animals. No microscopic evidence of a potential cause of death was found in dead and moribund animals in Experiment 1, though only a limited number of tissues were examined. Further, the serum levels of glucose seemed insufficient to explain the death of the animals, as they were not outside the range of serum levels seen in some rat diabetic models. Further work will be required to establish the actual cause of death in rats treated with PF 376304.

1876 Chem. Res. Toxicol., Vol. 20, No. 12, 2007

The significance of the increase in urinary creatine induced by PF 376304 is unclear. Creatine is considered one of the “usual suspects” in toxicological metabonomic evaluations and is frequently elevated by compounds as varied as vascular toxins, renal toxins, and hepatotoxins (5). Though some mechanistic interpretations of the significance of this interesting molecule have been attempted (29, 30), the role of this metabolite in the present case is not clear. Two alternatives to the PI3K inhibition scenario described above should be mentioned. First, elevated glucose is a common consequence of acute renal tubular toxicity. However, no evidence of renal lesions was noted in animals found dead or sacrificed moribund in Experiment 1. Additionally, there are several lines of evidence that suggest renal toxicity was not the cause of the glucosuria. First, we were unable to find reports of toxicant-induced glucosuria that approached the concentrations we saw in nondiabetic rat models in the absence of any other evidence of renal toxicity. Second, the glucosuria was concomitant with fairly remarkable increases in serum glucose and insulin and a decrease in serum triglycerides, which are not necessarily consistent with tubular toxicity. Third, the urine metabonomics profile was not consistent with tubular nephrotoxins that we or others have studied (31–34). Fourth, clinical pathology markers (blood urea nitrogen and creatinine) were unaffected in any surviving animal at termination. While tubular toxicity can resolve rather quickly, we would anticipate that there would have been some evidence of its effects in at least one surviving animal. Fifth, no effects were noted on serum electrolytes 1, 2, 4, 8, 24, or 48 h (the time of maximal effect) after dosing. The second alternative explanation of the glucosuria was suggested by the immediate and marked elevations in serum corticosterone. It could be argued that PF 376304 produced a stress-induced hyperglycemia. While stress may have contributed to the effect, there are several pieces of data that argue against stress as the sole explanation. First, even though corticosterone was elevated in all dose groups, glucose elevations were only minimally elevated in the 30 mg/kg group at one time point (2 h), and serum glucose levels in treated animals did not appear to correlate very well with concurrent corticosterone levels. Second, though the levels of corticosterone reached in this study were similar to those seen in rats subjected to stresses such as handling and anesthesia (35, 36), the concomitant increases in serum glucose were far greater. Third, the extent of the glucosuria was profound, in some cases constituting 10% (w/v) of the urine, which is far beyond what might reasonably be expected due to stress. Given the role of PI3K in glucose regulation, it may not be terribly surprising that an inhibitor of this enzyme has a profound effect on glucose levels in vivo. However, the severity of the glucose response upon PF 376304 treatment exceeded anything we anticipated. Because the doses employed in our studies were relatively high, it could be argued that the profound glucose dysregulation may have been a toxicological manifestation and not pharmacologically relevant, since the effect was not observed at 30 mg/kg, which was a pharmacologically active dose.1 Therefore, it could be argued that serum (or urine) glucose would serve as a simple biomarker for the untoward effects of the compound, as these effects were rapidly reversible even with continued dosing. This last observation presents one of the most intriguing findings of these studies: the fascinating reversibility of the hyperglycemia and glucosuria within as little as 72 h of 1

Unpublished observations.

Robertson et al.

administration of the first dose. While a simple explanation for this finding would be altered toxicokinetics over the time course of the study, this was clearly not the case, as drug levels in the surviving 100 mg/kg animals in Experiment 1 were actually slightly higher on day 7 than on day 1. Correspondingly, urine glucose excretion in the same group was markedly elevated (>20-fold) within 8 h of the first dose yet was normal by study termination (day 7). The ability of the rats to accommodate to the pronounced glucose dysregulation within 72 h in the presence of continuing drug exposure is startling and has no clear explanation. It raises the question of whether the reversal of the effect implies returning PI3K function, altered regulation of effectors downstream from PI3K, or altered regulation of some entirely unrelated pathway. All three possibilities present intriguing avenues for exploration of glucose regulation in rats and potential new targets for understanding metabolic disease. Finally, this study demonstrated the utility of metabonomic evaluation in early mechanistic toxicity evaluations. The cynic may argue that many of the conclusions drawn from this work could have as easily been made using a 25-cent dipstick, which seems obvious in hindsight. Of course, this argument assumes that the investigator would have had the foresight to collect samples within the first 48 h in this particular case. Metabonomics by its nature enables comprehensive analyses of numerous metabolites that can provide mechanistic insights into many types of toxicity. In this case, a single metabonomic study revealed effects that had gone unrecognized for a significant period of time.

References (1) Gibney, M. J., Walsh, M., Brennan, L., Roche, H. M., German, B., and van Ommen, B. (2005) Metabolomics in human nutrition: opportunities and challenges. Am. J. Clin. Nutr. 82, 497–503. (2) Griffin, J. L., and Nicholls, A. W. (2006) Metabolomics as a functional genomic tool for understanding lipid dysfunction in diabetes, obesity and related disorders. Pharmacogenomics 7, 1095–1107. (3) Schauer, N., and Fernie, A. R. (2006) Plant metabolomics: towards biological function and mechanism. Trends Plant Sci. 11, 508–516. (4) Robertson, D. G., Reily, M. D., and Baker, J. D. (2005) Metabonomics in preclinical drug development. Expert Opin. Drug Metab. Toxicol. 1, 363–376. (5) Robertson, D. G. (2005) Metabonomics in toxicology: a review. Toxicol. Sci. 85, 809–822. (6) Lindon, J. C., Holmes, E., and Nicholson, J. K. (2006) Metabonomics techniques and applications to pharmaceutical research & development. Pharm. Res. 23, 1075–1088. (7) Cantley, L. C. (2002) The phosphoinositide 3-kinase pathway. Science 296, 1655–1657. (8) Katso, R., Okkenhaug, K., Ahmadi, K., White, S., Timms, J., and Waterfield, M. D. (2001) Cellular function of phosphoinositide 3-kinases: Implications for development, immunity, homeostasis and cancer. Annu. ReV. Cell DeV. Biol. 17, 615–675. (9) Cohen, P. (2006) The twentieth century struggle to decipher insulin signalling. Nat. ReV. Mol. Cell Biol. 7, 867–873. (10) Taniguchi, C. M., Emanuelli, B., and Kahn, C. R. (2006) Critical nodes in signalling pathways: insights into insulin action. Nat. ReV. Mol. Cell Biol. 7, 85–96. (11) Vanhaesebroeck, B., Leevers, S. J., Panayotou, G., and Waterfield, M. D. (1997) Phosphoinositide 3-kinases: a conserved family of signal transducers. Trends Biochem. Sci. 22, 267–272. (12) Chantry, D., Vojtek, A., Kashishian, A., Holtzman, D. A., Wood, C., Gray, P. W., Cooper, J. A., and Hoekstra, M. F. (1997) p110δ, a novel phosphatidylinositol 3-kinase catalytic subunit that associates with p85 and is expressed predominantly in leukocytes. J. Biol. Chem. 272, 19236–19241. (13) Vanhaesebroeck, B., Welham, M. J., Kotani, K., Stein, R., Warne, P. H., Zvelebil, M. J., Higashi, K., Volinia, S., Downward, J., and Waterfield, M. D. (1997) P110δ, a novel phosphoinositide 3-kinase in leukocytes. Proc. Natl. Acad. Sci. U.S.A. 94, 4330–4335. (14) Zhao, J. J., Cheng, H., Jia, S., Wang, L., Gjoerup, O. V., Mikami, A., and Roberts, T. M. (2006) The p110R isoform of PI3K is essential for proper growth factor signaling and oncogenic transformation. Proc. Natl. Acad. Sci. U.S.A. 103, 16296–16300.

Metabonomic EValuation of a NoVel PI3K Inhibitor (15) Foukas, L. C., Claret, M., Pearce, W., Okkenhaug, K., Meek, S., Peskett, E., Sancho, S., Smith, A. J., Withers, D. J., and Vanhaesebroeck, B. (2006) Critical role for the p110R phosphoinositide-3-OH kinase in growth and metabolic regulation. Nature 441, 366–370. (16) Knight, Z. A., Gonzalez, B., Feldman, M. E., Zunder, E. R., Goldenberg, D. D., Williams, O., Loewith, R., Stokoe, D., Balla, A., Toth, B., Balla, T., Weiss, W. A., Williams, R. L., and Shokat, K. M. (2006) A pharmacological map of the PI3-K family defines a role for p110R in insulin signaling. Cell 125, 733–747. (17) Brachmann, S. M., Ueki, K., Engelman, J. A., Kahn, R. C., and Cantley, L. C. (2005) Phosphoinositide 3-kinase catalytic subunit deletion and regulatory subunit deletion have opposite effects on insulin sensitivity in mice. Mol. Cell. Biol. 25, 1596–1607. (18) Yip, S. C., El-Sibai, M., Hill, K. M., Wu, H., Fu, Z., Condeelis, J. S., and Backer, J. M. (2004) Over-expression of the p110β but not p110R isoform of PI 3-kinase inhibits motility in breast cancer cells. Cell Motil. Cytoskeleton 59, 180–188. (19) Okkenhaug, K., Bilancio, A., Farjot, G., Priddle, H., Sancho, S., Peskett, E., Pearce, W., Meek, S. E., Salpekar, A., Waterfield, M. D., Smith, A. J., and Vanhaesebroeck, B. (2002) Impaired B and T cell antigen receptor signaling in p110δ PI 3-kinase mutant mice. Science 297, 1031–1034. (20) Okkenhaug, K., Bilancio, A., Emery, J. L., and Vanhaesebroeck, B. (2004) Phosphoinositide 3-kinase in T cell activation and survival. Biochem. Soc. Trans. 32, 332–335. (21) Ward, S., Sotsios, Y., Dowden, J., Bruce, I., and Finan, P. (2003) Therapeutic potential of phosphoinositide 3-kinase inhibitors. Chem. Biol. 10, 207–213. (22) Hennessy, B. T., Smith, D. L., Ram, P. T., Lu, Y., and Mills, G. B. (2005) Exploiting the PI3K/AKT pathway for cancer drug discovery. Nat. ReV. Drug DiscoVery 4, 988–1004. (23) Ward, S. G., and Finan, P. (2003) Isoform-specific phosphoinositide 3-kinase inhibitors as therapeutic agents. Curr. Opin. Pharmacol. 3, 426–434. (24) Ruckle, T., Schwarz, M. K., and Rommel, C. (2006) PI3Kγ inhibition: towards an ‘aspirin of the 21st century’? Nat. ReV. Drug DiscoVery 5, 903–918. (25) Seol, J. W., Lee, Y. J., Kang, H. S., Kim, I. S., Kim, N. S., Kwak, Y. G., Kim, T. H., Seol, D. W., and Park, S. Y. (2005) Wortmannin elevates tumor necrosis factor-related apoptosis-inducing ligand sensitivity in LNCaP cells through down-regulation of IAP-2 protein. Exp. Oncol. 27, 120–124.

Chem. Res. Toxicol., Vol. 20, No. 12, 2007 1877 (26) Shimaya, A., Kovacina, K. S., and Roth, R. A. (2004) On the mechanism for neomycin reversal of wortmannin inhibition of insulin stimulation of glucose uptake. J. Biol. Chem. 279, 55277–55282. (27) Bradley, E. A., Clark, M. G., and Rattigan, S. (2007) Acute effects of wortmannin on insulin’s hemodynamic and metabolic actions in vivo. Am. J. Physiol. 292, E779–E787. (28) Chirieac, D. V., Davidson, N. O., Sparks, C. E., and Sparks, J. D. (2006) PI3-kinase activity modulates apo B available for hepatic VLDL production in apobec-1-/- mice. Am. J. Physiol. 291, G382–G388. (29) Clayton, T. A., Lindon, J. C., Everett, J. R., Charuel, C., Hanton, G., Le Net, J. L., Provost, J. P., and Nicholson, J. K. (2003) An hypothesis for a mechanism underlying hepatotoxin-induced hypercreatinuria. Arch. Toxicol. 77, 208–217. (30) Clayton, T. A., Lindon, J. C., Everett, J. R., Charuel, C., Hanton, G., Le Net, J. L., Provost, J. P., and Nicholson, J. K. (2004) Hepatotoxininduced hypercreatinaemia and hypercreatinuria: their relationship to one another, to liver damage and to weakened nutritional status. Arch. Toxicol. 78, 86–96. (31) Gartland, K. P., Bonner, F. W., Timbrell, J. A., and Nicholson, J. K. (1989) Biochemical characterization of p-aminophenol-induced nephrotoxic lesions in the F344 rat. Arch. Toxicol. 63, 97–106. (32) Anthony, M. L., Sweatman, B. C., Beddell, C. R., Lindon, J. C., and Nicholson, J. K. (1994) Pattern recognition classification of the site of nephrotoxicity based on metabolite data derived from proton nuclear magnetic resonance spectra of urine. Mol. Pharmacol. 46, 199–211. (33) Robertson, D. G., Reily, M. D., Sigler, R. E., Wells, D. F., Paterson, D. A., and Braden, T. K. (2000) Metabonomics: Evaluation of nuclear magnetic resonance (NMR) and pattern recognition technology for rapid in vivo screening of liver and kidney toxicants. Toxicol. Sci. 87, 326–337. (34) Lenz, E. M., Bright, J., Knight, R., Westwood, F. R., Davies, D., Major, H., and Wilson, I. D. (2005) Metabonomics with 1H-NMR spectroscopy and liquid chromatography-mass spectrometry applied to the investigation of metabolic changes caused by gentamicin-induced nephrotoxicity in the rat. Biomarkers 10, 173–187. (35) Balcombe, J. P., Barnard, N. D., and Sandusky, C. (2004) Laboratory routines cause animal stress. Contemp. Top. Lab. Anim. Sci. 43, 42– 51. (36) Vachon, P., and Moreau, J. P. (2001) Serum corticosterone and blood glucose in rats after two jugular vein blood sampling methods: comparison of the stress response. Contemp. Top. Lab. Anim. Sci. 40, 22–24.

TX7002036