Application of Genomics for Identification of Systemic Toxicity Triggers

Chem. Res. Toxicol. 2010, 23, 1025–1033

1025

Application of Genomics for Identification of Systemic Toxicity Triggers Associated With VEGF-R Inhibitors Hisham K. Hamadeh,*,† Marque Todd,†,§ Laura Healy,†,| James T. Meyer,‡ Annie M. Kwok,†,⊥ Marnie Higgins,† and Cynthia A. Afshari† ComparatiVe Biology and Safety Sciences, and Pharmacokinetics and Drug Metabolism, Amgen Inc., One Amgen Center DriVe, Thousand Oaks, California 91320 ReceiVed January 28, 2010

The key to the discovery of new pharmaceuticals is to develop molecules that interact with the intended target and minimize interaction with unintended molecular targets, therefore minimizing toxicity. This is aided by the use of various in vitro selectivity assays that are used to select agents most potent for the desired target. Typically, molecules from similar chemical series, with similar in vitro potencies, are expected to yield comparable in vivo pharmacological and toxicological profiles, predictive of target effects. However, in this study, we investigated the in vivo effects of two analogue compounds that similarly inhibit several receptor tyrosine kinases such as vascular endothelial growth factor receptor 1 (VEGFR/Flt1), vascular endothelial growth factor 2 (VEGFR2/kinase domain receptor/Flk-1), vascular endothelial growth factor receptor 3 (VEGFR3/Flt4), platelet-derived growth factor receptor (PDGFR), and Kit receptors, which bear similar chemical structures, have comparable potencies, but differ markedly in their rodent toxicity profiles. Global gene expression data were used to generate hypotheses regarding the existence of toxicity triggers that would reflect the perturbation of signaling in multiple organs such as the liver, adrenal glands, and the pancreas in response to compound treatment. We concluded that differences in pharmacokinetic properties of the two analogues, such as volume of distribution, half-life, and organ concentrations, resulted in marked differences in the chemical burden on target organs and may have contributed to the vast differences in toxicity profiles observed with the two otherwise similar molecules. We propose including select toxicokinetic parameters such as Vss, T 1/2, and T max as additional criteria that could be used to rank order compounds from the same pharmacological series to possibly minimize organ toxicity. Assessment of toxicokinetics is not an atypical activity on toxicology studies, even in early screening studies; however, these data may not always be used in decision making for selecting or eliminating one compound over another. Finally, we illustrate that in vivo gene expression profiles can serve as a complementary assessor of this activity and simultaneously help provide an assessment of on or off-target biological activity. Introduction The notion of body burden of contaminants is well recognized in the science of risk assessment and is used to study environmental exposures and their relationship to human dose (1). Biomathematical models permit the prediction of doses needed to target specific diseases in humans through the use of a variety of methods that include in vitro approaches as well as in vivo preclinical models (2–4). Here, we are reporting the investigation of two chemically related molecules that demonstrated similar enzymatic selectivity and potency profiles but yielded different toxicity profiles. Angiogenesis is the development of new vasculature from pre-existing blood vessels and/or circulating endothelial progenitor cells (5–7). Angiogenesis plays a critical role in many physiological processes, such as embryogenesis, wound healing, and menstruation, and in certain pathological events, such as solid tumor growth and metastasis. The recognition of vascular endothelial growth factor (VEGF) as a primary stimulus of angiogenesis in pathological * Corresponding author. Hisham K. Hamadeh. DABT, Amgen Inc., One Amgen Center Drive, Thousand Oaks, CA 91320. Phone: (805) 447-4818. E-mail: [email protected]. † Comparative Biology and Safety Sciences. ‡ Pharmacokinetics and Drug Metabolism. § Current address: Pfizer Inc., La Jolla, CA 92121. | Current address: Covance Inc., Thousand Oaks, CA 91320. ⊥ Current address: Allergan Inc., Irvine, CA 92612.

conditions has led to the exploration of many strategies to block VEGF activity (8). Inhibitory anti-VEGF receptor antibodies, soluble receptor constructs, antisense strategies, RNA aptamers against VEGF, and small molecule VEGF receptor tyrosine kinase inhibitors have all been developed to interfere with VEGF signaling (9). A number of target organ toxicities are observed in animals treated with tyrosine kinase inhibitors (10). In our study, compound-associated microscopic findings occurred in multiple organs including the liver, stomach, adrenal gland, pancreas, bone marrow, and teeth when Sprague-Dawley rats were exposed to molecules that concomitantly inhibit a variety of receptor tyrosine kinases including VEGFR1/Flt1, VEGFR2/ kinase domain receptor/Flk-1, VEGFR3/Flt4, platelet-derived growth factor receptor (PDGFR), and Kit receptors (11). Most of these effects can be attributed to the inhibition of one or more of the intended pharmacological targets (11). For example, incisors in rats grow continuously throughout their lifetime, and compromise of the blood supply to teeth leads to the impairment of normal growth and is considered to be an exaggerated pharmacological effect (12). We investigated whether common molecular triggers of toxicity were evident in multiple organs after the administration of multikinase inhibitors and whether mechanism(s) of toxicity could be uncovered.

10.1021/tx1000333  2010 American Chemical Society Published on Web 06/03/2010

1026

Chem. Res. Toxicol., Vol. 23, No. 6, 2010

Figure 1. Structures of the two analogues. Compounds were built with an azaindole ring system.

Fourteen-day rat studies were conducted with two structurally similar multikinase inhibitors, one associated with relatively minor toxicological findings in rats (AMG A) and the other exhibiting significantly more multiorgan toxicity (AMG B). The compounds displayed similar kinase-inhibition profiles and performed comparably in enzymatic assays. Global gene expression data were generated using whole genome DNA rat microarrays from the liver, adrenal, and pancreas tissues derived from the studies to aid in determining potential apparent off-target activity associated with AMG B exposure. It is hypothesized that gene expression measurements in organs can be divided into two main categories, one of which represents the direct perturbation of signaling that the compound induces and the other being the responses each organ uniquely mounts to deal with the perturbations triggered by the compound in question. It is how the molecular triggers from the first category of changes diverge within every organ that comprises the latter category of gene expression changes. It is relatively difficult to dissect the two categories when studying one organ, and as such, the use of multiple tissues is considered advantageous in this study since similarities across different organs in their response to the compound in question may highlight changes directly related to compound signaling rather than organ specific responses. Gleaning the toxicity triggers that are shared in most organs might allow the design of compounds without these attributes and thus minimizing the risk of toxicity in a variety of tissues. Using this approach, genes that were modulated in association with AMG B but not with AMG A were primarily indicative of perturbation of pathways related to pharmacological targets, including the VEGF pathway.

Experimental Procedures Compounds. The azaindole compounds AMG A and AMG B were synthesized at Amgen Inc. (Figure 1). The effective molecular weights of AMG A and AMG B are 414.491 and 429.375, respectively. The purity of both compounds was greater than 99.9%. Compound Administration and In-life Toxicology Evaluations. Fourteen-day studies in male Sprague-Dawley rats were conducted with AMG A and AMG B. AMG A was administered daily by oral gavage at doses of 3 or 30 mg/kg/day, and AMG B was administered daily by oral gavage at doses of 3 or 60 mg/kg/day. The vehicle used for each compound was OraPlus, pH 2.0 adjusted with methanesulfonic acid. The dose volume was 10 mL/kg. Doses for each compound were selected so that exposure (AUC0-24) would be comparable between the two compounds at both the low and high doses, on the basis of previous data. Four rats were euthanized at each of 3 time points (days 2, 5, and 15 after daily dosage; 24 h postdose) in each dose group for each compound. The study also included a group of vehicle dosed animals at each time point. Animals were fasted overnight (water was available). Blood (approximately 400 µL for hematology and 1.0 mL for clinical chemistry) was collected from the posterior vena cava at the time of terminal sacrifice. All surviving animals were

Hamadeh et al. food fasted overnight prior to the day of necropsy. Animals to be sacrificed were euthanized with carbon dioxide, the diaphragm punctured, bled for clinical pathology tests, exsanguinated, and necropsied. Histopathology was performed on select tissues. Toxicokinetics. Blood samples for toxicokinetic analysis were collected from animals in each dose group on study day 1 at 2, 4, 8, 12, and 24 h postdose. In addition, predose and 2-h postdose samples were collected from the 3-mg/kg/day dose group on study day 4 and from the 30- and 60-mg/kg/day dose groups on study day 14 for the determination of AMG A or AMG B levels in plasma. The plasma samples (50 µL) were extracted in 96-well format by protein precipitation using two times the sample volume of acetonitrile containing an internal standard. The sample mixtures were vortexed for 30 s and centrifuged at 1,600g for 5 min. One hundred microliters of supernatant was transferred to a fresh 96 well plate using Tomtec Quadra 3 SPE. An equal volume (100 µL) of mobile phase A was added to the supernatant, the solution was mixed, and the plates were loaded into the autosampler. Samples were separated by reversed-phase liquid chromatography on a Phenomenex Synergi Polar RP, 2.0 × 50 mm (4 µm) analytical column. The mobile phases were 0.1% formic acid in HPLC grade water (mobile phase A) and 0.1% formic acid in HPLC grade acetonitrile (mobile phase B). Isocratic liquid chromatography conditions at 70% B and a flow rate of 0.300 mL/min were used, with a total run time of 2.5 min. Plasma calibration curve standards were prepared at concentrations of 0.5, 1.0, 2.5, 5, 10, 25, 50, 100, 250, 500, 1000, 2000, and 5000 ng AMG A or B/mL Sprague-Dawley rat stock plasma. Solid tissue samples were homogenized in 4 times their weight of water. The plasma sample preparation methodology was then followed and concentrations reported as ng/g. Solid tissue calibration curve standards were prepared at concentrations of 2, 4, 10, 20, 40, 100, 200, 400, 1000, 2000, 4000, and 8000 ng AMG A or B/g in the stock tissue matrix. AMG A or AMG B concentrations were determined by LC-MS/ MS using electrospray ionization with multiple reaction monitoring in the positive ion mode. Peak areas were integrated using the Sciex Analyst, version 1.3.1 software (residing on a Windows 2000 computer). Following peak area integration, the data were exported to the software package Watson Non-GLP (version 7.0.0.0.1, InnaPhase Corp., Philadelphia, PA), where concentrations were determined by a weighted (1/x2) linear regression of peak area ratios (peak area of AMG A or AMG B/peak area of internal standard) versus the theoretical concentrations of the plasma calibration standards. AMG A or AMG B plasma concentration-time data for animals on day 1 were subjected to noncompartmental analysis. Concentrations that were below the limit of quantification (BQL) were set to zero. Nominal doses and sampling times were used in the toxicokinetic analysis. The Cmax and Tmax were taken directly from the concentration-time data. The area under the plasma concentration versus time curve from time 0 to 24 h (AUC0-24) was calculated using the linear/log trapezoidal method. Noncompartmental analyses of plasma AMG A or AMG B concentrations were performed using WinNonlin Professional (v.3.3 Pharsight Corporation, Mountain View, CA). Frozen Tissue Collection. Tissues (adrenal gland, pancreas, and liver) were harvested at necropsy and diced into 3 to 5 mm pieces. To protect sample integrity for later applications, samples were immediately snap-frozen in liquid nitrogen and stored at -70 °C until processed. RNA Preparation. Total RNA was isolated from 100 to 150 mg pieces of tissue according to the RNeasy extraction procedure (Qiagen, Valencia, CA). Tissues (adrenal, pancreas, and left lobe of liver) were homogenized in RLT lysis buffer containing β-mercaptoethanol using a hand-held rotor-stator homogenizer. Samples were applied to RNeasy midi (liver and adrenal) or maxi (pancreas) columns and processed according to the manufacturer’s instructions. An on-column DNase digestion was performed to remove any residual genomic DNA contamination. RNA concentration and yield were measured spectrophotometrically using Eppen-

Organ Chemical Burden and Toxicity

Chem. Res. Toxicol., Vol. 23, No. 6, 2010 1027

Table 1. Microscopic Findings for AMG B When Administered at Different Doses and Durations to Sprague-Dawley Ratsa necropsy dayb day 2

day 5

day 15

dose (mg/kg/day) organ

microscopic findingc

3c

adrenal

vacuolation (cortex) congestion/hemorrhage/infiltrates acinar cell atrophy (exocrine) acinar cell apoptosis mitotic figures necrosis vacuolation

1 (2.0)

pancreas liver

1 (1.0) 1 (1.0) 1 (1.0)

60c

1 (2.0)

60c

3c

1 (1.0)

1 (1.0)

1 (2.0)

4 3 4 2 2

1 (1.0) 1 (1.0)

2 (1.0) 1 (1.0)

a n ) 4/dose group/day. b Animals were necropsied 24 h after the last dose of AMG B. minimal, 2 ) mild, 3 ) moderate, and 4 ) marked.

dorf Biophotometer. Quality of the nucleic acid samples was evaluated with the RNA 6000 Nano chip kit (Agilent Technologies, Expert software). All samples were of good quality, as evidenced by distinct ribosomal 18S and 28S peaks and low baseline values. Gene Expression Data Generation. For microarray profiling, 5 µg of total RNA was reverse transcribed to double stranded cDNA, and biotinylated cRNA was generated using BioArray HighYield RNA Transcript Labeling Kit (Enzo Life Sciences, Farmingdale, NY). To increase hybridization efficiency, cRNA samples were fragmented prior to an overnight incubation with the rat RAE 230_2 array (Affymetrix, Santa Clara, CA). Gene expression ratios were calculated in Resolver by comparing intensity data from individual treated animals from each dose group with averaged intensities from control samples (n ) 4) at each time point. Data can be accessed from the Gene Expression Omnibus (GEO) database under submission series GSE10015. Computation Software and Statistical Analyses. Gene expression data were analyzed using Partek Pro software package v.6.05.0311, Rosetta Resolver v.5.0.0.3, and Ingenuity Pathways Analysis tool v.3.0. Partek Pro was used for analysis of variance (ANOVA) and list analysis. Ingenuity Pathways Analysis tool was used for pathway analysis.

3c

c

3 4 4 3 1

(1.7) (2) (2.75) (2.3) (1.0)

Incidence (average severity score). Severity scale: 1 )

Figure 2. Summary of averaged alanine aminotransferase (ALT) and aspartate aminotransferase (AST) values modulated in a test article related fashion in rats treated with AMG B (n ) 4). ALT and AST values were statistically significantly increased relative to controls at all time points only at 60 mg/kg/day (p < 0.05).

Table 2. Plasma AUC0-24 and Cmax Values Associated with AMG A and AMG B Measured after the First Dosea

Results In-life Toxicological Findings. No toxicologically relevant findings were observed in rats treated with AMG A up to 14 days of daily dosing. Exposure of rats to AMG B for 24 h at 3 or 60 mg/kg resulted in minimal to slight toxicologically significant findings in small numbers of treated rats. The incidence and severity of the findings for each dose group are summarized in Table 1. The findings included increases in hepatic transaminases (ALT and AST) (Figure 2), vacuolation of adrenal cortical cells, increased apoptosis and/or atrophy (degranulation) of pancreatic acinar cells, and acute inflammation of the pancreas. After 4 daily doses of AMG B, toxicologically significant lesions occurred in the 3- and 60-mg/kg dose groups (Table 1). Pancreatic and adrenal findings were similar to those seen 24 h after a single dose, but the changes occurred with greater frequency and were more pronounced after 4 doses in the 60mg/kg dose group. Hepatic effects were also noted and included mild increases in transaminases (Figure 2), slight to moderate increases in mitotic figures, slight vacuolar degeneration of hepatoctyes with or without apoptosis, and minimal necrosis. Liver, pancreas, and adrenal glands appeared to be target organs after 14 days of dosing with AMG B, and most of the findings were observed in the 60-mg/kg dose group (Table 1). Total bilirubin, albumin, and transaminases were mildly to moderately increased (Figure 2). These changes correlated histologically with minimal increases in hepatic mitotic figures in one animal. Clinical chemistry changes associated with the

(2.0) (2.0) (2.5) (1.0) (2.0)

60c

AMG A (non-toxic) AMG B (more toxic)

dose (mg/kg/day)

AUC0-24 (µg·h/mL)

Cmax (µg/mL)

3 30 3 60

2.53 ( 0.81 73.4 ( 25.9 3.62 ( 0.80 62.0 ( 13.90

0.819 ( 0.203 12.6 ( 2.8 0.322 ( 0.071 4.02 ( 0.51

a AUC0-24 AMG A was not statistically different from AUC AMG B when low or high doses were compared. Cmax AMG A was higher at both doses than AMG B (p < 0.05).

pancreas included mild increases in lipase and amylase; these parameters were markedly elevated in one rat. This animal also had the most severe histological lesions in many of the target organs. Minimal to moderate increases in acinar cell apoptosis and/or atrophy (degranulation) and arteritis were also observed in the pancreas. There were red foci in the stomach of one animal, but no histologic correlates were seen. In summary, daily administration of 3 and/or 60 mg/kg AMG B by oral gavage to rats for 14-days resulted in minimal pancreatic damage and clinical chemistry evidence of liver damage 24 h after a single dose. Incidence and severity of pancreatic and liver damage increased with increasing duration of exposure. By day 15, there was evidence that additional organs had sustained toxicity. Toxicokinetic Analysis. As shown in Table 2, administration of the high doses of AMG A (30 mg/kg/day) and AMG B (60 mg/kg/day) resulted in comparable exposure values (AUC0-24 and Cmax), despite the 2-fold difference in administered dose for the two compounds. Administration of the low dose (3 mg/ kg/day) of both AMG A and AMG B also produced similar

1028

Chem. Res. Toxicol., Vol. 23, No. 6, 2010

Hamadeh et al.

Table 3. Selectivity, KDR Target Potency, and Toxicokinetic Parameter Comparisons between AMG A and AMG B KDR, FLK1, VEGFR2 (IC50) FLT1, VEGFR1 (IC50) FLT4 (IC50) KIT (IC50) protein binding (fraction bound) T1/2 (rat) VSS (rat) Tmax (rat)

AMG B

AMG A

0.001 µM 0.000775 µM 0.000975 µM ∼0.001 µM >99% 5-7 h ∼3200 mL/kg ∼20,000 s

0.001 µM 0.000456 µM 0.001035 µM ∼0.001 µM >99% 0.8-1.3 h ∼400 mL/kg 5,000 s

exposure values. Additionally, for AMG A, there was a 15fold increase in Cmax and 29-fold increase in AUC0-24 observed from 3 to 30 mg/kg/day. For AMG B, there was a 12-fold increase in Cmax and 17-fold increase in AUC0-24 observed from 3 to 60 mg/kg/day. AMG A and AMG B had comparable activities on the pharmacological target (KDR) and other off-target kinases (Table 3) and were administered at doses resulting in similar plasma exposures in rats. Under these conditions, perturbation in transcript levels related to the intended pharmacological targets in multiple organs were observed exclusively in association with AMG B. This finding suggested that higher organ exposure and residency time for AMG B as compared to AMG A might account for the observed differences in toxicity and gene expression profiles. The hypothesis of differential organ exposure between the two compounds despite their comparable blood levels was further supported by their respective pharmacokinetic and physicochemical properties (Table 3). AMG B had a relatively longer half-life (T1/2), increased volume of distribution (Vss), Tmax, and log P values as compared with those of AMG A. These properties are typically associated with higher exposure level and longer residency times in the organs versus plasma. To confirm this hypothesis, we quantified concentrations of AMG A and AMG B in frozen liver tissue and compared these values to those observed in plasma samples (Figure 3). The results confirmed the retention of AMG B in liver samples at concentrations 5 to 10 times higher than those observed in blood. Low concentrations of AMG A were detected in both

Figure 3. Compound concentrations in plasma and frozen liver tissue. Higher concentrations of compound AMG B were found in liver and other tissues (data not shown) compared with those in plasma. Also, tissue concentrations of the compound associated with rat toxicity were much higher and had longer tissue residency time than AMG A, which was not associated with significant toxicity in rats. Individual bars represent individual animals. Adjacent bars correspond to serum or liver concentrations from the same animal as noted by the color. Serum levels of AMG B were statistically significantly different from those associated with AMG A (p < 0.05). Liver levels of AMG B were statistically significantly different from those associated with AMG A (p < 0.05). No statistically significant difference was observed for serum or liver levels for either compound.

Figure 4. Three-dimensional plots of toxicokinetic properties of a number of compounds from the same pharmacologic series with comparable efficacies in preclinical models. Panel A corresponds to values in dogs, while panel B indicates values in rats. Not all compounds had comparable toxicokinetic parameters in rats and dogs. AMG B was associated with a higher volume of distribution (Vss) and longer half-life in rats than in dogs and was relatively more toxic in rats than in dogs.

blood and liver 24 h after dose administration. Overall, these results suggest that gene expression measurement works well as a complementary assessor of compound exposure within a tissue. To further investigate whether the hypothesized relationship between increased chemical burden and observed toxicities could be extrapolated to other compounds, we collected Vss, T1/2, and Tmax data from several other internal multikinase inhibitors that share comparable target efficacy properties to AMG A and AMG B. Vss and T1/2 can offer an estimate of the duration of retention of compounds in various organs, while Tmax may be a more relevant descriptor of the residency time of the compound in gastrointestinal tissue, and thus, the latter may be an important consideration for agents associated with gastrointestinal toxicities. When evaluated concomitantly, these toxicokinetic properties represented a preliminary and simple surrogate of chemical burden in a biological system. Compounds with relatively higher values corresponding to these parameters were generally associated with increased toxicities in the gastrointestinal tract, liver, adrenals, and other organs (data not shown). We found these observations to be

Organ Chemical Burden and Toxicity

Chem. Res. Toxicol., Vol. 23, No. 6, 2010 1029

Table 4. Number of Transcripts Modulated by Each Treatment Regimen [p < 0.001 Rosetta Resolver Error Model] dose AMG A (nontoxic)

3 mg/kg/day

time

day day day 30 mg/kg/day day day day AMG B (more toxic) 3 mg/kg/day day day day 60 mg/kg/day day day day

1 4 14 1 4 14 1 4 14 1 4 14

adrenals liver pancreas NDa ND 1356 ND ND 1472 1229 1015 2405 2053 2123 5985

401 396 312 485 342 544 740 758 1140 1126 1237 3501

ND ND 1707 ND ND 2637 ND ND 2254 ND ND 5234

a ND: No gene expression data generated for the dose/time/tissue combination.

species-specific when evaluating rat and dog studies since compounds did not necessarily have similar toxicokinetic profiles pan-species. For each species, dots, representing individual compounds, that were proximal to each other, tended to result in similar toxicity profiles. For example, AMG B was associated with relatively higher tissue accumulation and increased toxicities than AMG A in rats, but in dogs, both compounds elicited similar pharmacokinetic profiles and tissue exposure which translated into decreased toxicity in dogs relative to that in rats (Figure 4A,B). These correlations held for most of the compounds that were investigated in this article with similar pharmacological effects but diverse chemical structures. Therefore, monitoring of select toxicokinetic parameters could offer a basis for investigating differences in toxicity outcomes in different species. Gene Expression Data Set. Ratio expression values for genes were exported from Rosetta Resolver, the database that housed the gene expression data, according to the following criterion: p < 0.0001 based on the default error model built into the software. If a gene was modulated by at least one treatment, it was included in the exported list of genes. This resulted in 10,599 sequences that were used for further analyses. Number of Gene Modulations. Table 4 shows the number of genes perturbed by each treatment regimen. Generally, more genes were modulated as dose and duration of exposure increased. In addition, relatively higher numbers of modulations were associated with AMG B in multiple organs and at various time points than were observed with AMG A. Toxicity Gene Profiles. We hypothesized that similar early molecular events, occurring in multiple organs, were related to the observed toxicities. Our focus in this article was not the sequelae of toxicity in different organs, but rather the early molecular profiles themselves, if they existed. Thus, our analysis focused on identifying genes, as singles or groups, that were modulated in a similar fashion in multiple organs in a doseand time-dependent manner and exclusively in association with AMG B (Figure 5). An ANOVA was performed to find genes that were differentially expressed in response to exposure to AMG B in a timeand dose-dependent fashion in the liver, pancreas, and adrenals. Genes were selected on the basis of p < 0.00001 for the model with a relatively conservative Bonferroni correction of p < 0.1. Thirty-five mRNAs met the statistical criteria (Table 5). Exposure to AMG A did not cause statistically significant modulation of any of the 35 identified mRNAs. The molecular process associated with each mRNA is also included in Table 5. These data were indicative of significant

Figure 5. Gene expression data analysis scheme highlighting the strategy for identifying molecular events associated with AMG B and using AMG A and as additional filter for common pharmacological effects.

Figure 6. Functional pathway analysis of top 35 sequences modulated in a dose- and time-dependent fashion with AMG B but not with AMG A. Analysis was conducted using Ingenuity Pathway Tool according to the manufacturer’s instructions. The significance values are a measure for how likely it is that genes from the data set under investigation participate in that function. The significance is expressed as a p-value, which is calculated using the right-tailed Fisher’s Exact Test. In this method, the p-value is calculated by comparing the number of input genes of interest that participate in a given function or pathway relative to the total number of occurrences of these genes in all functional/ pathway annotations stored in Ingenuity Pathways Knowledge Base. In the right-tailed Fisher’s Exact Test, only over-represented functional/ pathway annotations, annotations which have more Functions/Canonical Pathways Analysis Genes than expected by chance (right-tailed annotations), are used.

perturbation of elements relating to angiogenic processes. This finding was in agreement with the pathway analysis output derived from the Ingenuity pathway knowledgebase indicating a significant perturbation in VEGF and nitric oxide signaling pathways (Figure 6). Of notable mention is the observation of decreased transcript levels of VEGFR2, VEGFR1, and Neuropilin 1 (NRP1) in a dose- and time-dependent fashion in conjunction with exposure to AMG B but not AMG A (Figure 7). The pathway analysis was indicative of perturbations to axes related to the intended pharmacological targets.

Discussion The present study compared elicited gene expression effects in multiple organs, pharmacokinetic characteristics, and pharmacological and toxicological effects of a pair of compounds (AMG A and AMG B) in order to find explanations for the significant differences in the toxicity properties of the two

guanine nucleotide binding protein gamma subunit 11 complement component 1, q subcomponent, alpha polypeptide complement component 1, q subcomponent, C complement component 1, q subcomponent, beta polypeptide transmembrane receptor FcgammaRIII-X Toll interacting protein (predicted) nuclear undecaprenyl pyrophosphate synthase 1 homologue NAD(P)H dehydrogenase, quinone 1 dysferlin (predicted) neural precursor cell expressed, developmentally down-regulated gene 4A cathepsin L similar to CDV-3B

1367902_at

a

Averaged log10 ratio expression.

response to oxidative stress response to stress translational initiation transmembrane receptor protein tyrosine kinase signaling

proteolysis regulation of transcription, DNA-dependent response to chemical stimulus

nitric oxide biosynthesis plasma membrane repair protein modification

immune response inflammatory response metabolism

immune response

immune response

angiogenesis angiogenesis angiogenesis angiogenesis, cell adhesion apoptosis cell adhesion cell adhesion ER to Golgi vesicle-mediated transport G-protein coupled receptor protein signaling immune response

actin filament bundle formation angiogenesis angiogenesis

GO biological process

liver

0.29 -0.21 0.09

-0.05 -0.05 -0.03

0.10 -0.17 0.09 -0.12 -0.35 -0.14 -0.29 -0.21 -0.14 -0.24 -0.11

0.03 -0.02 -0.02 -0.06 -0.10 -0.02 -0.11 -0.12 0.12 -0.08 0.00

-0.19

-0.18 0.06 0.02

-0.05 0.00 -0.01

-0.05

-0.27

-0.06

0.06 0.03

-0.26

-0.09

0.02 0.02

-0.22

-0.09

-0.02 0.04 -0.19 0.08

-0.36 -0.22 -0.41 -0.14

-0.30 -0.23 -0.25

0.03 -0.28 0.10 -0.28

-0.04 -0.06 0.02 -0.09 -0.14 -0.12 0.02

-0.23

0.09 0.09

0.30 -0.13 0.13

-0.22 0.07 0.04

-0.44

-0.38

-0.35

-0.28

-0.09

0.00 -0.05

-0.11 -0.06 0.05

0.00 -0.01 -0.04

-0.12

-0.03

-0.07

-0.04

-0.09 0.04 -0.09 0.04

-0.03 -0.05 -0.08

0.02 -0.09 0.03 0.00

-0.56 -0.31 -0.23 -0.22

-0.27 -0.20 -0.54

0.17 -0.52 0.15 -0.61

-0.29

0.18 0.24

-0.06 0.03 0.03

0.43 -0.37 0.20

-0.54 0.08 0.13

-0.71

-0.62

-0.62

-0.34

-0.32 -0.22 -0.25 -0.39 0.09 -0.46 -0.28 0.11

-0.02 -0.16 0.09

-0.07 -0.06 -0.03

-0.10

-0.05

-0.11

-0.10

-0.08 -0.03 -0.01 -0.04 0.01 -0.05 -0.02 0.02

-0.32 -0.09 -0.15 -0.08

-0.16 -0.20 -0.15

0.08 -0.13 0.06 -0.09

-0.16

-0.01 0.09

0.00 -0.21 0.07

-0.06 0.11 0.05

-0.30

-0.29

-0.23

-0.12

-0.19 -0.08 -0.12 -0.10 0.04 -0.12 -0.17 0.10

-0.63 -0.17 -0.25 -0.10

-0.31 -0.31 -0.39

0.10 -0.30 0.09 -0.24

-0.43

0.04 0.18

-0.08 -0.40 0.10

-0.08 0.14 0.09

-0.52

-0.76

-0.54

-0.24

-0.27 -0.08 -0.19 -0.24 0.04 -0.14 -0.26 0.11

-0.15

-0.24 -0.16 -0.22 -0.23 0.03 -0.42 -0.16 0.05

-0.07 -0.11 -0.11

-0.03

-0.08 -0.07 -0.13 -0.13 -0.03 -0.12 -0.04 -0.01

-0.37 -0.15 -0.36

-0.14 -0.12 -0.12 -0.20 0.06 -0.25 -0.18 0.07

-0.05 -0.01 -0.03

-0.05 0.05 -0.03 -0.02 0.00 -0.06 -0.04 0.00

-0.15 -0.12 -0.31

-0.27 -0.18 -0.16

-0.02 -0.06 -0.12

-0.11 -0.09 -0.13

-0.04 -0.01 -0.01

day 1 60 mg/kg

day 14

3 60 3 60 3 60 3 mg/kg mg/kg mg/kg mg/kg mg/kg mg/kg mg/kg

day 7

day 14

day 14

pancreas

-0.10 -0.09 -0.20 0.00

-0.19 -0.15 0.09

-0.04 -0.07 0.00 -0.08

-0.11

-0.07 -0.04

-0.08 -0.19 0.00

0.10 -0.04 -0.12

0.03

-0.03

-0.01

-0.12

-0.17 -0.02 -0.17 -0.13 0.00 -0.15 -0.15 -0.01

0.00 -0.06 -0.25

-0.93 -0.17 -0.35 -0.08

-0.33 -0.45 -0.43

0.06 -0.35 0.09 -0.25

-0.46

0.01 0.11

-0.05 -0.53 0.08

-0.09 0.08 0.02

-0.45

-0.72

-0.71

-0.26

-0.34 -0.19 -0.28 -0.38 0.06 -0.32 -0.35 0.08

-0.36 -0.16 -0.44

-0.23 -0.06 -0.06 0.00

-0.14 -0.13 -0.18

-0.01 -0.05 0.03 -0.12

-0.06

-0.01 0.02

-0.16 -0.08 0.11

-0.09 -0.03 -0.04

-0.15

-0.19

-0.21

-0.12

-0.04 0.03 -0.10 -0.09 -0.03 -0.09 -0.08 0.02

-0.05 -0.11 -0.10

-0.85 -0.37 -0.13 -0.13

-0.47 -0.49 -1.09

0.11 -0.33 0.13 -0.46

-0.42

0.15 0.17

0.05 -0.58 0.16

-0.34 0.16 0.17

-0.65

-0.75

-0.72

-0.24

-0.37 -0.13 -0.17 -0.44 0.05 -0.66 -0.39 0.16

-0.42 -0.39 -0.55

0.05 0.09 0.00 0.03

-0.18 0.10 0.07

-0.07 0.04 0.03 0.04

-0.03

-0.04 0.00

-0.09 -0.03 -0.10

0.01 -0.09 -0.09

0.48

0.13

0.13

-0.01

0.05 -0.03 -0.08 -0.06 0.04 -0.02 0.18 -0.08

0.17 -0.01 0.08

-0.33 -0.11 -0.71 -0.18

-0.42 -0.14 -0.37

0.25 -0.28 0.47 -0.48

-0.25

0.18 0.18

0.37 -0.37 0.46

-0.16 0.11 0.11

-0.54

-0.54

-0.68

-0.29

-0.21 -0.09 -0.47 -0.29 0.20 -0.37 -0.19 0.10

-0.18 -0.17 -0.69

3 60 3 60 3 60 mg/kg mg/kg mg/kg mg/kg mg/kg mg/kg

day 7

adrenal

Chem. Res. Toxicol., Vol. 23, No. 6, 2010

1389553_at 1382078_at 1393235_at 1389541_at

1384405_at 1376623_at 1373785_at

1386958_at 1373818_at 1397674_at 1388784_at

1382601_at

1370245_at 1383028_at

1387599_a_a 1392786_at 1395157_at

1373523_at 1380808_at 1392522_at

1370215_at

1373025_at

1376652_at

similar to scavenger receptor type A SR-A (predicted) thioredoxin reductase 1 heat shock 90kD protein 1 eukaryotic translation initiation factor 3 colony stimulating factor 1 receptor precursor EST claudin-like protein 24 similar to phospholipase D family, member 4 dendritic cell inhibitory receptor 3, Dcir3 EST EST malignant fibrous histiocytoma amplified sequence 1 (predicted)

lymphocyte cytosolic protein 1 endothelial-specific receptor tyrosine kinase kinase insert domain protein receptor (KDR), VEGFR2 neuropilin 1 Angiomotin-like 1 (predicted) fms-related tyrosine kinase 1, VEGFR1 endomucin bifunctional apoptosis regulator stabilin 1, STAB1 intercellular adhesion molecule 2, Icam2 GTPase Rab6

1389210_at 1393067_at 1367948_a_a

1370570_at 1371627_at 1383019_at 1372587_at 1388733_at 1374247_at 1389235_at 1371103_at

gene name

probe set ID

day 1

Table 5. Genes That Were Modulated in a Statistically Significant Manner in a Dose- and Time-Dependent Fashion in Response to AMG B Treatment but Not Modulated with AMG A (Each Relative to Vehicle Treatment)a

1030 Hamadeh et al.

Organ Chemical Burden and Toxicity

Chem. Res. Toxicol., Vol. 23, No. 6, 2010 1031

Figure 7. Averaged log10 ratio expression of transcript corresponding to VEGFR2 (KDR) in test article-treated rats relative to time-matched controls. Statistically significant, dose- and time-dependent decrease in KDR expression was observed in tissues from rats exposed to AMG B, while no statistically significant perturbation was observed in tissues derived from rats treated with AMG A. Error bars are standard deviations (n ) 4).

multikinase inhibitors. The use of analogues with significantly different toxicity profiles is one approach to studying on- versus off-target toxicological effects of small molecules. This approach aims at neutralizing effects attributed to differences in pharmacological or known off-target properties of compounds under study. In addition, this approach is invaluable for deriving gene expression profiles relevant to toxicological processes. Despite achieving comparable blood exposures with two compounds that have similar pharmacological potencies and chemical structure, vast differences in toxicological findings were noted. In this study, the intent was not to, exhaustively, understand the toxicity in each organ but rather to use a multitude of organs to explain commonalities and convergences in the gene expression data that may explain universal triggering events associated with the compound treatments. The perturbation of the on-target kinases after statistical cutoffs only by the toxic compound and not by the nontoxic analogue was deemed important since these were the intended targets of the compounds being studied. Since little else was observed in terms of gene expression changes common to all the organs, it suggested to us an on-target effect mediated by the tissue burden associated with the toxic compound. Thus, differences in gene expression elicited in multiple organs in rat were consistent with the differences in toxicity. Two example genes that were differentially modulated are VEGF and NRP1. VEGF is a mitogen that specifically acts on endothelial cells. Its expression is increased by hypoxia, and its cell-surface receptor, known as Flk1 in mouse, is exclusively expressed in endothelial cells (13, 14). Flk1 is the mouse homologue of the kinase insert domain receptor (KDR) (15). KDR and its mouse homologue bind vascular endothelial growth factor with high affinity and have been implicated in the development of blood and blood vessels. NRP1 is a membrane-bound coreceptor to a tyrosine kinase receptor for both VEGF family members. NRP1 plays versatile roles in angiogenesis, axon guidance, cell survival, migration, and invasion. Soker et al. (16) confirmed that NRP1 binds VEGF165. They showed that when coexpressed in cells with KDR (VEGFR2), NRP1 enhances the binding of VEGF165 to KDR and VEGF165-mediated chemotaxis. Conversely, inhibition of VEGF165 binding to NRP1 inhibits its binding to KDR and its mitogenic activity for endothelial cells. Soker et al. (16) proposed that NRP1 is a novel VEGF receptor that modulates

VEGF binding to KDR and subsequent bioactivity and therefore may regulate VEGF-induced angiogenesis. Other genes demonstrating dose- and time dependency in response to AMG B without accompanying modulation with AMG A were related to cell adhesion. Intercellular adhesion molecule 2 (Icam2), which was decreased, binds to the leukocyte adhesion protein, lymphocyte function-associated antigen-1 (LFA-1). This protein may play a role in lymphocyte recirculation by blocking LFA-1-dependent cell adhesion. It mediates adhesive interactions important for antigen-specific immune response, NK-cell mediated clearance, lymphocyte recirculation, and other cellular interactions important for immune response and surveillance. Complement component 1q (C1q), which was decreased in our study in response to AMG B, is known to interact with various types of vascular endothelial cells leading to the production of biologically active proteins or expression of adhesive molecules, thus playing a major role in the induction of a diversity of functions, including adhesion and spreading (17). C1q, for example, has been reported to trigger the production of IL-8, IL-6, and monocyte chemo-attractant peptide-1 by human umbilical vein endothelial cells (HUVECs) (18). Similarly, C1q has been shown to enhance immune complex-mediated expression of adhesive proteins such as E-selectin, ICAM-1, and vascular cell adhesion molecule-1 (VCAM-1) by HUVECs (19). The mechanisms underlying the decrease in transcript levels of VEGFR2 and other intended pharmacological targets are not understood. Assuming that the decrease is directly related to inhibition of the target kinases, this measure might serve as an indicator of test article abundance and biochemical coverage or activity in the assayed organs. In addition to regulation by protein tyrosine phosphatases, receptor tyrosine kinases have evolved various autoinhibitory mechanisms to suppress basallevel activity. Although the modulation of catalytic activity through reversible phosphorylation/dephosphorylation of the kinase activation segment has been recognized for some time, recent studies have shown that the juxtamembrane region (between the transmembrane helix and the cytoplasmic kinase domain) of several receptor tyrosine kinase (RTK) subclasses has a prominent role in kinase activation (20). In crystal structures of the vascular endothelial growth factor receptor-2 (KDR) (21) and FLT3 (22), a Tyr residue or another hydro-

1032

Chem. Res. Toxicol., Vol. 23, No. 6, 2010

phobic residue in the kinase-proximal juxtamembrane region occupies the RC-β sheet cleft in a manner that is similar to Tyr984 in the insulin receptor, where the juxtamembrane region has been shown to negatively regulate its catalytic activity. Therefore, this autoinhibition mechanism is probably operative in KDR and other tyrosine kinases targeted by our molecules. The fact that expression of VEGFR2 and other related genes were perturbed by AMG B but not by AMG A was suggestive of an exaggerated on-target effect in the microenvironment of several organs. Higher compound concentrations at target organs of AMG B could have accounted for this observation. Taken together, the longer half-life and higher Vss associated with AMG B may have contributed to the higher accumulation of AMG B in various organs compared to that in blood and thus its associated toxicities. Therefore, the chemical burden in multiple organs was higher in association with exposure to AMG B compared with that to AMG A was evidenced by gene expression analyses, toxicokinetic properties, and direct measurement of compound concentrations in the target organs. Strong evidence exists that inter- and intrapatient pharmacokinetic variability of chemotherapy drugs is a major contributor to toxicity and treatment efficacy. This variability can be as high as 50-fold in plasma concentrations even when body surface area-based doses are administered (23–25). This difference in pharmacokinetics may be the result of many factors including genetic composition, physiological status, and external influences. Pharmacokinetic measurement reflects a patient’s absorption, distribution, metabolism, and elimination of an agent. Patients with lower clearance rates may have higher plasma concentrations, which can result in toxicity. Integration of these pharmacokinetic considerations into clinical practice is demonstrated in a study of cancer patients (26). Dose-limiting toxicity affects 31% to 34% of patients treated with 5-fluorouracil (5-FU). Bocci et al. (26) examined the pharmacokinetics of 5-FU and its catabolite, 5-fluorodihydrouracil (5-FDHU), in 188 patients, showing that the mean total body clearance rate for 5-FU was 65.67 L/h/m2, and Tmax for 5-FDHU was 26.63 min. The mean half-life of 5-FDHU was longer in patients who experienced grade 2-4 toxicity from 5-FU treatment than in patients with toxicity of grade 0-1, and Tmax > 30 min was associated with gastrointestinal and hematologic toxicity. The authors conclude that analysis of 5-FU and 5-FDHU pharmacokinetics before treatment would be helpful in identifying patients at risk from 5-FU therapy. Similar observations have been made when studying illudins, which are novel low molecular weight natural products that are cytotoxic to human tumor cells in vitro. The fact that some illudin analogues are more efficacious in vitro and in vivo than other analogues was explained by an elegant series of experiments using a radiolabeled drug (27). Cell lines sensitive at nanomolar concentrations displayed a saturable, energy-dependent accumulation of illudins with relatively low Km and high Vmax values, whereas a nonsensitive cell line, requiring millimolar concentrations to achieve in vitro toxicity, showed minimal illudin uptake with higher Km and lower Vmax values. Also, in one of the earliest experiments carried out to investigate the mechanisms of paraquat toxicity in rats, it was found that after oral administration, the plasma concentration of paraquat remained relatively constant over a period of 30 h, whereas the concentration in the lung rose progressively to several times that in the plasma (28). The lack of time-dependent accumulation in any other organ studied explained, in part, the selective toxicity of paraquat to the lung since it was this organ

Hamadeh et al.

that achieved the highest concentrations of paraquat after oral dosing (29), thus linking local chemical burden to toxicity. We propose the concomitant monitoring of parameters such as Vss, T1/2, and Tmax, as surrogates for chemical burden, via three-dimensional plots as shown in Figure 4 or more elaborate weighted computational models, as an approach that may be useful in prioritizing compounds based on their potential to yield relatively high tissue levels, potentially leading to undesired effects. The multivariate approach stems from the occasional limitation of univariate analysis. For example, AMG A is the nontoxic compound in this study, and care was taken to achieve dose equivalence based on the AUC values since that measure was thought to better represent exposure over a longer period of time. To do so, a higher dose of A was used and thus resulted in a higher Cmax. Despite the higher Cmax, compound A was not associated with significant toxicity of gene expression events suggesting Cmax as a singleton variable may not be a significant contributor to the observed toxicities or molecular changes. The recommendations based on data in this article are to aim for the lowest values of Vss, T1/2, and Tmax that still maintain the efficacy success thresholds. We hypothesize that sustained presence of kinase inhibitors in various organs might be associated with toxicological findings. This strategy, when deployed within a drug development program, requires careful review of compound attributes such as efficacy in preclinical models, known off-target hits, physicochemical properties, covalent binding due to metabolic activation, and transporter perturbations, to name a few, as these might be confounding factors when associating toxicokinetic properties with the development of toxicities. In conclusion, this article highlights a strategy to identify potential triggers that may underly the toxicity observed in association with exposure to xenobiotics. This strategy incorporates appropriate negative controls to enable the dissection of gene expression data and highlight transcripts relevant to toxicological processes of interest. This study uniquely demonstrates the ability to generate hypotheses using high density gene expression data via the study of the convergence of gene expression data from multiple organs, thus approximating a systems approach that focuses on the potential triggers of toxicity rather than the unique sequelae in individual organs. The study also highlighted the ability to follow up on potential leads using more classical techniques. For the evaluation of small molecule kinase inhibitors, our results highlight the importance of local tissue concentration and residency time of small molecule kinase inhibitors as additional parameters to consider when interpreting toxicological findings since our present study demonstrated the potential link between chemical burden and manifestation of toxicity in an in vivo system. As the relationship between these parameters and toxicity is better understood, the strategy we describe here can be used as a preliminary visualization tool to discriminate compounds from the same pharmacological class. Additional examples will be required to investigate the general applicability of modeling traditional pharmacokinetic parameters, alongside tissue drug concentrations (or surrogate gene expression profiles), toward decision making regarding discriminating and selecting potential pharmaceutics during the lead optimization phase of drug development. Acknowledgment. We thank Drs. Robert T. Dunn, Anthony Polverino, and Paul Hughes, and Gary Skiles for contributing to the valuable scientific discussions relevant to this article.

Organ Chemical Burden and Toxicity

Chem. Res. Toxicol., Vol. 23, No. 6, 2010 1033

References (1) Fiore, B. J., Anderson, H. A., Hanrahan, L. P., Olson, L. J., and Sonzogni, W. C. (1989) Sport fish consumption and body burden levels of chlorinated hydrocarbons: a study of Wisconsin anglers. Arch. EnViron. Health 44 (2), 82–88. (2) Hickey, J. L., and Reist, P. C. (1977) Application of occupational exposure limits to unusual work schedules. Am. Ind. Hyg. Assoc. J. 38 (11), 613–621. (3) Bertail, P., Cle´menc¸on, S., and Tressou, J. (2008) A storage model with random release rate for modeling exposure to food contaminants. Math Biosci. Eng. 5 (1), 35–60. (4) Tressou, J., Cre´pet, A., Bertail, P., Feinberg, M. H., and Leblanc, J. Ch. (2004) Probabilistic exposure assessment to food chemicals based on extreme value theory. Application to heavy metals from fish and sea products. Food Chem. Toxicol. 42 (8), 1349–1358. (5) O’Reilly, M. S. (1997) The preclinical evaluation of angiogenesis inhibitors. InVest. New Drugs 15 (1), 5–13. (6) Parangi, S., O’Reilly, M., Christofori, G., Holmgren, L., Grosfeld, J., Folkman, J., and Hanahan, D. (1996) Antiangiogenic therapy of transgenic mice impairs de novo tumor growth. Proc Natl Acad Sci U S A. 5 93 (5), 2002–7. (7) Folkman, J. (1972) . Anti-angiogenesis: new concept for therapy of solid tumors. Ann. Surg. 175 (3), 409–416. (8) Folkman, J. (1974) Tumor angiogensis: role in regulation of tumor growth. Symp. Soc. DeV. Biol. 30 (0), 43–52. (9) Siemeister, G., Martiny-Baron, G., and Marme, D. (1998) The pivotal role of VEGF in tumor angiogenesis: molecular facts and therapeutic opportunities. Cancer Metastasis ReV. 17 (2), 241–248. (10) Patyna, S., Arrigoni, C., Terron, A., Kim, T. W., Heward, J. K., Vonderfecht, S. L., Denligner, R., Turnquist, S. E., and Evering, W. (2008) Nonclinical safety evaluation of sunitinib: a potent inhibitor of VEGF, PDGF, KIT, FLT3, and RET receptors. Toxicol. Pathol 36 (7), 905–916. (11) Hall, A. P., Westwood, F. R., and Wadsworth, P. F. (2006) Review of the effects of anti-angiogenic compounds on the epiphyseal growth plate. Toxicol. Pathol. 34 (2), 131–147. (12) Wise, G. E., and Yao, S. (2003) Expression of vascular endothelial growth factor in the dental follicle. Crit. ReV. Eukaryotic Gene Expression 13 (2-4), 173–180. (13) Millauer, B., Wizigmann-Voos, S., Schnurch, H., Martinez, R., Moller, N. P., Risau, W., and Ullrich, A. (1993) High affinity VEGF binding and developmental expression suggest Flk-1 as a major regulator of vasculogenesis and angiogenesis. Cell 72 (6), 835–846. (14) Plate, K. H., Breier, G., Millauer, B., Ullrich, A., and Risau, W. (1993) Up-regulation of vascular endothelial growth factor and its cognate receptors in a rat glioma model of tumor angiogenesis. Cancer Res. 53 (23), 5822–5827. (15) Matthews, W., Jordan, C. T., Gavin, M., Jenkins, N. A., Copeland, N. G., and Lemischka, I. R. (1991) A receptor tyrosine kinase cDNA isolated from a population of enriched primitive hematopoietic cells and exhibiting close genetic linkage to c-kit. Proc. Natl. Acad. Sci. U.S.A. 88 (20), 9026–9030. (16) Soker, S., Takashima, S., Miao, H. Q., Neufeld, G., and Klagsbrun, M. (1998) Neuropilin-1 is expressed by endothelial and tumor cells

(17)

(18)

(19)

(20) (21)

(22) (23) (24)

(25) (26)

(27)

(28) (29)

as an isoform-specific receptor for vascular endothelial growth factor. Cell 92 (6), 735–745. Peerschke, E. I., Smyth, S. S., Teng, E. I., Dalzell, M., and Ghebrehiwet, B. (1996) Human umbilical vein endothelial cells possess binding sites for the globular domain of C1q. J. Immunol. 157 (9), 4154–4158. van den Berg, R. H., Faber-Krol, M. C., Sim, R. B., and Daha, M. R. (1998) The first subcomponent of complement, C1q, triggers the production of IL-8, IL-6, and monocyte chemoattractant peptide-1 by human umbilical vein endothelial cells. J. Immunol. 161 (12), 6924– 6930. Lozada, C., Levin, R. I., Huie, M., Hirschhorn, R., Naime, D., Whitlow, M., Recht, P. A., Golden, B., and Cronstein, B. N. (1995) Identification of C1q as the heat-labile serum cofactor required for immune complexes to stimulate endothelial expression of the adhesion molecules E-selectin and intercellular and vascular cell adhesion molecules 1. Proc. Natl. Acad. Sci. U.S.A. 92 (18), 8378–8382. Hubbard, S. R. (2004) Juxtamembrane autoinhibition in receptor tyrosine kinases. Nat. ReV. Mol. Cell Biol. 5 (6), 464–471. McTigue, M. A., Wickersham, J. A., Pinko, C., Showalter, R. E., Parast, C. V., Tempczyk-Russell, A., Gehring, M. R., Mroczkowski, B., Kan, C. C., Villafranca, J. E., and Appelt, K. (1999) Crystal structure of the kinase domain of human vascular endothelial growth factor receptor 2: a key enzyme in angiogenesis. Structure 7 (3), 319– 330. Griffith, J., Black, J., Faerman, C., Swenson, L., Wynn, M., Lu, F., Lippke, J., and Saxena, K. (2004) The structural basis for autoinhibition of FLT3 by the juxtamembrane domain. Mol. Cell 13 (2), 169–178. Krynetski, E. Y., and Evans, W. E. (1998) Pharmacogenetics of cancer therapy: getting personal. Am. J. Hum. Genet. 63 (1), 11–16. McDonald, G. B., Slattery, J. T., Bouvier, M. E., Ren, S., Batchelder, A. L., Kalhorn, T. F., Schoch, H. G., Anasetti, C., and Gooley, T. (2003) Cyclophosphamide metabolism, liver toxicity, and mortality following hematopoietic stem cell transplantation. Blood 101 (5), 2043–2048. Partridge, A. H., Avorn, J., Wang, P. S., and Winer, E. P. (2002) Adherence to therapy with oral antineoplastic agents. J. Natl. Cancer Inst. 94 (9), 652–661. Bocci, G., Barbara, C., Vannozzi, F., Di, P. A., Melosi, A., Barsanti, G., Allegrini, G., Falcone, A., Del, T. M., and Danesi, R. (2006) A pharmacokinetic-based test to prevent severe 5-fluorouracil toxicity. Clin. Pharmacol. Ther. 80 (4), 384–395. Kelner, M. J., McMorris, T. C., Montoya, M. A., Estes, L., Rutherford, M., Samson, K. M., and Taetle, R. (1997) Characterization of cellular accumulation and toxicity of illudin S in sensitive and nonsensitive tumor cells. Cancer Chemother. Pharmacol. 40 (1), 65–71. Smith, L. L., Wright, A., Wyatt, I., and Rose, M. S. (1974) Effective treatment for paraquat poisoning in rats and its relevance to treatment of paraquat poisoning in man. Br. Med. J. 4 (5944), 569–571. Rose, M. S., Lock, E. A., Smith, L. L., and Wyatt, I. (1976) Paraquat accumulation: tissue and species specificity. Biochem. Pharmacol. 25 (4), 419–423.

TX1000333