Acute Drug-Induced Vascular Injury in Beagle Dogs: Pathology and Correlating Genomic Expression

Abstract

Acute vascular injury that leads to vascular inflammation is a common finding in the preclinical toxicity testing of drugs in rats and dogs. However, the relevance of this finding for risk to humans is unclear. Concern about the safety of these drugs is heightened by the current lack of noninvasive clinical methods to predict the onset of vascular damage in animals or humans. Determining the relevance of this poorly understood preclinical outcome for humans requires a better understanding of the molecular mechanisms of injury in addition to the development of sensitive and specific leading biomarkers for the clinical diagnosis of acute vascular damage.

Most molecular research on this toxicity has been performed in rats, but recent development of canine gene expression microarrays makes transcriptomic studies now possible in the dog. In this study, we investigated the molecular mechanisms of drug-induced vascular injury in dogs using gene arrays. After treating Beagles with toxic doses of CI-947, an adenosine receptor agonist, we profiled gene expression in the coronary arteries and correlated those changes with histopathology at 16 and 24 hours after dosing. The results demonstrated that pathobiological processes such as stimulation of the innate immune response, increased extracellular matrix turnover and oxidative stress were active at times of very early injury.

Introduction

Acute vascular injury that leads to vascular inflammation has long been a common finding in the preclinical toxicity testing of drugs in dogs and rats (Boor et al., 1995; Kerns et al., 2005). However, the relevance of this finding for risk to humans is unclear. Concern about the safety of these drugs is heightened by the current lack of noninvasive clinical methods to accurately predict the onset of vascular damage in animals or humans. Determining the relevance of this poorly understood preclinical outcome for humans requires a better understanding of the associated molecular mechanisms of injury in addition to the development of sensitive and specific leading biomarkers for the clinical diagnosis of acute vascular damage.

Adenosine and many adenosine receptor agonists are potent vasodilators with demonstrated ability to significantly increase coronary arterial blood flow (Stepp et al., 1996). Adenosine receptor agonists also have been reported to cause coronary vascular injury in dogs (Macallum et al., 1991). CI-947 is a nonselective adenosine receptor agonist that has produced acute coronary arteriopathy detectable at 24 and 48 hours after oral administration of a single dose of 5 or 10 mg/kg (Metz et al., 1991). The histopathological changes at 24 hours were characterized by accumulation of proteinaceous material and erythrocytes in pockets within the tunica media. The early medial changes were often accompanied by minimal adventitial inflammatory cell infiltrations. At 48 hours, the lesions progressed to include transmural necrosis and more extensive and intense inflammatory cell infiltrations. When examined by transmission electron microscopy, the coronary arteries exhibited ultrastructural alterations that included endothelial retraction, subendothelial accumulation of fibrin and platelets, necrosis of smooth muscle cells, and mural infiltration of granulocytes and monocytes.

In this study, we built upon the previous investigations conducted with CI-947 to further examine the pathogenesis of vascular injury caused by this compound. We dosed dogs once orally with 2 or 10 mg/kg and evaluated the histopathologic changes in intramural and extramural coronary arteries at 16 and 24 hours postdosing. Using microarrays, we measured correlating changes in global gene expression of isolated left extramural coronary arteries from the same animals. The changes in transcriptomic patterns that correlated with lesion development revealed pathobiological processes active at times of early vascular injury, providing insight into the molecular mechanisms responsible for development of that injury.

Methods and Materials

Animals, Dosing, In-Life Evaluation

Twenty-four purebred male beagle dogs, 8 to 12 months old, were obtained from Marshall Farms (North Rose, NY). Males are more sensitive to CI-947 cardiovascular toxicity than females. The dogs were housed individually in conditions controlled for light cycle (12 hour light/dark), temperature (70 ± 5°F), and relative humidity (45 ± 5%). A diet of dry food was fed once daily (Certified Purina Diet 5L66) and water was provided ad libitum. All animal care and experimentation protocols were approved and carried out at Pfizer Laboratories under standard operating procedures and in accordance with the Guide for the Care and Use of Laboratory Animals (National Research Council, 1996). Pfizer Laboratories are AAALAC-accredited facilities.

A single dose of CI-947 was administered orally to each dog in gelatin capsules at 2 or 10 mg/kg. Control dogs were dosed with gelatin capsules only. Dogs in each dose group were scheduled for necropsy at either 16 or 24 hours post dosing, and each dose/time point group contained 4 animals. A PONEMAH P3 Digital Acquisition System (LDS, Charlotte, NC) was used to collect body temperature, heart rate (HR; beats per minute) and indirect systolic blood pressure (BP; mm Hg) data from conscious dogs. These measurements were obtained from dogs pre-dosing and ~4 hours post dosing, the time of maximum drug exposure (T_max).

Two sets of blood samples (collected in tubes with EDTA, citrate, and no additive) were taken from the jugular vein of each dog, one set immediately prior to dosing and the other immediately prior to necropsy. These samples were processed and evaluated routinely for hematology (Advia 120, Bayer Diagnostics, Tarrytown, NY), coagulation parameters (STA-Compact, Roche Diagnostics, Branchburg, NJ), and serum chemistry (Hitachi 917, Roche Diagnostics).

Cardiovascular (HR, BP) and clinical pathology parameters were analyzed by comparing pre-dose and post-dose values for individual animals, and by comparing treated group means to control means using t-tests after demonstrating that variance was homogeneous (Rohlf and Sokal, 1994).

Pathology

The dogs were anesthetized by intravenous injection of pentobarbital sodium and exsanguinated. After opening the thorax, the heart was immediately removed and rinsed in sterile saline. The right coronary artery and the circumflex and paraconal interventricular branches of the left coronary artery (Evans and Christenson, 1979) were carefully dissected away from adjacent myocardium and epicardial adipose tissue using microdissection scissors. Each artery was divided into 3 sections: one snap-frozen in liquid nitrogen, one embedded and frozen in OCT (Fisher Scientific), and one fixed in 10% neutral buffered formalin. The heart was opened and examined, then fixed in formalin with samples from other selected organs (lung, liver, kidney, testes, epididymides, urinary bladder, arteries of cranial mediastinum).

Fixed tissues were processed routinely into paraffin-embedded blocks, sectioned at 4 μm, placed on glass slides and stained with hematoxylin and eosin or Movat’s pentachrome. All slides were examined by light microscopy using an Olympus BX51 microscope. All affected arterial cross-sections were counted in the following sections: 4 sections taken from each of the isolated right, left circumflex and left paraconal coronary arteries; 4 sections taken from the heart (right atrium with ventricular free wall, left atrium with ventricular free wall and septum). A lesion score was calculated for each animal by summing the number of affected cross-sections in all slides. Other organs were examined routinely for histopathological evidence of subclinical disease, in particular to rule out occurrence of idiopathic polyarteritis (Clemo et al., 2003).

RNA Isolation and Microarray Analysis

Total RNA was isolated from frozen left circumflex coronary arteries by extraction with an RNeasy lipid tissue mini kit (Qiagen, Valencia, CA). Labeled cRNA probes were generated from 100 ng of total RNA from each sample using a 2-cycle cDNA synthesis kit and in vitro transcription labeling kit (Affymetrix, Santa Clara, CA). The quantity and quality of total RNA and labeled cRNA were assessed by UV absorption spectrophotometry and by using Agilent’s Lab-on-a-Chip total RNA nano-biosizing assay (Agilent, Palo Alto, CA). Biotinylated cRNA probes were hybridized to Canine Genome Array chips, containing 23,838 probe sets for canine genes and expressed sequence tag (EST) clusters, using the Affymetrix Fluidics Station 400 according to the manufacturer’s standard protocol (Affymetrix, Santa Clara, CA).

The image data on each microarray chip was normalized using dChip software (www.dChip.org), and the model-based expression values were used in analyses to find differentially expressed genes. Genes having average expression values less than 200, detection flags absent for both comparison groups or fold changes less than 1.5 (up- or down-regulated) were excluded from further statistical analyses. To statistically identify genes that were differentially expressed between groups, a 2-sample t-test (significance level set at p = 0.05) was performed for each gene. To correct for multiple comparisons, the Benjamini and Hochberg false discovery rate was applied to adjust the p-values obtained from the t-tests (Reiner et al., 2003). Preliminary analyses of the canine coronary artery gene expression data revealed significant variation across samples (data not shown). Instead of comparing untreated and treated animals with different doses at different time points, we increased statistical power by comparing all the animals with arteriopathy to control animals to identify genes that were significantly changed. Principal component analysis (PCA) (Sherlock, 2001) was performed on differentially expressed genes to classify and visualize the results. Differentially expressed canine ESTs were annotated when possible by comparing the putative canine protein sequences with sequences of homologous human proteins in GenBank using BLASTx (Altschul et al., 1990).

Results

In-Life and Pathology

Significant clinical changes between treated and control dogs were limited to dose-dependent decreases in systolic BP and increases in HR at 4 hours postdosing (Figure 1).

Treated animals developed coronary arteriopathy with incidence and severity that were both dose and time dependent (Table 1). The arteriopathy was segmental and primarily affected small to medium branches of the extramural coronary arteries (~200–500 μm diameter), although a few small intramural branches also had similar histopathological changes. Left and right coronary arteries were affected, without obvious predilection for either. First pathological changes at the light microscopic level appeared in the tunica media. Minimal early changes in the vascular smooth muscle of some arteries were characterized by the formation of hyaline droplets and pockets containing proteinaceous material and/or erythrocytes. In more advanced lesions, some erythrocyte-filled pockets became transmural, and the media also contained pyknotic and karyorrhectic debris from necrotic and apoptotic smooth muscle cells and a few transmigrating leukocytes (Figure 2). Affected arteries sometimes had hypertrophic endothelium and a few marginated or adherent leukocytes. Additionally, some injured arteries were surrounded by small perivascular hemorrhages and/or minimal inflammatory infiltrations that generally consisted of a few mononuclear cells and neutrophils scattered among hypertrophic adventitial fibroblasts. The ventricular myocardium of a few dogs contained small focal subendocardial hemorrhages.

Hematology changes that accompanied development of the arteriopathy were confined to mild elevations in the neutrophil count of 3 of 4 dogs from the 16-hour and 24-hour 10-mg/kg dose groups (individual animal increases of 1.3 to 4.0-fold above predose values). Plasma fibrinogen values also increased in the same individual animals in these dose groups (1.3- to 1.5-fold increases above pre-dose values). Mean neutrophil counts and fibrinogen levels also were increased statistically for these groups when compared to controls (data not shown). One animal (21) in the 2 mg/kg/16-hour group had focal subclinical pneumonia and was excluded from the statistical analyses.

Gene Expression Profiling

Data analysis yielded 485 differentially expressed genes, of which 243 could be identified putatively as canine homologues of human genes or more definitively as corresponding to sequenced canine genes. PCA of the differentially expressed genes demonstrated a credibly robust segregation of dogs that had arteriopathy from those with no lesion (Figure 3). Of the genes that could be identified, twenty-five with significant differential expression and the highest fold changes up or down are presented in Tables 2 and 3 along with short functional annotations.

Discussion

At the pathophysiological and histopathological levels, this study demonstrated that dogs receiving a single dose of CI-947 exhibited significant hemodynamic changes at T_max (4 hours) followed by acute coronary vascular injury that was evident at 16 hours postdosing, 8 hours earlier than previously reported. The BP and HR changes were indicative of systemic vasodilation and reflex tachycardia caused by this compound. Coronary arteries in dogs are particularly sensitive to vasodilation produced by adenosine (Crystal et al., 1984). Concurrent with vasodilation, adenosine induces a 4-to 5-fold increase in coronary arterial blood flow that follows a very steep dose curve (Stepp et al., 1996) and persists in the face of continued supraphysiological exposure. Although increased blood flow is one of the most frequently cited pathophysiological mechanisms of drug-induced vascular injury (Kerns et al., 2005), the cellular- and molecular-level determinants of acute flow-induced injury are not well understood. Recent studies suggest that when increased flow generates turbulence, altered shear stresses can perturb the homeo-static balance of normal endothelial cells to initiate injury (Resnick et al., 2003). Laminar flow, unlike turbulent flow, maintains a vascular environment that is anti-inflammatory, anti-apoptotic and protected from oxidative stress (Desai et al., 2002; Wasserman and Topper, 2004). Consistent with a turbulent-flow mechanism, arterial diameter and blood-flow measurements made by fluorescence microangiography have demonstrated increased shear stress in the coronary arterial bed of dogs treated with adenosine in vivo (Stepp et al., 1995).

In this study with CI-947, gene expression was analyzed in RNA isolated from dissected whole arteries, so that the resulting changes reflected responses not only of endothelium, but also of other cell types including vascular smooth muscle, peripheral blood leukocytes, and adventitial stroma, as well as perivascular adipose and cardiac tissue that could not be completely dissected away. The gene expression profiles correlating with development of CI-947-induced coronary arteriopathy revealed a number of pathobiological responses active in this complex tissue. Up-regulated genes such as monocyte chemotactic protein-2 (MCP-1), complement component-3 (C3), lysozyme and the cathepsins were indicative of a pro-inflammatory response involving stimulation of the innate immune system (Staros, 2005). Increased matrix remodeling was evident in the up-regulation of genes such as fibronectin, gallectin, the matrix metalloprotease inhibitor TIMP-1, and 3UDP-glucose dehydrogenase, which synthesizes glycosaminoglycans (Tayebjee et al., 2005). Expression of oxidative stress response genes such as glutathione S-transferase and haptoglobin was also increased (Sadrzadeh and Bozorgmehr, 2004) contemporaneously with glucose-regulated protein, which is an indicator of hypoxia and nutrient stress (Flores-Diaz et al., 2004). Increased thrombospondin-1 gene expression indicated perturbation of normal hemostatic balance (Boffa and Karmochkine, 1998).

Many genes with decreased expression were of cardiomyocyte or smooth muscle origin. The largest group consisted of structural genes encoding myofibrillary, cytoskeletal and calcium-binding proteins (Stromer, 1995). Another major class of down-regulated genes included enzymes of the citric acid cycle and glycolysis, such as the rate-limiting phosphofructokinase, reflecting alterations in energy metabolism associated with depletion of energy stores and oxygen in the coronary tissue (Berg et al., 2002).

Turbulent blood flow can initiate a cascade of gene expression changes that override the protective endothelial cell phenotype induced by laminar flow. Although the gene expression changes seen in this study overlapped some with the patterns induced in endothelium by disturbed shear, e.g., strongly elevated MCP-1 (Yu et al., 2002), they may also reflect events distal to those initiated by disturbed flow as well as other potential molecular mechanisms of injury involving additional cell types. Adenosine receptors are expressed on many different cells, including endothelial and vascular smooth muscle cells as well as leukocytes and platelets (Sawynok and Liu, 2003). There are 4 known receptor subtypes (A1R, A2AR, A2BR, and A3R) that at pharmacological exposures mediate a wide range of sometimes opposing functions in the vasculature and innate immune system (Howlett et al., 2005) (Table 4). When activated, the A2Rs are responsible for the vasodilation of coronary arteries (Tabrizchi and Bedi, 2001), but this response is negatively modulated by A1R (Tawfik et al., 2005). Also, stimulation of various AR subtypes expressed on neutrophils, monocyte/macrophages, mast cells and platelets can positively or negatively modulate responses such as degranulation, release of cytokines and expression of adhesion molecules (Hasko and Cronstein, 2004). The complexity of adenosine receptor pharmacology suggests that the summation of effects produced by suprapharmacologic doses of a nonselective adenosine agonist such as CI-947 would be equally complex, involving hemodynamic changes as seen in this study as well as interactions among various blood constituents and the vasculature that could be drug-induced and/or secondary to turbulence (Yamamoto et al., 2003).

Gene expression profiles of injured coronary arteries in this study provide molecular evidence of pathobiological responses at work, but do not demonstrate clear signatures of a dominant cellular/molecular mechanism of injury at the time points examined. Some of these processes, such as matrix remodeling, correlated well with histopathological changes in the arterial wall. However, genes from some processes evident in the histopathology, e.g., apoptosis, were not represented among the most strongly modulated genes, an outcome probably due to the highly segmental nature of the lesion. It is apparent from the advanced state of the vascular lesion at 16 and 24 hours that injury must have been initiated earlier post dosing, likely implicating maximum altered hemodynamic forces, rather than sustained hemodynamic change, in the initiation of the coronary arterial damage. Future studies will examine effects of in vivo CI-947 treatment at time points prior to 16 hours to investigate the initiating events of the injury and dissect early molecular mechanisms away from later secondary events that contribute to lesion development. We also will mine more information from the transcriptomic data through in-depth bioinformatics analysis to determine the temporal progression of biological pathway perturbations. Complementary in vitro studies will be aimed at evaluating the contribution of particular cell types, especially the endothelium and peripheral blood leukocytes, in promoting vascular injury. Better understanding of the cellular and molecular mechanisms of earliest injury should, in turn, allow development of leading biomarkers that are clearly linked to vascular lesion development.