Gastric and Pancreatic Lesions in Rats Treated with a Pan-CDK Inhibitor

Introduction

This article describes lesions in preclinical studies produced by a pan-Cyclin Dependent Kinase (CDK) inhibitor (AG-12986) not previously reported. This class of compounds is currently an important component of the relatively novel approach of targeted therapy against cancer.

The Center for Disease Control (CDC) has reported that cancer is the second leading cause of mortality in the United States (USA), and it is expected to become the leading cause of death in the present decade. The therapy against cancer continues its evolution, but still relies heavily on traditional DNA damaging agents or blockers of DNA synthesis as the first line of treatment (Foye, 1995; Sawyers, 2004). Nowadays, the increasing knowledge of the molecular pathways that participate in tumor growth or progression has allowed the entrance of targeted therapy into the arena of cancer treatment (Sawyers, 2004). By increasing the specificity against the target, this novel approach is intended to augment the effectiveness of the compounds while reducing the adverse effects of the treatment.

Many of the targeted therapies against cancer are focusing on one of the foundations of tumorigenesis, the cell cycle (Buolamwini, 2000; Senderowicz, 2003; Sawyers, 2004). Numerous studies have demonstrated that abnormal expression or function of molecules involved in the cell cycle are the origin of a variety of malignant neoplasms (Hunter and Pines, 1994; Massague, 2004). Thus, targeting the cell cycle controlling molecules is one of the most dynamic and promising approaches for anticancer treatment (Buolamwini, 2000).

In the context of cell cycle regulation, CDKs are recognized as critical molecules for the control and activation of the cell cycle in normal cells (Schafer, 1998). At the same time, the overexpression or loss of CDK functions may lead to unregulated cellular proliferation and generation of tumors (Pardee, 1989). CDKs are serine/threonine kinases activated by binding to cyclins within the nucleus. Different cyclin isoforms bind to distinct CDKs and participate in the dynamic phases of the cell cycle (Schafer, 1998; Johnson and Walker, 1999). In a simplified overview, CDK4 and CDK6 interacting with D Cyclins participate in the progression from G1 to S phase. CDK2 bound to Cyclin A regulates the progression through S phase. Finally, CDK1 in concert with Cyclins B and A controls the G2-M transition and the mitosis (for a detailed review on cell cycle, see Schaffer, 1998). Regarding the clinical application of CDK inhibitors, some reports have created uncertainty due to the contradictory results observed between preclinical studies and knockout models (Buolamwini, 2000; Kozar et al., 2004; Malumbres et al., 2004). In spite of the initial uncertainty, the development of CDK inhibitors has continued after showing interesting preclinical effects on cell cycle progression, induction of apoptosis, promotion of cellular differentiation, inhibition of angiogenesis, and modulation of transcription (Senderowicz and Sausville, 2000). Furthermore, some clinical trials have demonstrated a positive effect of some CDK inhibitors in combination therapy (Arguello et al., 1998; Zhai et al., 2002; Senderowicz, 2003).

In the next sections, we present in more detail interesting preclinical toxicologic findings of a pan-CDK inhibitor interrupted in development due to safety concerns. To our knowledge, the lesions we observed in the pancreas and stomach have not been previously described in the literature for compounds targeting CDKs.

Material and Methods

Test Substance

AG12986 is a Pfizer proprietary compound belonging to the aminothiazole series. It was studied for its properties as a potent inhibitor of CDK1, CDK2, CDK4, and CDK5 at nM concentrations within a 10-fold range for these 4 kinases. The vehicles utilized were D5W/Acetic Acid for the second experiment and 0.1 M Acetic Acid plus 0.1 NaCl for the other 2.

Tissue Samples and Processing

Tissue samples were obtained from male Sprague–Dawley rats (Charles River Laboratories, Boston, MA) included into 3 different preclinical experiments to assess in vivo toxicity. The animals weighed between 180–200 g for the first and third experiments, and between 300–340 g for the second experimental procedure at the start of dosing. The toleration studies were designed as follows: (1) one 5-day IV repeat dose study with pathologic evaluation at 6 and 26 days postdosing. The doses utilized in this study were 3, 10, and 75 mg/kg; (2) 1 single dose intravenous (IV) toxicology screening study with pathologic evaluation at 3 and 8 days postdosing. The dose selected for this experiment was 20 mg/kg; (3) and 1 single IV dose toxicokinetic study with pathologic evaluation at 24 hours postdosing, utilizing doses of 10 and 50 mg/kg.

The studies were conducted in accordance with the current guidelines for animal welfare (Guide for the Care and Use of Laboratory Animals, 1996; Animal Welfare Act, 1966, as amended in 1970, as amended in 1976, and 1985, 9 CFR Parts 1,2,3). The procedures used in this study were reviewed and approved by the internal Institutional Animal Care and Use Committee.

The sampled tissues were fixed in 10% formalin, processed by routine histologic techniques, and paraffin embedded. Five-micron sections were stained with hematoxylin and eosin and analyzed under the light microscope. In addition to pancreas and stomach, we examined a more comprehensive set of tissues in all the studies.

We analyzed in greater detail the pancreas and stomach for the presence of single-cell death semiquantitatively, assigning subjective values to the samples depending on the amount of individually necrotic cells. The features utilized include the presence of cells with cytoplasmic hyperchromasia and rounding, separation from adjacent cells, nuclear pyknosis, nuclear fragmentation, pericellular clear halo, and phagocytosis by adjacent cells. In the endocrine pancreas, some cells showed cytoplasmic swelling concurrent with marked nuclear pyknosis. These cells were also considered individually necrotic for scoring purposes. A total of 21 pancreata and 18 stomachs from control animals, and 32 pancreata and 25 stomachs from treated animals were examined. The scoring criteria for the H&E stained sections in the exocrine pancreas and gastric mucosa was established as follows in 10 randomly chosen 20× magnification fields:

• 0– no evidence of single cell death;

• 1– less than 10 individual cell deaths;

• 2– > than 10 < than 30 single-cell deaths;

• 3– > than 30 < than 50 single-cell deaths;

• 4– >than 50.

For the endocrine pancreas (islets), a total of 10 islets were examined. The scoring system was the following:

• 0–no single-cell death;

• 1–1 to 2 single-cell deaths in 10 islets;

• 2–3 to 10 single-cell deaths in 10 islets;

• 3–11 to 20 single-cell deaths;

• 4–more than 20 single-cell deaths.

We evaluated and scored samples of gastric mucosa for the presence of single-cell death following the same criteria as for the exocrine pancreas.

Immunohistochemistry

Selected samples (pancreas and stomach) were processed for single immunohistochemistry with rabbit anticaspase 3 antibody (cleaved form) (Cell Signaling Technology) and mouse anti-topoisomerase II-α(clone SWT3D1, DakoCytomation). For dual staining, guinea pig antiinsulin and rabbit antiglucagon antibodies (DakoCytomation) were also utilized. Briefly, for the anticaspase 3 IHC the sections were placed on charged slides, deparaffinized, and washed with Tris buffer. Antigen retrieval was developed with a Decloaking Chamber (Biocare Medical) in Reveal buffer pH 6 (Biocare Medical) for 5 minutes at 125°C, followed by cooling down at 90°C, and rinsing in tap water and Tris buffer. Peroxidase blocking was done by incubation with 3% hydrogen peroxide in distilled water for 10 minutes at room temperature. The immunohistochemistry was developed in an automatic immunostainer (DakoCytomation). After stabilizing in TBS-Tween20, the samples were preincubated with Protein Block-Serum Free (DakoCytomation), and followed by incubation with a 1/100 dilution of the primary antibody for an hour at room temperature. Subsequently, the samples were incubated with a 1/200 dilution of secondary goat-anti-rabbit antibody bound to a conjugate polymer-Horseradish peroxidase (HRP) (DakoCytomation). Finally, the slides were incubated with DAB substrate (Invitrogen Corporation) and counterstained with hematoxylin.

In the case of dual labeling, the primary antibodies, ready-to-use antiinsulin or antiglucagon, were detected with HRP conjugates (DakoCytomation). The second antibody anticaspase 3 was detected with an avidin-biotin conjugate bound to alkaline phosphatase (Vector Laboratories).

The staining for topoisomerase II was developed in a Ventana Discovery XT Immunostainer (Ventana Medical Systems). Specific settings for the protocol included proprietary antigen retrieval system (standard CC1), avidin-biotin blocking for 8 minutes, 1-hour incubation with 1/500 dilution of the primary antibody at room temperature, and 32 minutes, incubation with a rat-serum-preadsorbed secondary horse-anti-mouse (Vector Laboratories). The staining was revealed utilizing the DAB map kit (Ventana Medical Systems).

We applied the same scoring system for the anticaspase 3 immunohistochemistry and the histopathology, except for the stomach, where we modified the criteria to the following:

• 0–no evidence of cleaved caspase 3 positive cells;

• 1–<30 positive cells in 10 × 20 magnification fields;

• 2–30–50 positive cells in 10 × 20 magnification fields;

• 3–51–70 positive cells in 10 × 20 magnification fields;

• 4–>70 positive cells in 10 × 20 magnification fields.

Topoisomerase II positive cells were counted in representative animals, obtaining a semiquantitative estimate of the amount of positive cells in 7–10 high-power fields. The total number of positive cells was compared between the different groups.

Results

Pathology

We observed dose-related mortality at 24 hours in animals treated at doses of 20 mg/kg and higher (20, 50, and 75 mg/kg) (Table 1). Grossly, animals dosed at 10 mg/kg or higher had decreased thymus size. Histologically, the lymphoid organs (thymus and lymph nodes), the bone marrow and the spleen had leukocytolysis, which was also seen in small capillaries of different organs (heart, lung, kidney) (Figure 1). Single-cell necrosis occurred in the liver and intestinal mucosa (small and large intestine), as well as in the epithelial lining of the epididymis. In addition, the exocrine and endocrine pancreas, and the stomach (fundus and pylorus) exhibited parenchymal and mucosal single-cell necrosis. In the next sections, we will focus on the lesions of the pancreas and stomach.

In the exocrine pancreas of untreated animals, we observed the presence of rare single-cell death. In contrast, rats administered repeated doses, or in acute phase at a single dose, showed a dose-related increase in individual cell death (from 10 to 75 mg/kg) (Figures 2A and 2B). The single nec- rotic cells were often rounded and separated from the adjacent cells. The cytoplasm was darker and amphophilic. The nucleus was round and pyknotic. Frequently there was a clear halo around dead cells. Occasionally, the adjacent epithelial cells engulfed the necrotic elements, but in general, the single-cell death did not induce a local inflammatory response.

At the highest doses (50 and 75 mg/kg) acinar cells had decreased perinuclear basophilia (reduced nucleic acids in the RER) and the exocrine acini were in some cases disaggregated. Decreased amount of zymogen granules in exocrine cells was present at lower doses (10 and 20 mg/kg). Lesions in the exocrine pancreas were associated with increased serum amylase and lipase at 75 mg/kg (data not shown) in the only experiment in which these enzymes were evaluated.

The endocrine pancreas was also affected by the treatment, showing a lower degree of susceptibility to the toxic effects compared to the exocrine portion. This is demonstrated by the low severity of lesions at a dose of 10 mg/kg (single or repeated), compared to the effects at the same dose in the exocrine portion (Table 1, experiments 1 and 3). Animals treated with doses starting at 20 mg/kg showed dose-related toxicity to the endocrine cells, manifested as nuclear pyknosis, fragmentation, and cytoplasmic vacuolation. Frequently, the peripheral cells of the islets were the targets for toxicity at 20 mg/kg, but at the higher doses (50 and 75 mg/kg) central cells were also affected (Figures 2A and 2B). The lesions were not present in all islets within a pancreatic tissue sample.

When the animals were treated with a single dose (at 20 mg/kg), a recovery period (5 days) allowed reversibility of the characteristic lesions in the exocrine and endocrine pancreas (see in Table 1, experiment 2, the difference in the scoring between day 3 and day 8).

The gastric mucosa was another of the unexpected target organs for toxicity. Stomach samples from control animals showed sparse individually necrotic cells at the neck and upper portion of the glands. Fragmented cells were rarely observed in the lower half of the glands. In animals treated with this pan-CDK inhibitor cellular fragmentation (single-cell death) was seen in the mid (isthmus) and basal gland areas of the fundus at doses of 10–20 mg/kg. At higher doses, the full length of the glands was affected by the toxic effect, with minimal effects on the superficial layer of mucus cells (Figures 2C and 2D). Similar to the fundus, sections of the pylorus showed individual cell necrosis, mainly in the basal 2/3 of the glands, occasionally progressing to deep mucosal erosions with colonizing bacteria (Figure 3). The damaged cells in the gastric mucosa were typically rounded, with hyperchromatic and pyknotic nucleus. Nuclear dust and cell separation were commonly seen (Figures 2D and 3).

Immunohistochemistry

Cleaved (active)-caspase 3 immunostaining identified a similar distribution of single necrotic cells as described with routine histopathology. The positive signal presented as diffuse or finely granular cytoplasmic staining (Figures 2F and 2H). We saw small numbers of immunopositive cells in the exocrine pancreas and gastric glands in the controls animals, but no positive cells in the islets. In contrast, the IHC showed immunopositive cells in the treated animals starting at 10 mg/kg and in a dose-related fashion. This effect was obvious at single or repeated doses (Figures 2E and 2F).

In the stomach of untreated animals, immunohistochemistry for activated caspase 3 demonstrated the presence of scattered positive cells within gastric glands, often in the luminal half of the glands. Scattered positive cells were also seen at the transition squamous-glandular mucosa in both mucosal sides. After treatment, a few animals administered a low dose (3 mg/kg) showed an increase in the number of positive cells at the isthmus of gastric glands, but infrequently at the basal portion of the glands. With increasing dose (10–20 mg/kg), the individually necrotic cells appeared regularly at the basal portion of the glands as well. Lastly, at the highest dose (75 mg/kg), the full length of the gland was frequently positive (Figures 2G and 2H).

Due to the particular distribution of some of the damaged cells in the islets, we decided to utilize dual antigen labeling in an effort to identify a specific target population. The primary antibodies selected were antiglucagon (α-cells) and antiinsulin (β-cells), in each case combined with anti-activated caspase 3 as a secondary antibody. We could not find a specific correlation between apoptosis and a defined cell population.

The evaluation of cell proliferation was accomplished utilizing an antibody against topoisomerase II-α, which detects cells in late S-phase and G2 of the cell cycle (Woesner et al., 1991). We did not observe differences in the proliferative index in pancreas (exocrine and endocrine), and stomach at 3, 10, and 20 mg/kg (Figure 4). A reduced number of positive cells (average of 9.1 per high magnification field versus average of 40.7 for the other groups) was observed at a dose of 75 mg/kg in the stomach, but not in any of the pancreatic compartments (Figure 4). Some of the positive cells at the high dose had morphologic features indicative of apoptosis (Figure 1H, inset).

Discussion

In this article, we show that the treatment of Sprague–Dawley rats with a pan-CDK inhibitor exerts toxicological effects in stomach and pancreas (exocrine and endocrine), targets not previously reported. Studies of the drug tissue distribution showed presence of parental compound in the pancreas up to 48 hours and in the stomach up to 192 hours after single IV injection (data not shown). The exact mechanism inducing these effects is currently unknown. The positive results by IHC for active-caspase 3 indicated that apoptosis plays a major role in the induction of these toxic phenomena.

Analyzing the results, we have observed a good correlation between single-cell necrosis recognized by morphologic features of histopathology and active-caspase 3 immunohistochemistry, indicating death by apoptosis. The immunohistochemistry showed increased sensitivity as it detects cells in initial stages of apoptosis not recognized by morphological features. The activation of caspase 3 is one of the early stages in the induction of some forms of apoptosis (Slee et al., 2001).

In the pancreas, the histopathology and immunohistochemistry revealed presence of low numbers of apoptotic cells in the exocrine pancreas of untreated animals, which suggests epithelial renewal in this fraction of the organ. Some authors have previously shown that the replication of cells is closely linked to apoptosis and they often show similar distribution (Aguda and Algar, 2003). In addition to the primary effect of single-cell necrosis observed in the H&E, some animals treated with the highest dose (75 mg/kg) showed a progression to pathological phenomena (acinar disaggregation, loss of perinuclear basophilia) common to other pancreatic diseases, as in models of pancreatitis (Weaver et al., 1994). This suggests a common final outcome for processes inducing pancreatic acinar destruction.

In clear contrast to the exocrine portion of the pancreas, it was extremely rare to find any evidence of apoptosis in the islets of controls animals. This is suggestive of an inferior regenerative capacity, as previously indicated by some authors (Bouwens, 1998). In fact, some researchers have hypothesized that the β-cells in the islets originate from migrating acinar cells, and that there are no regenerative stem cells in the islets (Bouwens, 1998; Bock, 2004). In contrast, other authors have reported the existence of replication within the islets (Magami et al., 2002). Despite this controversy, the alleged low replication index might explain the lower susceptibility of the endocrine pancreas to alterations of the cell cycle.

In the animals that showed a toxic effect on the islets, the lesions at low doses (10–20 mg/kg) were located at the periphery of the islets, location where glucagon cells are usually present (seen by IHC, data not shown). Nevertheless, the dual antigen labeling did not reveal a preferential targeting of a defined cell population because cleaved caspase-3 positive cells were seen throughout the islets. Curiously, the glucagon cells have been reported to have the highest replicative rate among the islet populations (Magami et al., 2002). This characteristic might make glucagon cells more susceptible to the interference with the cell cycle.

The interference with other cellular functions might instead explain the induction of apoptosis in the endocrine cells. For example, it has been reported that CDK5 has a role in the fusion of secretory granules in β-cells of the pancreas, suggesting the possibility of secretory disregulation as another possible cause of cell damage observed in the pancreatic islets (Lilja et al., 2001).

In the stomach of control animals, we detected small numbers of active-caspase 3 positive cells not observed by routine histology. These cells were mainly located at the squamous-glandular mucosal transition and at the isthmus of the gastric pits, areas of cell division (Modlin et al., 2003). The IHC for topoisomerase II-α revealed a higher number of positive cells at the glandular isthmus, indicative of higher rate of replication (Figure 4A). Similar to the pancreas, the caspase 3 positive cells in the stomach were present in areas of replication, as reported by other authors (Hoshi et al., 1998).

In the stomach, the most striking phenomenon in animals treated with a high dose was the extension of single-cell necrosis throughout the full length of the gastric glands, into areas where replication is not predominantly observed. The explanation for the extension of the apoptosis to the full length of the glands is not clarified by our findings, but at least 2 possible interpretations come to mind, which could be applied also to pancreatic lesions. First, the apoptosis in areas of replication might extend to adjacent cells by stimuli delivered from neighboring apoptotic cells. This phenomenon has been observed in in vitro paradigms (Kim, 2001; Lo et al., 2005). Alternatively, the interference with the function of CDKs or other off-target kinases might encompass cellular processes not directly bound to cell division. In this sense, different authors have described cell-cycle independent roles of the CDKs in cellular functions, as in cytoskeletal reorganization (Slangy et al., 1995), cell motility (Fahraeus and Lane, 1999), and DNA repair (Esashi et al., 2005), among others. In fact, Sausville, a recognized authority in cancer therapy with CDK inhibitors, has indicated recently that the antitumor activity of most of these pan-CDK inhibitors might be driven by the inhibition of RNA synthesis, because CDKs 7, 8, and 9 play an important role in the phosphorylation of RNA polymerase II (Sausville, 2005). Nevertheless, due to the fact that the lesions in the pancreas and stomach have not been described for other pan-CDK inhibitors, we consider an off-target effect specific for AG-12986 the most likely cause of the phenomena.

The decrease of positive topoisomerase II staining in the gastric mucosa at a dose of 75 mg/kg, concurrent with the presence of apoptosis indicates a correlation between both processes. The detection of a few topoisomerase II-α positive cells with morphology of apoptosis suggests that at least in some cases the cells are entering into apoptosis during cell division. However, the specific mechanism of induction of apoptosis (target or off-target effect) has not been investigated.

The development and application of cell-cycle targeted therapies against cancer has opened up new avenues to control the disease, but at the same time has created new paradigms of toxicologic effects (Arguello et al., 1998; Sausville et al., 1998). We have presented additional target sites in preclinical models, which might be specific for the compound studied and not derived from the intended pharmacological effect. This information may contribute to a more comprehensive monitoring in preclinical and clinical studies.

Several pan-CDK inhibitors are currently in advanced stage of clinical development. Flavopiridol is the frontrunner for the class and is a pan-CDK inhibitor with a broad spectrum of activity against CDKs. There are also other compounds, such as Olomucine and Roscovitine, currently in clinical development with a different inhibitory profile (Fischer and Gianella-Borradori, 2003). For most of these compounds the limiting toxicities in humans are the effects on white blood cells and the gastrointestinal tract, being diarrhea and neutropenia the dose-limiting signs (Arguello et al., 1998; Stadler et al., 2000; Zhai et al., 2002; Fischer and Gianella-Borradori, 2003). Besides, flavopiridol has shown important vascular thrombotic events and asthenia in phase II trials (Stadler et al., 2000). The previously described toxicological effects in pre-clinical models have a similar outline, but the morphology of the lesions is poorly described. Toxic effect on bone marrow, thymus, spleen, and intestine are reported (Sedlacek et al., 1996; Arguello et al., 1998; Sausville et al., 1998).

After examination of these 3 studies, we have observed a variety of lesions that included effects on the immune system (thymus, bone marrow, spleen, lymph nodes) and the gastrointestinal tract (large and small intestine). Somewhat unexpected toxic effects were seen in the exocrine and endocrine pancreas, and the stomach. The toxic effects in the different cell types are, at least in part, mediated by apoptosis, as we have shown by active-caspase 3 immunohistochemistry, and in agreement with previous data (Arguello et al., 1998; Senderowicz and Sausville, 2000). From a mechanistic point of view, the induction of apoptosis by some CDK inhibitors has been partially counteracted in vitro by caspase inhibitors, but not with overexpression of Bcl-2 nor Bcl-XL (Ribas and Roix, 2004). This has been interpreted as a non-mitochondrially driven apoptotic pathway induced by the inhibition of the action of CDK1 on survivin and the release of caspase 9 from the nucleus into the cytosol for this in vitro model (Ribas and Roix, 2004).

Flavopiridol, Olomucine, and Roscovitine, all CDK inhibitors with different spectra, have shown induction of apoptotic death in in vitro and in vivo models, suggesting a common mechanism of action derived from the pharmacological effects (Ribas and Roix, 2004). The precise mechanism causing the lesions in the pancreas and stomach in rats is not yet defined based on this morphologic study.

Despite the effects on the cell cycle of CDK inhibitors and the induction of apoptosis, clinical trials have shown that monotherapy is not currently a primary option for CDK inhibitors tested to this date. Yet, in combination with traditional chemotherapy they have demonstrated a beneficial effect (Fischer and Gianella-Borradori, 2003). The numerous ongoing clinical trials for this class may continue adding information to improve the therapeutic regimes against cancer.

In summary, we have described additional target organs for undesirable toxicological effects in preclinical models for a pan-CDK inhibitor. The effects were considered a direct compound effect and dose related, but the exact mechanism inducing the lesions (direct CDK inhibition or off-target effect) is unknown. Given the lack of these reported lesions with other CDK inhibitors, the effects observed with AG12986 may be due to its unique CDK inhibitory pattern, inhibition of off-target kinases, interaction with non-kinase macromolecules, or unique metabolism/distribution properties. We have demonstrated the induction of apoptosis in both the pancreas and stomach after treatment, as shown by the immunopositive staining with anti-active caspase 3 antibody. Immunostaining with topoisomerase II-α shows that some of the cells entered into apoptosis during cell division. Apoptosis in general can be induced by a variety of mechanisms that converge into a common internal pathway of self-destruction.