Retinal and Peripheral Nerve Toxicity Induced by the Administration of a Pan-Cyclin Dependent Kinase (cdk) Inhibitor in Mice

Abstract

Cyclin-dependent kinases (cdks) play a crucial role in cell cycle regulation and are considered promising targets for cancer therapy. Intravenous administration of AG-012986, a pan-cyclin-dependent kinase inhibitor (cdk_i), resulted in unexpected retinal and peripheral nerve toxicity in mice. AG-012986 was administered daily to CD-1 or B6C3F1 mice for 5 consecutive days. Mice were euthanized 24 h after the last dose (study day 6) or after a 21-day post-dose period (study day 26). Compound related microscopic findings were seen in the sciatic nerves (axonal degeneration) of both strains and in the retina (retinal degeneration/atrophy) of CD-1 mice only after the post-dose period. Although retinal degeneration/atrophy was not detected by routine histology in mice euthanized on day 6, apoptotic retinal cells were evident at this time using TUNEL assay. To our knowledge retinal or peripheral nerve toxicity secondary to the administration of cdk_is has not been previously reported. Although the pathogenesis of these lesions is unclear, the toxicities may reflect the unique profile of cdk inhibition, off-target kinase inhibition or receptor binding, or metabolism/distribution properties of AG-012986. Multi-targeted-inhibitors may interfere with cdks and other kinases involved in a wide range of functions other than cell cycle regulation, which could result in unexpected toxicities that may hinder their clinical applications.

Introduction

Cyclin-dependent kinases (cdks) are a family of enzymes first characterized for their control of progression through the cell cycle and their dependence on cyclin proteins for activity (Lee and Nurse, 1987). Since deregulated cdk activity is known to be a hallmark of malignancy, cdks are currently viewed as promising targets for cancer therapy. Cdk_is are novel chemotherapeutic agents that can arrest proliferating cells at either G1-S and/or G2-M interface and are capable, under certain circumstances, of inducing apoptosis in a p53- and pRb-independent manner (Garret and Fattaey, 1999).

The toxicities expected from inhibiting the cell cycle via cell cycle associated kinases are those common to traditional chemotherapeutic cytotoxic agents that selectively target tissues with a high rate of cell proliferation; namely gastrointestinal and bone marrow toxicities. Non-selective muti-targeted Cdk_is however, have the potential to inhibit cdks and possibly other off-target kinases involved in cellular processes outside the scope of cell cycle regulation which may lead to toxicities not limited to those observed in cell populations with high mitotic activity.

In this publication we report the unexpected findings of retinal and peripheral nerve toxicities in outbred albino mice (Crl:CD-1(ICR)BR) from a study designed to evaluate the potential toxicity and pharmacokinetic profile of AG-012986, a pan-cyclin-dependent kinase inhibitor that binds in the ATP binding site. Retinal toxicity resulting from systemic exposure to compounds during drug development is a rare but serious finding, often leading to compound attrition.

Materials and Methods

Mouse Studies

Crl:CD-1(ICR)BR mice (referred to as CD-1 in the text; 20/sex/dose; Charles River Laboratories, Portage, MI) were treated with either vehicle alone (0.1M acetic acid, 0.1M NaCl, pH 4.5) or 5, 15 or 50 mg/kg of AG-012986 4-{[4-amino-5-(2,6-difluorobenzoyl)-1,3-thiazol-2-yl]amino}-N-[(1R)-2-(dimethylamino)-1-methylethyl]benzamide by daily intravenous tail vein injections for 5 consecutive days. Mice were approximately 6 weeks old at the start of the study. Mice were euthanized by carbon dioxide inhalation and necropsied 1 day (10/sex/dose) or 21 days (10/sex/dose) after the last dose administration. A similar safety study was performed in male B6C3F1 (pigmented) mice (referred to as B6 in the text; 6/dose; Charles River Laboratories, Portage, MI) with necropsies 1 (day 6), 10 (day 15), and 21 (day 26) days after dosing. A single dose study was also performed in CD-1 mice (20/sex/dose) at doses of 0 (vehicle), 10, 30, or 100 mg/kg. Necropsy occurred either 5 or 21 days after dosing. Clinical observations were recorded daily throughout the studies. All studies mentioned in this report were approved by an Institutional Animal Care and Use Committee and conform to the Guide for the Care and Use of Laboratory Animals.

Histopathology

At the time of necropsy, both eyes, including the optic nerve, were fixed whole by immersion in 6% glutaraldehyde. The right sciatic nerve and other tissues routinely sampled in drug safety studies were fixed in neutral buffered formalin. The brain and cervical, thoracic and lumbar segments of the spinal cord were collected and examined microscopically in studies conducted with CD-1 mice only. Tissues were dehydrated, embedded in paraffin, sectioned and stained with hematoxylin and eosin (H&E) for microscopic examination. In this report only microscopic findings in the eyes and sciatic nerves will be discussed.

Staining for Apoptosis

Using the manufacturer’s specified protocol, paraffin-embedded sections of eyes from the first 5 CD-1 male and female mice necropsied 1 day after treatment from the vehicle control and high dose (50 mg/kg) groups were stained for apoptotic nuclei using the ApopTag Plus Fluorescein Detection Kit terminal deoxynucleotidyl transferase (TdT)-mediated dUTP nick end labeling (TUNEL) assay (Chemicon International, Inc., Temecula, CA). Two slides from each group were manually quantitated for the total number of cells and apoptotic cells within the outer and inner nuclear layer of the retina. The remaining 3 slides per group were semi-quantitatively assessed as yielding similar results as the 2 representative slides.

Results

Repeat Dose Study of AG-012986 in CD-1 Mice

CD-1 mice were treated intravenously for 5 consecutive days with AG-012986 at doses of 0 (vehicle control), 5, 15, or 50 mg/kg/day and necropsied either 1 (day 6) or 21 days (day 26) after the last dose. While clinical signs were recorded daily throughout the study, no evidence of retinal or nerve toxicity was observed. Microscopic examination of HE-stained sections of the eyes and sciatic nerves of AG-012986-treated CD-1 mice euthanized 1 day after the last dose administration revealed no compound-related findings. A normal control retina is shown in Figure 1. One female in the control group and one female in the high dose (50 mg/kg/day) group had retinal atrophy characterized by a total absence of photoreceptors and the outer nuclear layer of the retina (Figure 2). These microscopic changes were interpreted as compatible with those of naturally occurring retinal degeneration/atrophy occasionally seen in this and other strains of mice (Hubert et al., 1999).

Clear compound-related retinal degeneration/atrophy and nerve fiber degeneration of the sciatic nerves were observed in males and females of the high dose (50 mg/kg/day) group euthanized after a 21-day post-dose period. Nerve fiber degeneration in the sciatic nerve was present in all mice from the high-dose group and was characterized by minimal to mild axonal degeneration with formation of “digestion chambers” containing occasional axonal fragments and myelin debris (Figure 3). A summary of the incidence and severity of the sciatic nerve degeneration from CD-1 mice evaluated after the 21-day post-dose period is shown in Table 1. Trace sciatic nerve degeneration was observed in 1/10 of the control males and 1/10 of the females in the low (5 mg/kg/day) and mid dose groups (15 mg/kg/day) and were not considered compound related. Sciatic nerve degeneration was observed in all of the high-dose mice (50 mg/kg/day). In males, the incidences of trace and mild sciatic nerve degeneration were 3/10 and 7/10, respectively. In females, the incidences of trace and mild nerve degeneration were 7/10 and 3/10, respectively. Sections of the brain and cervical, thoracic and lumbar segments of the spinal cord were examined microscopically but were free of compound-related findings (data not shown).

Compound-related retinal degeneration/atrophy observed in CD-1 mice was characterized by a variable decrease in the thickness and cellularity of the outer layers of the retina, mainly the photoreceptor and outer nuclear cell layers (Figure 4). A summary of the incidence and severity of the retinal atrophy observed in CD-1 mice after the 21-day post-dose period is shown in Table 2. Retinal atrophy was observed in 1 low-dose male. Because of the complete absence of the photoreceptor and outer nuclear cell layer observed in this case, this change was interpreted as compatible with naturally occurring retinal degeneration/atrophy and therefore nonrelated to the administration of AG-012986. Trace retinal atrophy was observed in 6/10 males and females in the high-dose group. The compound-related retinal effects were diffuse and bilateral in nature. No microscopic evidence of nerve fiber degeneration was observed in the optic nerves (data not shown).

Single Dose Study of AG-012986 in CD-1 Mice

A single dose study of AG-012986 was also performed in CD-1 mice at doses of 0 (vehicle), 10, 30, or 100 mg/kg with histologic assessment of tissues taken 4 (day 5) or 21 (day 22) days after dosing. The doses and resulting acute plasma levels were approximately twice as high as those used in the repeat-dose study, while total compound exposure over the course of the study was less from the single dose (data not shown). Delayed peripheral neuropathy similar to that described in the multidose study was present at day 21 in the 30 and 100 mg/kg dose group. A summary of the incidence and severity of sciatic nerve lesions resulting from single dose administration of AG-012986 to CD-1 mice is shown in Table 3. Trace degeneration was observed in 1 control female and was not considered treatment-related. No sciatic nerve degeneration was seen at the low dose (10 mg/kg), while trace degeneration was observed in 7/10 males and 4/10 females in the mid dose (30 mg/kg). In the high-dose group (100 mg/kg), trace sciatic nerve degeneration was observed in 2/10 males and 1/10 females, mild degeneration was seen in 8/10 males and females, and moderate degeneration was observed in 1 female. Sections of the brain and cervical, thoracic and lumbar segments of the spinal cord were examined microscopically but were free of compound-related findings (data not shown). Retinopathy was not observed in the single dose study.

Unlike the repeat dose study, clinical signs associated with peripheral neuropathy were observed in the single dose study (Table 4). The incidence of abnormal gait, splayed limbs, and tremor was ≤1 per gender on days 1–7 in the high-dose group, and in the lower dose groups throughout the study (data not shown). Signs of abnormal gait were observed in high dose (100 mg/kg) females beginning on day 8, followed by splayed limbs on day 9. By day 11, both high-dose males and females presented signs of abnormal gait, splayed limbs, and tremors. The incidence generally increased from the time of first observation until day 16 at which time the incidences remained roughly constant through the end of the study. By day 21, 6/10 males and 8/10 females displayed tremors, 4/10 males and 5/10 females presented with splayed limbs, and 2/10 males and 4/10 females showed abnormal gait. The presence of clinical signs and slight increase in severity of sciatic nerve degeneration observed in the single dose study versus the repeat-dose study implies that the sciatic nerve degeneration was related to the maximum plasma concentrations of the compound rather than the cumulative exposure.

Repeat Dose Study of AG-012986 in B6 Mice

Pigmented (B6) mice were treated with AG-012986 under conditions that elicited retinal atrophy in CD-1 mice in order to investigate the role of retinal pigmentation in the observed toxicity. Male B6 mice were treated with either vehicle (control) or 50 mg/kg/day of AG-012986 for 5 days and necropsied 1 (day 6), 10 (day 15), or 21 days after dosing (day 26). The day 15 time point was added to better characterize the time course of occurrence of the microscopic lesions. Microscopic examination of HE-stained sections of the eyes and sciatic nerves of AG-012986-treated B6 mice euthanized 1 day (day 6) after the last dose administration revealed no compound-related findings. Compound-related nerve fiber degeneration of similar severity to that observed in the sciatic nerves of CD-1 mice was observed in all high-dose (50 mg/kg/day) animals necropsied after a 10 (day 15) and 21 (day 26) day post-dose period (Table 5). Sciatic nerve degeneration was not observed in any of the vehicle treated mice while trace degeneration was observed in 4/6 B6 mice on both days 15 and 26. Mild nerve degeneration was observed in 2/6 B6 mice at both time points. Clinical signs were recorded daily but none related to sciatic nerve degeneration were observed. Microscopic evaluation of the retinas from the AG-012986-treated B6 mice revealed no evidence of retinal atrophy at any time point.

Staining for Apoptosis in Retinal Sections

Although light microscopy examination revealed no morphologic evidence of retinal toxicity in CD-1 mice necropsied 1 day after the last administration of the high dose of AG-012986 (50 mg/kg/day), the possibility of scattered apoptosis was explored. TUNEL staining of retinal sections from the first 5 mice in the control and high-dose male and female mice revealed scattered apoptotic nuclei detected in the retina of compound-treated mice (Figure 5A). Semi-quantitative assessment of the slides from CD-1 mice treated with 50 mg/kg/day revealed that approximately 1–3% of the cells (~15–40 cells in a field containing ~900–1600 cells, data not shown) stained positive for apoptotsis, while less than 0.1% of the cells in slides from control mice stained positive for apoptosis (Figure 5B). The majority of the apoptotic cells were detected in the outer nuclear layer containing the cell bodies of the rods and cones. Microscopic re-evaluation of the HE-stained sections of the TUNEL-positive retinas failed to detect apoptotic nuclei at this time point.

Discussion

Cyclin-dependent kinases play a critical role in cell cycle regulation. Since deregulation of cdk activity appears to be a consistent feature of malignant transformation, pharmacologically induced cdk inhibition is thought to be promising for the treatment of cancer. However, not all cyclin-dependent kinases are involved in cell cycle regulation (Fischer and Lane, 2000). Cdk 5, for example, may play an important role in neuronal development and in the maintenance of neuronal cell function (Nikolic et al., 1996). More recently cdk5 has also been shown to be a key regulator of phototransduction in the retina (Hayashi et al., 2000; Matsuura et al., 2000).

In the present article, we report the unexpected finding of drug-induced retinal and peripheral nerve toxicity in outbred CD-1 albino mice treated with AG-012986, a pan-cdk_i. To our knowledge no ocular or peripheral nerve toxicity has been previously reported in pre-clinical or clinical studies secondary to the administration of cdk_is. The pathogenesis of the ocular lesions (retinal degeneration/atrophy) is unclear and investigative studies to elucidate its cause are warranted.

Interestingly, peripheral nerve but no eye lesions were detected in high-dose B6 mice (50 mg/kg) sacrificed after a 21-day drug-free recovery period. The presence of melanin in the eyes of these mice may have played a protective role against AG-012986-induced retinal injury. The mechanisms by which melanin neutralizes the toxic effects of AG-012986 are being currently investigated. It is possible that retinal toxicity in CD-1 mice may have been mechanistically driven through cdk5 inhibition since cdk5 is known to play a significant role in phototransduction (Hayashi et al., 2000; Matsuura et al., 2000) and may be implicated in some forms of retinal degeneration (Yamazaki et al., 2002). However, because these lesions have not been reported with other Cdk_is, the effects of AG-012986 are more likely a result of unidentified off-target kinases, receptors, or compound-specific metabolism/distribution qualities. Interaction of AG-012986 with other modulators of phototransduction (e.g., phosphodiesterases) has not been assessed but remains a possible explanation for the toxicities observed in the retina.

Peripheral neuropathy is a relatively common drug-induced manifestation of neurotoxicity and it is usually observed secondary to the administration of antineoplastic and antiretroviral medications (Arne-Bes, 2004). Specific chemotherapeutic agents implicated in the development of peripheral neuropathies are plant alkaloids, interferons, antimitotics, taxanes, and platinum-based compounds. Drug-induced peripheral neuropathy is usually sensory, dose-related and cumulative and often delayed, appearing weeks after initiation of therapy (Visovsky, 2003). Even though most drug-induced neuropathies are often considered to be distal (dying-back) axonopathies, the result of a disorder of axonal transport, the biochemical mechanisms underlying the presumptive impairment in axonal transport are still unclear (Masson et al., 1992; Arne-Bes, 2004).

As with the retinal degeneration/atrophy, the peripheral neuropathy observed in mice treated with AG-012986 could also be mechanistically driven by cdk5 inhibition. Cdk5 is expressed in the mature and developing central nervous system (Ino et al., 1994) and plays an essential role in neuronal differentiation (Nikolic et al., 1996), axonal transport and phosphorylation of neurofilaments in cultured neurons (Shea et al., 2004). Although it seems feasible that the unintended chemical manipulation of this key enzyme may have resulted in impaired axonal flow and the onset of peripheral neuropathy, the novelty of these findings among published Cdk_i activities makes other mechanisms of toxicity appear more likely.

Also noteworthy is the fact that peripheral nerve lesions were observed after single doses of 30 and 100 mg/kg but not after 5 days of dosing with 15 mg/kg. Additionally, the lesions observed from a single 100 mg/kg dose resulted in functional deficits noted in the clinical signs as abnormal gait, splayed limbs, and tremors, which were not observed after 5 days of dosing with 50 mg/kg. Conversely, retinal lesions were only observed after multiple days of dosing at 50 mg/kg and not after a single 100 mg/kg dose. These observations indicate that the nerve lesions were driven by maximum plasma levels while the retinal toxicity was a result of total exposure. This difference in the pharmacokinetic relationship of the two pathologies suggests the possibility of separate mechanistic pathways.

The ocular and peripheral nerve findings in AG-012986 treated CD-1 mice underscore the risks and challenges associated with drug development in general, and the therapeutic use of active site-directed kinase inhibitors in particular. The challenge of designing molecules selective for a small family of kinases within a protein family of >500 members is significant, especially given the limited ability to assess inhibitory potency against all kinase sub-families. In addition to their potential beneficial effects on tumor growth and progression, multi-targeted Cdk_is could inhibit cdks and other kinases involved in a wide range of cellular functions other than cell cycle regulation. The unintended targeting of these kinases or unrelated protein receptors or pathways may result in unexpected toxicities that could offset potential benefits and may hinder the clinical applications of these novel and promising therapeutic agents. The fact that these lesions have not been previously reported in associations with other pan-cdk_is indicates that the toxicities may reflect the unique profile of cdk inhibition (including cdks associated with functions other than cell cycle control), off-target kinase inhibition or receptor binding, or metabolism/distribution properties of AG-012986.