|Home | About | Journals | Submit | Contact Us | Français|
The Hedgehog (Hh) signaling pathway is an essential regulator of embryonic development and appears to play important roles in postnatal repair and cancer progression and metastasis. The teratogenic Veratrum alkaloid cyclopamine is a potent Hh antagonist and is used experimentally both in vitro and in vivo to investigate the role of Hh signaling in diverse biological processes. Here, we set out to establish an administration regimen for cyclopamine-induced teratogenicity in the mouse. The dysmorphogenic concentration of cyclopamine was determined in vitro via mouse whole-embryo culture assays to be 2.0μM. We administered cyclopamine to female C57BL/6J mice at varied doses by oral gavage, ip injection, or osmotic pump infusion and assessed toxicity and pharmacokinetic (PK) models. Bolus administration was limited by toxicity and rapid clearance. In vivo cyclopamine infusion at 160 mg/kg/day yielded a dam serum steady-state concentration of ~2μM with a corresponding amniotic fluid concentration of approximately 1.5μM. Gross facial defects were induced in 30% of cyclopamine-exposed litters, with affected embryos exhibiting cleft lip and palate. This is the first report describing the PKs and teratogenic potential of cyclopamine in the mouse and demonstrates that transient Hh signaling inhibition induces facial clefting anomalies in the mouse that mimic common human birth defects.
The Hedgehog (Hh) signaling pathway plays a critical role in embryonic development and tumorigenesis. In several animal models, Hh signaling inhibition during embryogenesis causes severe craniofacial defects including failure of cerebral fission, facial clefting, and cyclopia, a condition referred to as holoprosencephaly (HPE). In humans, HPE is estimated to occur in 1/15,000 births and most cases have an unknown etiology (reviewed in Cohen and Shiota, 2002).
The Sonic hedgehog knockout (Shh−/−) mouse exhibits a severe HPE phenotype characterized by craniofacial defects similar to those of sheep born to dams ingesting Veratrum californicum, a plant that contains cyclopamine (Chiang et al., 1996; Keeler, 1970). These teratogenic effects have been attributed to inhibition of the Hh signaling pathway by cyclopamine (Incardona et al., 1998).
In the absence of Hh ligand, its receptor, Patched (Ptc1), inhibits Smoothened (Smo) activity presumptively through a small molecule mediator (Bijlsma et al., 2006; Taipale et al., 2002). Upon Hh binding to Ptc1, inhibition of Smo is relieved, triggering a complex downstream signaling cascade that culminates in target gene activation via the Gli family of transcription factors (reviewed in Ingham and McMahon, 2001). Cyclopamine inhibits die Hh pathway by binding to and preventing the activation of Smo, preventing downstream target gene regulation (Chen et al., 2002).
Cyclopamine has been extensively used to study the role of Hh signaling in diverse biological processes. Analysis of Hh signaling in cell and tissue culture models has been aided by utilization of cyclopamine as a chemical inhibitor. These models allow for direct media-borne drug exposure with accompanying assays for pathway activity and cytoxicity. However in vivo cyclopamine exposure has been less straight forward. A myriad of in vivo mouse studies have used a wide range of dosing regimens and routes of administration including ip and sc injection, oral gavage (po), and infusion by microosmotic pump (OsP) (Table 1). While Table 1 shows only murine-based studies, cyclopamine has been used experimentally in many animal species, and to our knowledge, correlate serum concentrations have never been reported.
While different administration routes of cyclopamine presumably produce unique pharmacokinetic (PK) profiles with important experimental implications, no manuscript has described these analyses. The purpose of our study was to establish PK profiles for cyclopamine administered by po, ip and by OsP, and to assess the in vivo teratogenic potential of cyclopamine in the mouse. We measured serum cyclopamine concentrations to generate PK parameters while monitoring for overt toxicity with each administration route and dose. We used an in vitro mouse whole-embryo culture model of cyclopamine-induced HPE (Nagase et al., 2005) to establish a dysmorphogenic concentration of cyclopamine. Using our PK profiles, we endeavored to mimic these in vitro conditions in vivo and assess the teratogenic potential of cyclopamine in the mouse.
All animal procedures were done in accordance with the University of Wisconsin animal care guidelines. PK studies utilized naive female C57BL/6J mice at 12–16 weeks of age. For embryonic studies, timed pregnancies were established in-house. Three female C57BL/6J mice at 12–16 weeks were housed with one male C57BL/6J mouse overnight on a light cycle of 16 h on and 8 h off. The presence of a vaginal plug the following morning was considered embryonic day 0.5.(E0.5). Following cyclopamine administration, animals were regularly monitored for signs of overt toxicity including lethargy and dystonia. Mice exhibiting these signs were immediately euthanized by anesthesia followed by cervical dislocation and assigned “toxicity” in Table 2.
Cyclopamine was the kind gift of Infinity Pharmaceuticals (Cambridge, MA) and was dissolved in a sodium phosphate/citrate buffer (pH = 3) containing 2-hydropropyl-β-cyclodextrin (HPBCD) (Sigma-Aldrich, St. Louis, MO). Jervine was purchased from ChromaDex (Irvine, CA) and Toronto Research Chemicals (North York, Ontario).
Cyclopamine was administered by single ip injection at 10, 50, and 160 mg/kg or po at l0 or 50 mg/kg in 10% HPBCD (wt/wt) solution in a volume of 500 μl. We utilized sc-implanted Alzet microosmotic pumps (Durect, Cupertino, CA) for cyclopamine infusion. Cyclopamine was administered by three pump models: model 1007D (100 μl volume dispensed at 0.5 μl/h for 208 h), model 1003D (100 μl 1.0 μl/h for 102 h), and model 2001D (200 μl at 8.0 μl/h for 31 h). Pumps were filled with 1.5 mg of cyclopamine/100 μl 30% HPBCD (wt/wt) to achieve approximate corresponding dispensation rates of 10, 20, and 160 mg/kg/day.
At given time points ~50 μl of blood was collected via maxillary vein sampling. Following centrifugation at 2040 × g for 20 min, serum was collected and stored at −20°C. Two to three samples were collected per mouse at time points that were empirically selected based upon the route and duration of drug administration.
Dams were euthanized by isofluorane anesthesia followed by cervical dislocation at 9.25 days of gestation. Embryos encased in deciduas and uterine muscle were removed and placed in PBS. A small section of uterine muscle was removed to expose decidual tissue. Amniotic fluid was then collected with a syringe attached to a 28.5-gauge needle placed through the opening in the uterine muscle through decidua, visceral yolk sac, and amnion. Extracted fluid was centrifuged at 2040 × g for 20 min and the supernatant was stored at −20°C.
Cyclopamine was assayed by high performance liquid chromatography-mass spectrometry using a protocol modified from the one developed by Infinity Pharmaceuticals. Briefly, serum samples (10 μl) were diluted with 65 μl water containing 0.1% formic acid and 0.1% dimethyl sulfoxide. After adding 150 μl of methanol containing 4.5 ng jervine, an internal standard, samples were centrifuged through a 0.2-μm Nylon filter (Costar Spin-X HPLC microcentrifuge filters, Corning Inc., Corning, NY). The filtrate was diluted 1:1 with water prior to analysis. Analysis was performed using an Agilent 1100 Series LC/MSD quadrupole mass spectrometer system (Agilent Technologies, Santa Clara, CA) fitted with 3.5 μm particle size Symmetry C18 columns (2.1 × 20 mm analytical and 2.1 × 10 mm guard; Waters Corp., Milford, MA). Samples (20 μl) were injected with a flow rate of 1 ml/min in 5% solvent B (solvent A: water/0.1% formic acid [vol/vol]; solvent B: acetonitrile/0.1% formic acid [vol/vol]). Analytes were eluted from the column after a 1-min hold (5% B) with a linear 5 min gradient to 95% B. After a 1-min hold at 95% B, the solvent was returned to 5% B in 0.5 min and the column re-equilibrated for 2.5 min before the next injection. The optimized electrospray source conditions were 2500V capillary, 120V fragmentor, 50 psi nebulizing gas (N2) at 121/min with a source temperature of 350°C. Analyte ions (412.3 cyclopamine, 426.3 jervine) were detected by single ion monitoring in positive ion mode. Standards ranged from 2 to 2000 pg per injection with an interday variation of 7.6% and intraday variation of 7.0%. The lower limit of quantitation (LLOQ) was 6.32 pg per injection or 0.035μM. Values occurring within 24 h of the last dose but below the LLOQ and above background were given a value of 0.01μM. Values at or below background were given a null value.
Cyclopamine concentrations from all animals were processed simultaneously with a nonlinear, mixed-effects modeling program (NONMEM v6 with Wings for NONMEM v6 Icon development solutions, Ellicott City, MD) (Beal, 2006). Population analysis did not include evaluation of covariates other than route and dose. The computer fitting of all data included assessment of one and two compartments, induction of elimination, nonlinear elimination, and variable bioavailability by route. Bioavailability for the ip bolus was defined as 100%. Confirmation of the optimal model was made by inspection of the residual plots, minimization of the objective function, and a 1000-iteration nonparametric bootstrap analysis (Parke et al., 1999).
Dams were euthanized at 8.5 days of gestation. Embryos were removed from the uterus and placed in Tyode’s salts (9.6 g/l in distilled water) (Sigma) supplemented with 1.0 g/1 sodium bicarbonate. To separate individual embryos, uterine muscle was cut and removed. Subsequently, decidual tissue was carefully removed and Reichert’s membrane was trimmed back to the ectoplacental cone. Embryos were cultured in 70% rat serum (Harlan Bioproducts for Science, Indianapolis, IN), 30% Tyrode’s salts supplemented with 1.0 g/l sodium bicarbonate with 100 U Penicillin/Streptomycin. Cyclopamine was dissolved in HPBCD (≤0.2% in solution), yielding working concentrations of 1.0, 2.0, 5.0, and 20.0μM. Up to five embryos were cultured in 5-ml media per 30-ml glass vial with silicon rubber stopper at 37°C in a rotisserie incubator rotating at 32 rpm. At 0 and 8 h of culture, vials were gassed with a 5% O2, 5% CO2, and 90% N2 mixture. At 24 and32 h, vials were gassed with a 20% O2, 5% CO2, and 75% N2 mixture. Also at 24 h, to compensate for embryonic utilization, 50 mg/dl glucose was added to culture media. Following culture, embryos were fixed in 10% formalin and mounted in 5% methylcellulose for imaging.
Embryos at E16.5 were euthanized and initially fixed in 70% ethanol and visually examined for defects under a dissection microscope. For images shown in Figure 4, embryos were briefly soaked in PBS with ethidium bromide and imaged under ultraviolet light. Images were subsequently converted to grayscale.
Administration by ip injection caused minimal toxicity at doses of 10 mg/kg and 50 mg/kg (Table 2), while the highest dose tested (160 mg/kg) demonstrated toxicity (n = 4/4), preventing concentration measurements. Maximum serum concentration (Cmax) was reached 20 min (Tmax) following injection of 10 or 50 mg/kg doses with corresponding Cmax values of 2.58 and 19.9μM , respectively (Fig. lA).
Administration by po caused toxicity only at 50 mg/kg (n = 4/6). Relative to ip injection, Cmax values following administration of 10 and 50 mg/kg were considerably lower (0.39 and 2.07μM) and temporally delayed with a Tmax of 1.9 h (Fig. 1B). Circulating concentrations 24 h following bolus administration of cyclopamine were approaching or below our LLOQ (0.035μM).
The toxicity and rapid clearance associated with ip and po cyclopamine administration are significant hindrances to experimental application. To circumvent these problems, we simulated drug infusion via OsP surgically implanted sc.
Cyclopamine administration by OsP caused limited toxicity at the infusion rate of 160 mg/kg/day (n = 6/24). Increasing the infusion rate by 50% to 240 mg/kg/day caused consistent toxicity (n = 3/3). The sc infusion at 10, 20, and 160 mg/kg/day yielded semm steady-state concentrations (Css) of 0.093, 0.38, and 1.92μM, respectively (Table 2; Fig. 1C).
Long term in vivo cyclopamine exposure has more often been achieved by repeated bolus administrations at given intervals than OsP infusion (Table 1). To model this approach, we administered 10 mg/kg/day by daily ip injection. By assaying serum cyclopamine concentrations before and after subsequent injections, we found that circulating levels did not change with repeated bolus doses (Fig. 1D).
To our knowledge, transplacental transfer of cyclopamine has not been reported. To determine the amniotic concentration of cyclopamine relative to dam serum, we implanted OsPs dispensing 160 mg/kg/day for 31 h into pregnant dams at E8.25. Analysis of dam serum 24 h after pump implantation demonstrated a mean cyclopamine concentration of 2.26μM, whereas the mean concentration in amniotic fluid of exposed embryos was 1.45μM (Fig. 2).
Nagase et al. (2005) found that whole-mouse embryos cultured in vitro in the presence of cyclopamine between embryonic days 8.5 (E8.5) and E10.5 exhibited an embryonic phenotype of HPE correlated with markedly reduced expression of reliable Hh pathway activity indicators Ptcl and Glil. We used this model as it allows for direct media-borne drug exposure, to determine the concentration of cyclopamine required to induce a HPE-like phenotype in vitro. Mouse embryos exposed to HPBCD carrier alone achieved facial development (Fig. 3A) similar to embryos grown in vivo (data not shown). However, embryos exposed to 20μM cyclopamine exhibited a marked failure of mediolateral and proximodistal expansion of the frontonasal prominence (FNP); an embryonic phenotype consistent with HPE and mirroring the findings of Nagase et al. (2006) (Fig. 3B).
We found that this phenotype could be precisely measured using the distance between the lateral edges of the FNP relative to width of the midbrain (MB), which was unaffected by treatment. This calculation yielded a frontonasal prominence to midbrain (FNP/MB) ratio. Using a similar approach, Higashiyama et al. (2007) found a significant reduction of FNP width in ethanol-exposed holoprosencephalic mouse embryos compared to vehicle-exposed embryos. We found that vehicle-exposed embryos grown in vitro were not significantly different than those developed in vivo, while embryos exposed to cyclopamine demonstrated a concentration-dependent reduction in FNP/MB ratio where the lowest observable effect level (LOEL) was 2.0μM (Fig. 3C).
To assess in vivo teratogenic potential, OsPs infusing cyclopamine at 160 mg/kg/day for 31 h were implanted at E8.25 to achieve a Css concurrent with development at E8.5. Gross facial defects were present in three of 10 cyclopamine-exposed litters with an intralitter penetrance of ~50% (n = 9/19 embryos). Affected embryos were slightly smaller than normal littermates and exhibited mild blunting of the snout as well as cleft lip and palate (CL/P) (Fig. 4). Embryos exhibited unilateral (2/9) (Fig. 4D) and bilateral complete cleft lip (7/9) (Fig. 4E) with clefts extending into the primary and secondary palate (Fig. 4F). Facial clefts were often accompanied by open eyelid defects and in one embryo by forelimb syndactyly (data not shown). No gross defects were seen in vehicle-exposed litters (n = 8) (Fig. 4A–C), and there was no apparent difference in litter size or resorption rate between treatments (data not shown).
Cyclopamine concentrations from all animals were processed simultaneously with a nonlinear, mixed-effects modeling program. Standard approaches to the use of NONMEM for population pharmacokinetic (PopPK) modeling led to the decision to accept a two-compartment model to describe the pharmacokinetics of cyclopamine. The terminal elimination half-life of cyclopamine administered ip or po was 4 h (Table 3). The fit to a one-compartment model was generally acceptable but failed to predict the high concentrations seen after ip injection. A mixed-order (Micaelis-Menten) model did not improve the fit for the basic model. Figure 1 shows the sampled concentrations for each route and dose. Each line plot represents the predicted result for that dose based upon the mean of the parameter’s value for the population.
Construction of the error model for cyclopamine PopPK led to improvements in fit and decreased objective function values when intersubject variation (exponential model, eη) was applied to both clearance (CL) and central distribution volume (Vc). Intersubject variability was also evaluated for bioavailability (F), absorption rate (Ka), distribution clearance (Q), and tissue volume (Vt). Data rounding errors indicated that these additions to the error model were unnecessary and did not significantly improve the fit of observed data. A proportional intrasubject error model proved to be superior to both the additive and proportional + additive options. Confirmation of the stability of the PK parameters was tested by a 1000-iteration nonparametric bootstrap.
In vivo cyclopamine exposure has been commonly used to inhibit Hh signaling in murine models of development and disease. However, administration routes widely vary and a comparison of these approaches has not been published. Here, incorporating data from three administration routes, we characterized a two-compartment PK model yielding route- and dose-specific toxicity and PK profiles.
We found that administration by ip injection and oral gavage was limited by toxicity and a relatively short elimination half-life. Administration by oral gavage was particularly suboptimal given associated toxicity at higher concentrations and limited bioavailability (0.33 relative to ip administration). The toxicity associated with ip injection manifested as severe dystonia within 2 h of administration while toxicity associated with po and OsP administration manifested as lethargy at least 6 h after administration. The data, presented in Table 2 suggests that toxicity of was not a direct function Cmax or area under curve (AUC) values.
Visual inspection of the oral and ip bolus plots in Figure 1 may suggest that the terminal elimination slope should be shallower, reflecting a longer elimination half-life. These late, sustained concentrations may in fact be outliers, as most serum cyclopamine concentrations 24 h after the bolus were below the LLOQ.
As opposed to bolus dosing, OsP infusion yielded sustained concentrations without high, transient peak concentrations. However, detailed analysis yielded somewhat unexpected results. OsP infusion of 20 and 160 mg/kg/day yielded Css levels and AUC values proportional to dose, while values from 10 mg/kg/day infusion was less than proportional (Table 2) due to increased clearance. Further, there appears to be a gradual decline in the measured concentrations of cyclopamine over time, particularly notable in the 101-h infusion (Fig. 1C). It is not known whether this is a real effect, and if so if it is due to a decrease in pump efflux, degradation or adsorption of drug within the pump, or perhaps enzyme induction. Enzyme induction does not appear likely as there is no evidence of decreasing concentrations during tile daily × five ip injections. Similarly, with an elimination half-life of 4 h, there was no significant accumulation of drug in serum with daily ip injections.
Overall, however, OsP infusion yielded mostly stable concentrations and generally circumvented the toxicity of ip and po administration likely caused by large initial concentration spikes. Moreover, periods of sustained concentrations may be better suited for certain studies as opposed to intermittent concentration spikes produced by bolus administration (Figs. 1C and ID).
Here, we set out to establish an administration regimen for cyclopamine-induced teratogenicity in the mouse. We found that in vivo infusion of 160 mg/kg/day yielded amniotic concentrations of ~1.5μM (Fig. 2), while the in vitro LOEL for reduction of FNP expansion was 2.0μM (Fig. 3C). Achieved in vivo concentrations approximating the in vitro LOEL may explain the incomplete inter- and intralitter penetrance of teratogenicity. Inter- and intralitter variations in embryonic staging may also contribute to the incomplete penetrance. In affected litters, overall body size of embryos with gross defects was slightly smaller than unaffected littermate embryos. It is unclear whether this reflects increased sensitivity of earlier staged embryos within a litter at time of exposure or if it is a subsequent effect of cyclopamine exposure in sensitive embryos.
While the Veratrum alkaloids, cyclopamine (11-deoxyjervine) and jervine, have been considered “universal” teratogens, the mouse appears to be somewhat resistant. Comparing the interspecies teratogenic activity of cyclopamine and jervine, Keeler (1975) found that while golden hamster fetuses were extremely sensitive, rats were less sensitive and Swiss Webster mice were apparently resistant to both related compounds.
The cyclopamine-induced facial defects described here mimic those reported by Omnell et al. (1990) who found that of 20 litters of C57BL/6J dams exposed to 70 mg/kg jervine via po at E8.5, 53 of 126 living fetuses were abnormal, predominately displaying cleft lip/palate often accompanied by open eyelid and limb and digit defects. We attempted to precisely replicate these experimental conditions but did not find defects in a total of 22 fetuses from four surviving litters (Lipinski, unpublished data). However, this sample size may have been insufficient given the low frequency of cyclopamine-induced defects produced by the cyclopamine administration regimen described here.
Omnell also reported an approximate LD50 concentration for jervine of 120 mg/kg, and Keeler (1975) found that teratogenic doses of cyclopamine caused varying lethality in hamsters. These data, along with our finding that while cyclopamine infusion at 160 mg/kg/day caused teratogenicity, infusion at 240 mg/kg/day caused dam toxicity, suggests that the concentration required for Veratrum alkaloid-induced terato-genicity approaches dam-toxic concentrations.
The mouse serves as a valuable tool to study craniofacial development because of the high fidelity between embryonic development of the face in mouse and man (Diewert and Wang, 1992). The methodology for teratogen-induced CL/P described here provides an additional model, which along with genetic mouse models (reviewed in Juriloff and Harris, 2008) provide tools to study the morphological processes underlying CL/P. However, the low frequency of affected embryos by the presented administration methodology mitigates its practical utility. The mouse strain utilized in this study appears particularly resistant as spontaneous CL/P is not seen in C57BL/6J colonies. Conversely, some mouse strains demonstrate reliable spontaneous frequencies of CL/P, such as A strain (2-24%) and CL/Fr (36%) (Juriloff and Harris, 2008; Millicovsky et al., 1982; Trasler and Marchado, 1979). In fact, using embryo transfer techniques, Martin et al. (1995) found that the frequency and severity of CL/P was significantly reduced in CL/Fr blastocysts implanted into C57BL/6J dams compared to donor strain dams. Given these observations, exploration of relative strain sensitivity to cyclopamine-induced CL/P may yield a more tractable model.
CL/P arises from failed fusion of the median nasal prominence with the maxillary prominences. Several factors may contribute, including failed growth of the FNP effectively preventing or delaying contact of the prominences, causing failed or deficient fusion (Juliloff and Hmis, 2008). Our in vitro whole-mouse embryo assays demonstrated that exposure of 2μM cyclopamine caused a subtle but significant decrease in the mediolateral expansion of the FNP, providing a likely mechanism for the CL/P defects presented by embryos exposed in vivo.
Shh expression in the neuroectoderm is required for induction of Hh signaling in the adjacent face and for expansion of the FNP in chick (Marcucio et al., 2005). Hh signaling blockade following establishment of Shh in the forebrain but prior to its induction in the face results in facial defects without detectable effects on the forebrain (Cordero et al., 2004). Similarly, the findings here demonstrate that chemical inhibition temporally targeting Hh signaling during FNP expansion induces isolated facial clefting in the mouse that phenotypically mimic human anomalies.
While HPE is a rare clinical occurrence (~1/15,000 live births), non-syndromic CL/P is much more common (~1/700 live births). The etiological bases for CL/P in humans appear complex cmd multifactorial, likely involving genetic and environmental factors (reviewed in Murray, 2002). The finding here that tsansient inhibition of Hh signaling induces CL/P in mice is significant given recent findings that numerous structurally diverse small molecules inhibit Hh signaling with varying potencies (Chen et al., 2002; Frank Kamenetsky et al., 2002, Lipinski et al., 2007; Williams et al., 2003). Taken together, these findings argue tllat further efforts to identify and characterize Hh signaling inhibitors of human exposure may provide important insights into the underlying etiology of cleft lip/palate, one of the most common and morbid human birth defects.
The authors are deeply grateful to Infinity Pharmaceuticals for their gift of cyclopamine and sharing of HBPCD carrier and cyclopamine detection methodology. We thank Michael Bentz and Jill Helms for thoughtful discussion and Ruth Sullivan for technical assistance and critical review of the manuscript. We also thank Steven Attia, Travis Jerde and Jerry Gipp for critical review of the manuscript and Lisa Krugner-Higby for technical assistance. Mass spectrometry was performed at the University of Wisconsin School of Pharmacy Analytical Instrumentation Center.
FUNDING National Institutes of Health (DK065303-03); National Institute of Environmental Health Sciences (T32-ES00715 to R.L.); National Institute of Environmental Health Sciences (T32-ES07295 to P.W.H. and R.N.)