Teratogenicity and Fetal-Transfer Assessment of the Retinoid X Receptor Agonist Bexarotene

Abstract

Bexarotene, a retinoid X receptor (RXR) agonist, is used to treat cutaneous T-cell lymphoma, and drug repositioning research has also been reported, despite warnings of teratogenicity. However, fetal transfer of bexarotene and its effect on rat fetal bone formation have not been examined. In this study, we conducted a detailed teratogenicity and fetal transferability assessment of bexarotene in rats. Repeated administration of bexarotene during pregnancy caused marked fetal atrophy and bone dysplasia. Although fetal transfer was not detectable by dynamic imaging of [¹¹C]bexarotene by means of positron emission tomography, transfer to the fetus was confirmed by using a gamma counter. Similar levels were found in mother and fetus. In addition, we found that bexarotene was accumulated in the placenta. These findings will be useful for the toxicity assessment of bexarotene as well as for drug discovery research targeting RXR agonists, which are expected to have therapeutic effects in various diseases.

Bexarotene (Targretin, Figure 1) has been used since 1999 as a therapeutic agent for cutaneous T-cell lymphoma (CTCL),^1,2 which is a rare type of lymphoma associated with overgrowth of T-cells in the skin, and affects 0.3 to 0.9 people per 100 thousand annually.³ Bexarotene induces apoptosis and suppresses proliferation of tumor cells by selectively binding to retinoid X receptors (RXRs) in the nucleus and activating transcriptional activity.⁴ RXRs function by forming homodimers as well as by forming heterodimers with other nuclear receptors.^5,6 Heterodimer partners of RXRs include liver X receptor (LXR), thyroid hormone receptor (TR), retinoic acid receptor (RAR), peroxisome proliferator-activated receptor (PPAR), and Nurr1. Consequently, RXR ligands such as bexarotene have a wide variety of physiological activities, and drug repositioning studies targeting various diseases have been conducted. Examples include type 2 diabetes,⁷ central nervous system diseases such as Alzheimer’s disease and Parkinson’s disease,^8,9 inflammatory bowel disease,¹⁰ and COVID-19.¹¹ However, bexarotene also causes serious adverse events, such as hypothyroidism¹² and hypertriglyceridemia.¹³ In addition, teratogenicity has been reported in a zebrafish model,^14,15 and bexarotene is contraindicated for administration to pregnant women. This is a major reason why drug-repositioning studies of RXR ligands have recently been attracting less interest.

Figure 1.

Chemical structures of bexarotene and [¹¹C]bexarotene.

Teratogenicity refers to the property of causing irreversible morphological abnormalities in developing embryos and fetuses. The incidence of congenital neonatal deficiency in humans is about 2–3%, and it is thought that pharmaceuticals are the cause of about 1–3% of this number.¹⁶ This is not a large percentage, but the Food and Drug Administration of the United States (FDA) classifies drugs that may be administered to pregnant women by teratogenic risk.¹⁷ One of the factors contributing to teratogenicity is the low levels of fetal plasma proteins and drug-metabolizing enzymes so that drugs transferred to the fetus via the placenta may have a greater-than-expected effect. In this context, drug transfer to the fetus may occur by passive diffusion or active transport via transporters, which may mediate influx or efflux transport.

The fetal transferability of pharmaceuticals is generally evaluated by systemic autoradiography (ARG) or comparison of maternal and fetal blood levels of radioactivity after administration of drugs labeled with radioisotopes such as carbon-14.^18,19 Although this method is quantitative, it cannot image the fetal-transfer dynamics. To address this, we have employed positron emission tomography (PET) imaging of carbon-11-labeled propionic acid produced by the intestinal flora of pregnant rats.²⁰ If this PET method could be applied to the evaluation of fetal transferability of pharmaceuticals, it would not only allow imaging of the dynamics of fetal transfer but also contribute to reducing the number of experimental animals required.

The FDA data for bexarotene describe fetal appearance abnormalities observed in a toxicity test using rats, but no details of skeletal abnormalities have been reported.²¹ In addition, the fetal transferability of bexarotene has not been clarified so far. A detailed teratogenicity and fetal transferability assessment may provide useful insights for the development of next-generation bexarotene substitutes. Therefore, in this study, we investigated the teratogenicity of bexarotene by examining skeletal specimens of fetuses after administration to pregnant rats, and we also investigated the fetal transferability using [¹¹C]bexarotene.

Results and Discussion

Abnormal Fetal Morphology Caused by Bexarotene

Pregnant rats were orally administered bexarotene at 30 mg/kg/day from the 1st to the 19th day of gestation (Supporting Information Figure S1). According to the FDA, morphological abnormalities occur at the dose of 16 mg/kg,²¹ so 30 mg/kg was considered to be sufficient for teratogenicity assessment. On the 20th day of gestation, the fetuses were excised and the number of implantations, number of survivors, weight and height of fetuses, and weight of the placenta were measured (Figure 2). There were no significant differences in the numbers of implantations and survivors between the normal and bexarotene-administered groups (Figure 2B). However, the fetuses in the bexarotene-administered group were smaller than those in the normal group (Figure 2A), and their average weight and height were significantly smaller (Figure 2C,D). In addition, there was a decrease in placental weight (Figure 2E). Among the maternal organs, we found a marked decrease in uterine weight and increases in liver, kidney, and stomach weight in the bexarotene-administered group (Supporting Information Figure S2). Also, maternal serum showed significantly greater values of total bilirubin (TBIL), glutamic pyruvic transaminase (GPT), and glutamic oxaloacetate transaminase (GOT) (Supporting Information Figure S3). Bexarotene is known to cause hepatomegaly in rodents,²² and increased liver weight and abnormal serum parameters are thought to correlate with maternal liver dysfunction.

Figure 2.

Effects of administration of bexarotene (30 mg/kg, 19 days) on fetuses. (A) Photograph of fetuses and placenta. Comparison of (B) total and surviving numbers of fetuses, (C) weight of fetuses, (D) length of fetuses, and (E) weight of placenta. Data are mean ± SD. Normal, 4 mothers and 45 fetuses; bexarotene, 3 mothers and 26 fetuses. Student’s t-test vs normal group; ****, p < 0.0001; NS: not significant.

Fetal Skeletal Abnormalities Caused by Bexarotene

Fetal skeletons were prepared according to the previous study,²³ and bone dysplasia in the skull, front and back foot length, ribs, and vertebrae were evaluated. There were few hard bones in the skull (stained in red) (Figure 3A,B). In the bexarotene-administered group, bone dysplasia was observed at all points of the frontal bone (F), temporal bone (Tm), maxilla (Mx), mandible (Mn), upper occipital bone (Uo), and tympanic ring bone (Ty) (Figure 3C). In total, 19% of the normal group and 100% of the bexarotene-administered group showed bone dysplasia, indicating that bexarotene causes abnormalities in skull formation.

Figure 3.

(A,B) Photographs of skull. F, frontal bone; Tm, temporal bone; Mx, maxilla; Mn, mandible; Uo, upper occipital bone; Ty, tympanic ring bone. (C) Comparison of rates of skull dysplasia between the normal and bexarotene-administered groups. N = 11–16.

To evaluate the forefoot and hindfoot, we measured the lengths of the ulna (U) and tibia (T), respectively, which are both stained as hard bones (Figure 4A,B). Shortening was observed in the bexarotene-administered group. The lengths of the ulna and tibia of the left leg per whole body height in the bexarotene-administered group were significantly shorter than those in the normal group (Figure 4C,D).

Figure 4.

(A,B) Photographs of limbs. U, ulna; T, tibia. Comparison of the length ratio of (C) left ulna and (D) left tibia per body length of the normal and bexarotene-administered groups. Data are mean ± SD (N = 10–16). Student’s t-test vs normal group, *, p < 0.1; **, p < 0.01.

Focusing on the ribs, the normal group had regular stiff bones, while the bexarotene-administered group showed abnormalities such as fusion (F), short (S), wavy (W), branching (B), and trace (T) structures (Figure 5A,B). There was no significant difference in the number of ribs between the two groups, but the rate of various abnormalities was higher in the bexarotene-administered group than in the vehicle control group (Figure 5C). Indeed, the bexarotene-administered group showed five times more abnormalities than the normal group, indicating that bexarotene is teratogenic.

Figure 5.

(A,B) Photographs of ribs. F, fusion; S, short; W, wavy; B, branching; T, trace. (C) Comparison of rates of rib dysplasia between normal and bexarotene-administered groups (N = 11–16).

The vertebrae are divided into cervical, thoracic, and lumbar, and in this study, we focused on the thoracic and lumbar vertebrae (Supporting Information Figure S4). The vertebrae consist of spines (Sp) and transverse (Tr) processes. When bone dysplasia in the thoracic and lumbar vertebrae was examined, separation, shortness, fusion, and insufficient formation were confirmed in the bexarotene-administered group (Supporting Information Figure S4). Thus, repeated administration of bexarotene caused marked fetal atrophy and bone dysplasia.

Fetal Transferability of Bexarotene

Since bexarotene was confirmed to be teratogenic, we next evaluated its fetal transferability. In order to visualize the dynamics of fetal transfer, we employed PET imaging. Radioactive nuclides such as carbon-11 (¹¹C) or fluorine-18 (¹⁸F) can be used for quantitative, non-invasive PET imaging²⁴ but have a short half-life (¹¹C; 20 min, ¹⁸F; 108 min), making them difficult to synthesize and handle. Therefore, we selected ¹¹C as a label. Synthetic methods in which ¹¹C is introduced into the carboxylic acid site have been reported for RXR agonists such as bexarotene (see Table S1). Rotstein et al. reported the synthesis of [¹¹C]bexarotene by the introduction of [¹¹C]carboxylic acid in the presence of a copper catalyst using a borate ester substituent as a precursor.²⁵ However, the copper catalyst used in this method forms a chelate with the target product, and the reaction was carried out at a high temperature. We succeeded in introducing [¹¹C]carboxylic acid into the RXR agonist CBt-PMN using a trimethyltin precursor at low temperature,^26,27 and in 2020, the synthesis of [¹¹C]bexarotene using a tributyltin precursor and copper catalyst at high temperature was also reported.²⁸ In this study, we aimed to synthesize [¹¹C]bexarotene at low temperature using a tributyltin precursor in the presence of a catalytic amount of tetramethylethylenediamine (TMEDA) to activate n-BuLi, with [¹¹C]CO₂ gas as an ¹¹C-source. The tributyltin precursor was synthesized as shown in Supporting Information Scheme S1. In a COLD run using non-radioactive CO₂ gas, we found that addition of TMEDA resulted in a 4-fold improvement of the product yield based on the HPLC peak area ratio of the precursor and the target product (Supporting Information Scheme S2, Figure S5). Therefore, we next performed a HOT run using [¹¹C]CO₂ gas in the presence of TMEDA. However, the target product was not formed (Supporting Information Scheme S2, Entry 3). We considered that the acidification by 0.5% formic acid might be insufficient based on the amount of TMEDA added, and indeed, a 5-fold greater amount of formic acid resulted in the formation of the ¹¹C-labeled tracer with a radiochemical yield of 0.04% and a radiochemical purity of 99.8% (Supporting Information Scheme S2, entry 4, Figure S6B–D). Radiochemical purity was confirmed by comparison of the HPLC retention time and TLC behavior with those of non-radioactive bexarotene as the short half-life of ¹¹C requires rapid analysis. Interestingly, the product was obtained in almost the same yield when the amount of TMEDA added was halved (Supporting Information Scheme S2, entry 5). Since we have previously reported a method using p-toluenesulfonic acid (pTSA),²⁹ we tried the reaction using pTSA, but the target product was not obtained (Supporting Information Scheme S2, entry 6). Changing the reaction temperature to room temperature was ineffective (Supporting Information Scheme S7).

The obtained [¹¹C]bexarotene was intravenously administered to pregnant rats on the 16th to 17th day of gestation, and PET imaging was performed for 60 min. Time-dependent imaging of the migration of [¹¹C]bexarotene was attempted by superimposing the PET images on computed tomography (CT) images, but radiation could not be detected in the fetus at the injected dose (Supporting Information Figure S7A). Therefore, regions of interest (ROIs) were set based on the PET/CT images, and time-dependent radiation accumulation per volume was plotted (Supporting Information Figure S7B). The results showed a large accumulation in the liver, followed by the heart (reflecting radioactivity in the blood). On the other hand, the accumulations in the uterus, kidney, and muscles were extremely small. These results indicated that time-dependent fetal-transfer imaging by PET would not be feasible. Instead, after PET imaging, blood was collected under anesthesia, and the uterus and other organs were removed. The radioactivity in fetus and placenta was measured with a gamma counter (Figures 6, Supporting Information Figure S8). We found that the level of radioactivity in the fetus was the same as that in the maternal blood, and significantly higher radioactivity was detected in the placenta (Figure 6A). As already mentioned, the maternal blood and fetus are not in direct contact with each other, and these results indicate that although bexarotene is transferred to the fetus, it is accumulated in the placenta. Maternal blood passes through the endometrial arteries and into the villous space, where nutrients, oxygen, drugs, and so forth are transferred through the blood-placenta barrier to fetal blood by passive diffusion or active transport (Figure 6B). Thus, the presence of significantly higher radioactivity in the placenta than in maternal blood may be a consequence of non-specific binding to cell membranes during the fetal-transfer process due to the high lipophilicity of bexarotene.

Figure 6.

(A) Radioactivity in fetus and placenta, determined with a gamma counter. Data are mean ± SD (N = 3). Bonferroni test; *, p < 0.1; NS: not significant. (B) Schematic diagram of the placental structure and fetal migration of bexarotene.

The teratogenicity of bexarotene has been reported in the zebrafish embryo model,¹⁵ but the mechanism has not been established, although one possibility would be activation of RAR heterodimers involved in cell differentiation. Indeed, Jiang et al. examined the teratogenicity of RXR ligands in a chondrocyte differentiation system using mouse embryonic limb bud cells and found a correlation between the RAR activity and teratogenicity of the RXR ligands.³⁰ All-trans retinoic acid (ATRA) and 9-cis retinoic acid (9cisRA) are both natural RXR ligands, but 9cisRA has been reported to show weaker teratogenicity than ATRA in pregnant mice.³¹ This can be explained by the different fetal transferability of the two compounds. Although bexarotene is considered to be an RXR-selective agonist, it also shows weak activity toward RAR homodimers.³² Therefore, we examined RAR/RXR heterodimer activation by bexarotene in all three subtypes of RAR, compared with that of ATRA (Figure 7). Our results, together with the previous findings, suggest that the fetal transferability and RAR activity of bexarotene contribute to its teratogenicity. As regards fetal transferability, assuming that permeation through the blood-placental barrier is associated with the high lipophilicity of bexarotene, RXR agonists with less lipophilicity, such as Wy14,643,³³ oxaprozin derivatives,³⁴ NEt–3IB,³⁵ and CATF-PMN,³⁶ may exhibit reduced teratogenicity. In recent years, an in vitro model has been developed to evaluate blood-placental barrier permeability.³⁷ Thus, a twofold strategy of suppression of RAR activity and reduction of the blood-placental barrier permeability may contribute to the development of less teratogenic RXR ligands in drug discovery research.

Figure 7.

Relative transcriptional activities of bexarotene toward RAR/RXR heterodimers vs ATRA (1 μM). Red, green, and blue represent activities toward RARα/RXRα, RARβ/RXRα, and RARγ/RXRα, respectively.

Conclusions

In this study, we examined the teratogenicity of the RXR agonist bexarotene by investigating bone dysplasia in fetal skeletal specimens. Repeated administration of bexarotene for 19 days caused fetal atrophy as well as bone dysplasia, including insufficient bone formation, fusion, and curvature. Also, we first developed an improved synthetic method for [¹¹C]bexarotene at low temperature. We could not image the dynamics of fetal transfer of [¹¹C]bexarotene by means of PET, but transfer to the fetus was confirmed using a gamma counter. The levels of radioactivity were similar in maternal and fetal blood, but higher levels of radioactivity were accumulated in the placenta. These results will be helpful for the toxicity evaluation of bexarotene as well as in drug discovery to develop new RXR agonists.

Materials and Methods

In Vivo Studies

Animal preparation. All animal experiments were performed in accordance with institutional guidelines, and the protocol was approved by the Animal Research Committee of Okayama University. Eight-week-old male/female Wistar and SD rats (200–250 g) were used for the repeated oral administration and PET experiments, respectively.

Preparation of Pregnant Rats

After 1 week of acclimation of female rats, vaginal smears were taken with a cotton swab soaked in saline and cells were stained with Giemsa (Sigma-Aldrich). Female rats confirmed to be in estrus were housed with males overnight. For animals with a vaginal plug, the next day was set as the 0th day of pregnancy. Pregnant animals were housed individually.

Repeated Oral Administration of Bexarotene to Pregnant Rats

Bexarotene was synthesized according to the previous report.⁴ Rats were assigned to two groups with matching body weights on day 0. Bexarotene (N = 3) was orally administered to one group at 30 mg/kg/day (in 0.5 w/v % CMC solution) from day 0 to day 19. The normal control group (N = 4) was given 0.5 w/v % CMC solution alone. On day 20, blood was collected from the abdominal aorta with a 20G needle under isoflurane anesthesia, and the uterus, ovary, stomach, thymus, liver, kidney, and spleen were removed and weighed. The fetus and placenta were separated, and the weight and height of each fetus and the weight of the placenta were measured. Serum was obtained by centrifugation, and various parameters were measured using a Fuji Dry Chem system (Dry Chem 4000 V, Fuji Medical Co.).

Preparation of Skeletal Specimens

This experiment was performed according to the previous report.²³ Fetuses were fixed in ethanol for 2 weeks; then the skin was peeled off and the internal organs were removed. Ethanol was removed, and the fetus was soaked in Alcian Blue stain solution (Alcian Blue solution/acetic acid/ethanol = 2:3:7) for 3–4 h. The Alcian Blue solution was then removed, and the fetus was soaked in a series of 95% ethanol, 75% ethanol, 40% ethanol, 15% ethanol, and H₂O. After 24 h, the fetus was soaked in trypsin solution for transparency (trypsin in sat. sodium tetraborate aqueous solution: H₂O = 3:7) at 35 °C for 5 h. Then, Alizarin Red stain solution (Alizarin Red stain in 0.5% KOH aq) was added. After 24 h, the stain solution was removed and the fetus was soaked successively in 0.5% KOH aq, 20% glycerol in 0.5% KOH aq, and 50% glycerol in 0.5% KOH aq. The resulting skeletal specimens were stored in a refrigerator.

Small Animal PET and CT Imaging

This experiment was performed according to the reported method.²⁰ For intravenous injection studies, a 29-gauge needle connected to a PE 10 catheter was inserted into a lateral tail vein. All PET scans were performed using a ClairvivoPET (Shimadzu, Japan), which is designed for laboratory animals. The animals were anesthetized by inhalation of 1.5–2% isoflurane in room air, and their body temperature was maintained at 36 °C with a heating pad. A tracer (0.1–10 MBq, 0.1–0.15 mL) was injected intravenously. A dynamic emission scan was acquired for 30 min in the list mode with an energy window of 400–650 keV, and frames were collected in the following manner: 5 × 60, 5 × 60, 5 × 60, 5 × 60, 20 × 60 and 20 × 300 s. CT scanning was performed using an Aquilion TSX-01A (Tokyo Medical Systems, Japan). Images were reconstructed using FORE-FBP. Regions of interest (ROIs) were drawn over the brain and heart in axial images on the basis of the corresponding CT image using PMOD 3.0 software (PMOD Technologies Inc.). Decay-corrected radioactivity was expressed as a percentage of the injected dose per milliliter of tissue (% ID/mL) or a percentage of the injected dose per tissue (% ID), and time-course data were plotted.

Chemistry and Radiochemistry

General

The progress of all reactions was monitored by thin-layer chromatography (TLC) on 0.2 mm thick TLC plates (Merck, glass-backed, silica gel 60 F245), and spots were detected under UV light. Silica gel 60 (Kanto Chemical, particle size 0.04–0.05 mm) was used for purification by flash column chromatography. ¹H NMR (600 MHz) and ¹³C NMR (150 MHz) spectra were recorded on a Varian NMR System PS600 at room temperature. Deuterated chloroform (CDCl₃) was used as the solvent for all routine NMR measurements. Chemical shifts are reported in ppm relative to the respective deuterated solvent peak, δ 7.26 ppm for ¹H NMR and δ 77.2 ppm for ¹³C NMR, and coupling constants are given in Hz. FAB-MS spectra (low- and high-resolution mass spectra) were measured on a JEOL JMS-700 mass spectrometer. The purity of tested compounds was >95%, as confirmed by HPLC.

Determination of Bexarotene Purity and Reaction Progress Monitoring by HPLC

A Shimadzu liquid chromatographic system (Japan) consisting of a LC-20AT pump, SPD-20A detector, and CTO-10AS column oven was used. Data were processed using Labsolutions software. The reaction mixture (20 μL) was injected onto an Inertsil ODS-3 column (4.6 mm i.d. × 100 mm, 5 μm, GL Sciences, Tokyo, Japan) fitted with a guard column of Inertsil ODS-3 (4.0 mm i.d. × 10 mm, 3 μm, GL Sciences) at 40 °C; the mobile phase was MeOH/H₂O = 90/10 + 0.1% formic acid. The flow rate was 0.7 mL/min. A photodiode array (PDA) detector was used to monitor absorbance at 260 nm.

HPLC Conditions for Isolating the ¹¹C-Labeled PET Tracer

The HPLC system was a CFN-MPS-100 system, consisting of a PU-2086 Plus pump and UV-2075 Plus UV–vis spectrophotometric detector (Sumitomo Heavy Industry, Japan), equipped with an RI detector. Chromatographic isolation was performed on a YMC Pack ODS-AM (10 mm i.d. × 250 mm, YMC Co., Ltd., Japan). Data were processed using Chromeleon version 6.50 software. The mobile phase was MeOH/H₂O = 98/2 + 0.1% formic acid. The flow rate was 4.0 mL/min, and the absorbance at 260 nm was monitored.

HPLC Conditions for Analyzing the ¹¹C-PET Tracer

The HPLC system was a Shimadzu liquid chromatographic system (Japan) consisting of an LC-20Ai pump, an SPD-20A detector, GABI* (Raytest, Germany), a CTO-20A column oven, and Labsolutions software. The flow rate was 0.7 mL/min on an Inertsil ODS-3 column (4.6 mm i.d. × 100 mm, 3 μm, GL Science, Japan) at 40 °C; the mobile phase was MeOH/H₂O = 95/5 + 0.1% formic acid. Monitoring was done at 260 nm. The identity of the product was confirmed based on the UV spectrum acquired with a PDA detector.

Synthesis

See the Supporting Information.

Glossary

Abbreviations

9cisRA

9-cis retinoic acid

ARG

autoradiography

ATRA

all-trans retinoic acid

CMC

carboxy methylcellulose

CT

computed tomography

CTCL

cutaneous T-cell lymphoma

FDA

Food and Drug Administration of the United States

GOT

glutamic oxaloacetate transaminase

GPT

glutamic pyruvic transaminase

HPLC

high-performance liquid chromatography

LXR

liver X receptor

PET

positron emission tomography

PPAR

peroxisome proliferator-activated receptor

pTSA

p-toluenesulfonic acid

RAR

retinoic acid receptor

ROIs

regions of interest

RXR

retinoid X receptor

TBIL

total bilirubin

TLC

thin-layer chromatography

TMEDA

tetramethylethylenediamine

TR

thyroid hormone receptor

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsptsci.2c00126.

• Experimental flow chart, organ parameters, serum parameters, bone dysplasia, synthetic method of the ¹¹C-labeled PET tracer, tissue transferability of [¹¹C]bexarotene, experimental section, NMR, and MS charts (PDF)

Author Contributions

Y.T. and H.K. conceived and designed the project. M.F-T. and M.N-A. performed the oral administration study and prepared skeletal specimens. I.K. evaluated bone dysplasia. Y.T., M.A., and H.H. synthesized and analyzed the ¹¹C-labeled PET tracer. Y.T., M.W., and R.N. performed animal experiments. The manuscript was written by Y.T. and H.K.

This work was partially supported by the Ube Industrial Foundation (to H.K.) and JST SPRING (grant number JPMJSP2126 to Y.T.). This study was also funded in part by AIBIOS K.K. This funder was not involved in the study design, data collection, analysis, interpretation of data, writing of the article, or the decision to submit it for publication. All authors declare no other competing interests.

The authors declare the following competing financial interest(s): This study was also funded in part by AIBIOS K.K. This funder was not involved in the study design, data collection, analysis, interpretation of data, writing of the article or the decision to submit it for publication. All authors declare no other competing interests.