Molecular Mechanisms of the Teratogenic Effects of Thalidomide

1. Introduction

Thalidomide (Figure 1A) was first developed by Chemie Grünenthal (West Germany) in 1957 and was soon in use worldwide as a sedative. The use of thalidomide spread to more than 40 countries, but this drug was withdrawn from the market in 1961, as it was revealed to cause teratogenicity when taken during early pregnancy [1,2,3,4]. Clinical studies, however, demonstrated the therapeutic efficacy of thalidomide in several intractable diseases. First, in 1965, thalidomide was reported to be effective in erythema nodosum leprosum (ENL), an inflammatory complication of leprosy [5]. During the 1980s to early 1990s, thalidomide was shown to be effective in certain autoimmune diseases such as rheumatoid arthritis, Behcet’s disease, and chronic graft versus host disease [6,7,8,9]. Furthermore, in the early 1990s, thalidomide was reported to inhibit tumor necrosis factor (TNF)-alpha production and human immunodeficiency virus (HIV) replication [10,11,12]. In 1994, thalidomide was demonstrated to have anti-angiogenic activity, which suggested anti-cancer activity [13]. In 1999, thalidomide was shown to be effective against multiple myeloma, a malignant B cell lymphoma [14]. Based on these findings, in 1998 and 2006, thalidomide was approved by the Food and Drug Administration (FDA) for the treatment of ENL and multiple myeloma, respectively [15,16]. As the precise molecular mechanisms of thalidomide teratogenicity remains unclear, thalidomide prescription is strictly controlled by a program called the Thalidomide Risk Evaluation and Mitigation Strategy (REMS), formerly known as the System for Thalidomide Education and Prescribing Safety (STEPS) [17,18]. In Brazil, however, where leprosy is a common disease among the poor, the use of thalidomide led to a tragic increase in birth defects. Although the package was marked with a pictogram to prohibit its use by pregnant women, it was mistaken for a contraceptive due to poor literacy. Elucidating the molecular mechanisms of thalidomide embryopathy remains an urgent matter [19,20,21].

As the therapeutic efficacy of thalidomide was demonstrated, many thalidomide derivatives with greater potency were developed, yet the molecular mechanisms underlying the effects of thalidomide, such as inhibition of oxidative stress or angiogenesis, remained uncertain [22,23,24,25]. The most important question was to identify the direct target of thalidomide.

A decade ago, we identified cereblon (CRBN) as a primary target of thalidomide teratogenicity [26]. Since then, our understanding of the mechanisms of action of thalidomide have advanced significantly. Currently, CRBN is thought to act basically as a subunit of a ligand-dependent E3 ubiquitin ligase complex whose substrate recognition can be controlled by thalidomide or its related compounds [27]. CRBN is required for both the teratogenic effects and the therapeutic effects of thalidomide and its derivatives. Recently, CRBN-binding drugs have vividly been developed [28]. In this review, we introduce the basic functions of CRBN and discuss our current understanding of the molecular mechanisms of thalidomide, mainly focusing on its teratogenicity.

2. Teratogenic Activity of Thalidomide

When pregnant women took thalidomide between day 20 and day 36 after fertilization, multiple birth defects occurred [29]. A single tablet (50 mg) of thalidomide was enough to induce developmental defects [29]. A broad spectrum of birth defects was reported, including malformations of the limb, ear, eye, internal organs, face, genitalia, and central nervous system [29,30,31]. Even during the thalidomide-sensitive time period, by comparison, the earlier stages are particularly prone to more serious damages. First, the damage caused by taking thalidomide between day 20 and day 24 after fertilization appears as missing external ear. Thalidomide intake after day 24 causes multiple phenotypes such as damage in the inner ear, ear deformation, ocular anomalies, and upper limb damage (phocomelia, amelia), or hip dislocation. Damage to the lower limbs is seen in the comparatively late intake of the drug during the thalidomide sensitive time window, which is after day 27. Malformations in thumbs are seen by taking thalidomide from day 24 and even after day 31 [29]. The mortality rate for infants with thalidomide-induced birth defects was reported to be 30%–40%. Thalidomide caused imperforate anus and other gastrointestinal deformities in many infants, contributing to early death [31,32]. In addition, an unknown number of miscarriages were caused by thalidomide.

Limb defects were very frequently observed [31]. Both upper limbs and lower limbs were affected by thalidomide. Two types of limb defects are induced by thalidomide, phocomelia and amelia. The limb is composed of the stylopod, zeugopod, and autopod. Phocomelia describes an abnormal limb with a stylopod, a truncated or absent zeugopod, and a nearly intact autopod, while amelia is a complete loss of the limb [29,33,34]. Polydactyly was also observed in deformed limbs, including phocomelia [1,29,35,36]. Defects of the shoulder and hip joint points were reported [34]. Auricular defects were also very frequently observed, including anotia, mild malformation of the external ear, and hearing loss [29,31,35]. Ocular anomalies included uveal coloboma, glaucoma, and microphthalmia. With respect to the internal organs, kidney malformations, heart defects, and structural chest defects were frequently observed [34,37,38]. Facial palsy and facial asymmetry were also common [31,35]. Thalidomide was also reported to potentially affect facial muscles and facial nerves [29,31,35]. Autism and intellectual disability were also reported [31,35,39].

Thalidomide causes limb defects in humans, monkeys, rabbits, chicks, and zebrafish [27,30,33]. In monkeys and rabbits, both amelia and phocomelia occur. In chicks, only amelia occurs [33]. In zebrafish, thalidomide inhibits the development of the pectoral fins along the proximodistal axis [26,40,41]. Although the fin is structurally different from the limb in mammals and chicks, the molecular pathways are evolutionary conserved. Thalidomide inhibits chondrogenic differentiation in pectoral fins. Pectoral fins are composed of the endoskeletal disc, the scapulocoracoid, and the cleithrum. Thalidomide treatment in the early stage of development resulted in severe defects in chondrogenesis and retardation of the endoskeletal disc and cleithrum [40]. Rodents are resistant to limb deformities induced by thalidomide. It was reported that thalidomide did not induce limb defects in rats even at doses of up to 4000 mg/kg [42]. Why mice and rats are resistant to thalidomide teratogenicity remains unknown.

In vertebrates, fibroblast growth factor 8 (FGF8) is essential for the development of limbs, including fins [43,44]. FGF8 is expressed in the apical ectodermal ridge (AER), the distalmost end of the developing limbs. Thalidomide was shown to reduce the expression of FGF8 in the AER in rabbits, chicks and zebrafish [26,45]. Downregulation of FGF8 leads to induction of pro-apoptotic genes and therefore to malformation of the limbs. Therefore, the effects of thalidomide are likely mediated through evolutionarily common signal transduction pathways in different vertebrates.

3. The Direct Target of Thalidomide

In 2010, the identification of CRBN as a thalidomide-binding protein represented a major advance in understanding the molecular mechanism of thalidomide [26]. We immobilized various small chemical bioactive compounds onto ferrite glycidyl methacrylate (FG) beads to study their targets [46,47,48]. Affinity purification using thalidomide-immobilized FG beads led to the identification of CRBN and DNA damage-binding protein 1 (DDB1) as thalidomide-binding proteins. The function of CRBN was unknown at the time, although it was thought to be related to mental retardation and intellectual disability [49]. A clue to the function of CRBN was DDB1, the protein co-purified with CRBN. DDB1 forms a complex with Cullin RING type E3 ubiquitin ligase (CRL4) [50,51,52,53]. CRBN forms a complex with Cullin 4 (Cul4), DDB1, and regulator of Cullins-1 (Roc1) and functions as a substrate of this CRL4 complex (CRL4^CRBN). The autoubiquitination of CRBN was shown to be inhibited by thalidomide. It was found that the CRBN^{Y384A/W386A(YW/AA)} mutant did not bind to thalidomide. The zebrafish and chick developmental model systems were utilized to evaluate whether CRBN was genuinely involved in mediating thalidomide-induced teratogenicity. When CRBN^YW/AA was overexpressed, the teratogenic phenotypes of thalidomide were reversed in both chicks and zebrafish. The expression of FGF8 was restored by CRBN^YW/AA expression even after thalidomide treatment. These findings demonstrated that CRBN was a primary target of thalidomide and critically involved in thalidomide teratogenicity.

4. CRBN as a Therapeutic Target of Thalidomide and Its Derivatives

The finding that CRBN was a critical, direct target of thalidomide and functioned as a subunit of a CRL4 E3 ligase greatly advanced our understanding of the molecular mechanism of thalidomide and its derivatives. The therapeutic effects of thalidomide and its derivatives, rather than the teratogenic effects, became the focus of further research. As mentioned previously, the remedial effect of thalidomide against multiple myeloma led to the development of its derivatives lenalidomide and pomalidomide—both of which are now approved by the FDA. Pomalidomide is a compound in which an amino group is added to the phthalimide of thalidomide, and lenalidomide has the structure of pomalidomide without the carbonyl group on the phthalimide moiety (Figure 1B,C). Both compounds are called immunomodulatory drugs (IMiDs) and are reported to have more potent immunomodulatory activity than thalidomide [15,54].

An intriguing question was whether CRBN was involved in the therapeutic effects of lenalidomide and pomalidomide. In 2011, Stewart and colleagues found that knockdown of CRBN by RNA interference (RNAi) blocked the inhibition of cell proliferation by lenalidomide or pomalidomide in several multiple myeloma cell lines [55]. The group also reported that the expression of CRBN was considerably lower in pomalidomide-resistant cell lines [55]. Celgene Corporation and our group confirmed their data in 2012. We also demonstrated that compounds containing a glutarimide moiety, such as lenalidomide and pomalidomide, bind to CRBN [56], and lenalidomide and pomalidomide bound to CRBN more strongly than thalidomide. These results suggested that CRBN is required not only for the teratogenic effects but also the therapeutic effects of thalidomide and its derivatives.

5. Ligand-Dependent Substrate Recognition of CRL4^CRBN

Since CRBN was shown to be involved in the anti-cancer effects of thalidomide and its derivatives, researchers next investigated the relevant CRBN substrates. Since then, several CRBN substrates have been identified.

6. Structure of the CRBN–Drug–Neosubstrate Complex

The structure of the direct target of thalidomide has been intriguing to structural biologists. In 2014, Thoma, Fischer, and colleagues reported the X-ray structure of a chimeric complex of human DDB1 and chick CRBN bound to thalidomide [69]. Celgene and our group also reported the X-ray structure of human CRBN and human DDB1 bound to lenalidomide [70]. Chick CRBN is highly homologous to human CRBN. Chick CRBN is composed of at least three domains, a seven-stranded β-sheet located in the amino-terminal domain (NTD, residues 1–185), an α-helix bundle domain (HBD, residues 186–317) containing seven helices, and a carboxy-terminal domain (CTD, residues 318–445) composed of eight β-sheets. The NTD and the HBD are related to the N-terminus of Lon protease. The structure of DDB1 is composed of three β-propeller blades (BPA, BPB, and BPC). The HBD of CRBN binds BPA and BPC. The CTD contains a zinc finger domain and the thalidomide-binding domain (TBD). The zinc finger consists of C323, C326, C391, and C394 in humans. The physiological significance of the zinc finger is still unknown, but it is noteworthy that the C391R mutation of CRBN was reported to be associated with intellectual disability [71]. TBD possesses three tryptophans (W380, W386, and W400) that form a pocket called the tri-trp pocket. The glutarimide moiety of thalidomide and its derivatives can be inserted into the pocket.

Subsequent reports by these same groups in 2016 showed that the ligands form a molecular glue between CRBN and its neosubstrates. The Thoma group reported the X-ray structure of the CK1α–lenalidomide–CRBN/DDB1 complex and Celgene and our group reported the structure of the GSPT1–CC-885–CRBN/DDB1 complex [64,72]. Both CK1α and GSPT1 contain a β-hairpin motif, which is essential for binding to CRBN. In the CK1α complex, lenalidomide acts as a molecular glue between CRBN and CK1α. The G40 residue of CK1α in the hairpin motif is sandwiched between phthalimide and CRBN. The glycine is critical for the interaction with lenalidomide. One of the carbonyl groups in the phthalimide moiety prevents steric clash. Therefore, lenalidomide degrades CK1α more efficiently than thalidomide or pomalidomide. In the GSPT1 complex, CC-885 connects CRBN to GSPT1. The chloro-methyl-phenyl group of CC-885 is important for binding to GSPT1. The G575 of GSPT1, like the G40 of CK1α, is essential for its interactions with CC-885 and CRBN. The F150 residue of CRBN, which is in the NTD, is important for binding to GSPT1, but is not essential for interactions with CK1α and Ikaros. Although the functions of the NTD of CRBN are still largely unknown, the NTD is required for its interactions with several neosubstrates. In 2018, the structure of CRBN bound to the zinc finger motif of Ikaros or the related neosubstrate ZNF692 in the presence of pomalidomide was determined, and the amino group of pomalidomide was shown to contribute to the interaction with Ikaros/ZNF692 and CRBN [68].

Thalidomide and its derivatives have one chiral center. Thalidomide and other IMiDs are usually used as a mixture of the (S)-isomer and the (R)-isomer. Previously, researchers thought that thalidomide exerts different effects depending on the enantiomer [73]. However, the isomers rapidly racemize under physiological conditions [74]. It therefore remained unclear whether thalidomide acted differently on different optical isomers. Researchers have shown that the (S)-enantiomer bound approximately 10-fold more strongly to CRBN [70,75] than the (R)-enantiomer. The crystal structures of CRBN and each enantiomer revealed that both enantiomers bind the tri-trp pocket. However, the bound form of (S)-enantiomer to CRBN exhibited a more relaxed conformation of its glutarimide ring [69,70,75]. The conformation of the (R)-enantiomer is twisted to avoid steric clashes, resulting in a weaker binding activity. The (S)-enantiomer more effectively decreased Ikaros protein in multiple myeloma and induced fin defects in zebrafish [75]. Although the (R)-enantiomer possesses weak CRBN-binding activity, it might be a supplier of (S)-enantiomer by racemization under physiological conditions.

Through structural biological analyses, species-specific effects of thalidomide and its derivatives have been partly elucidated. In rodents, neither teratogenic effects nor therapeutic effects of thalidomide and its derivatives have been observed [76,77]. Mouse CRBN is 95% homologous to human CRBN and can bind to thalidomide. However, it was shown that lenalidomide and pomalidomide did not degrade neosubstrates such as Ikaros or CK1α in mouse cells or human cells in which CRBN was replaced with mouse CRBN [63]. There are two critical amino acids in human CRBN, E377 and V388. In rodents, these amino acids are substituted to valine and isoleucine, respectively. The V388I substitution of CRBN abolishes the interaction with Ikaros and CK1α in the presence of thalidomide or IMiDs [63]. Structural studies have shown that CRBN binds to Ikaros and CK1α at V388 [68,72]. The V388I mutation induces steric clash and therefore prevents binding. The E377V substitution abolished the interaction with GSPT1 in the presence of CC-885 because E377 binds to the urea of CC-885 [62]. Lenalidomide has been shown to degrade mouse CK1α in “humanized” mouse cells expressing mouse CRBN^I391V [63]. Researchers constructed humanized mice expressing mouse CRBN^I391V [78,79] and found that lenalidomide acted on the hematopoietic stem cells in humanized mice. Thalidomide induced fetal loss in these mice, but limb defects were not observed [79]. Thalidomide resistance in rodents therefore remains a mystery.

7. Teratogenic Mechanisms Associated With CRBN

While the substrates of thalidomide and its derivatives and their roles in its therapeutic effects have been elucidated, the molecular basis of the teratogenic effects remained unclear. Four downstream factors have been identified as candidate players in thalidomide—CRBN-mediated teratogenicity.

8. Concluding Remarks

This decade saw remarkable progress in our understanding of the targets and underlying molecular mechanisms of thalidomide. First, thalidomide binds to CRBN, which recruits a neosubstrate, then ubiquitinates the bound neosubstrate. Ubiquitinated neosubstrates are degraded via the ubiquitin–proteasome pathway, and various effects occur, depending on the neosubstrate.

More thalidomide derivatives, now called cereblon E3 ligase modulators (CELMoDs), have been developed and are now being tested in clinical trials [66,103]. CC-122 (avadomide) (Figure 1F) has broader effects than IMiDs and is effective against diffuse large B cell lymphomas (DLBCLs) and solid tumors [104,105]. CC-220 (iberdomide) (Figure 1G) has stronger Aiolos/Ikalos degradation activity than IMiDs and is being studied for the treatment of systemic lupus erythematosus (SLE) [106,107,108]. Among the most recently developed CELMoDs is CC-92480 (Figure 1H), which is being studied for the treatment of lenalidomide-resistant multiple myeloma [109].

Furthermore, new approaches utilizing the drug-dependent CRL4^CRBN ubiquitination activity are being established. By combining thalidomide or other CRBN-binding compounds with other low-molecular-weight compounds that interact with pathogenic proteins, CRL4^CRBN can be recruited to degrade specific proteins of interest (POIs). This technology is called Proteolysis-Targeting Chimeras (PROTACs) and was originally proposed by Crews and Deshaies [110]. To achieve this targeted protein degradation approach, many PROTAC molecules are being synthesized and tested for clinical use [111,112,113,114,115]. For example, dBET1 (Figure 1I), among the earliest CRBN-based PROTACs to be developed by Bradner and colleagues, is a fusion of thalidomide and JQ1 [113]. JQ1 is an acetylated-lysine-like compound that binds to bromodomain and extra-terminal (BET) motif-containing transcription factors such as BRD4 (Bromodomain containing 4) and exerts its effects by competitively inhibiting the binding of these transcription factors to acetylated histones. As silencing of BRD4 downregulates the expression of the MYC (v-Myc Myelocytomatosis Viral Oncogene Homolog) oncogene, JQ1 is a potent anti-cancer molecule [116]. Nevertheless, dBET1 has improved anti-cancer activity, as it degrades the transcription factor itself directly, which leads to comparatively rapid inhibition.

Although expectations are high for the clinical use of CRBN-binding compounds as a means of targeted protein degradation, the potential for serious adverse effects needs to be taken into consideration. Downstream targets of thalidomide are gradually being identified, but we have a long way to go to completely elucidate the mechanism of teratogenicity. The structural comparison of the therapeutic neosubstrates with the teratogenic neosubstrates might contribute to develop new safer CRBN-binding drugs [117].

Finally, the function of CRBN in the absence of thalidomide or its derivatives remains mostly uncharacterized. As briefly mentioned above, CRBN was originally reported to be related to intellectual disability, and it was also reported that CRBN interacts with the BK_Ca channel α subunit (Slo) (BK_Ca stands for Large-conductance Ca²⁺- and voltage-gated big K⁺) [118,119]. The function of CRBN without thalidomide or other drugs has been examined in animals. Forebrain-specific conditional CRBN knockout mice displayed learning disabilities, and knockdown of CRBN in zebrafish impaired brain development [120,121]. These studies suggest that CRBN function is related to the developing brain, but the function of ligand-unbound CRBN remains unknown. Uridine was also reported to bind to CRBN, and a high concentration of uridine mimicked the thalidomide fin deformity in zebrafish [122]. There might be additional natural ligands or metabolites that bind to CRBN [123]. Further understanding of the basic function of CRBN may lead to the discovery of new biological phenomena and will contribute to the development of safer and more effective drugs.

Author Contributions

Conceptualization, T.I.; writing—original draft preparation, T.A.-O. and T.I.; writing—review and editing, T.A.-O., T.I. and H.H.; supervision, T.I. and H.H.; funding acquisition, T.I. All authors have read and agreed to the published version of the manuscript.

Funding

This work was supported by a Grant-in-Aid for Scientific Research (KAKENHI) from the Japanese Society for the Promotion of Science (JSPS) (grant number 18H05502, to T. I.).

Conflicts of Interest

The authors have no conflicts of interest to declare.