Retinoic acid (RA) is a morphogen that plays a key role in vertebrate embryogenesis. It is produced from provitamin A in mesodermal tissues that express representatives of the retinaldehyde dehydrogenase family of enzymes. RA is primarily a differentiation inducer. It competes with growth stimulating factors, such as those of the FGF family, and with other developmental regulators, such as those belonging to the Wnt and Hox families, to exert its effects.
The balance among a host of interacting factors, which changes with time during embryogenesis and is dependent on localisation within the embryo, determines the fate of individual cells in individual locations at distinct time points during development. RA plays a key role in the formation of the vertebrate body plan, being involved in anterior-posterior patterning, axial differentiation of the neural tube, caudal-ventral specification within the central nervous system as well as hindbrain development. Moreover, it regulates neural crest cell migration, contributing to the formation of a host of tissues and organs, such as facial structures, heart, the hematopoietic system, limb innervation and peripheral ganglia. RA activity is determined by the local presence, subtypes, and density of retinoid receptors, which have been grouped in RAR and RXR receptor families. Though RA receptors seem ubiquitous throughout the embryo, specific representatives of these receptor families have specific spatial distributions within embryonic tissues (Rowe et al., 1992; Viallet and Dhouailly, 1994; Elmazaar et al., 1996; Romand et al., 1998; Mandal et al., 2013). This may explain differences in embryotoxic characteristics among various embryotoxicants that all interfere with RA homeostasis.
In addition, RA is metabolised through CYP26 isoforms, which also show a subtype, time- and location-specific expression during embryogenesis. Other mechanisms such as sequestering to RA-binding protein 1 and 2 may also contribute to this regulation. For example, RA plays a crucial age- and cell-specific role in cranio-facial morphogenesis, including palatogenesis. Over-expression of RA at specific foetal ages can disrupt these processes and cause teratogenic effects, including the induction of cleft palate. Since catabolism by CYP26 is the most important pathway, inhibition of this enzyme in a particular tissue, such as the developing head, would result in increasing RA levels (Chambers et al., 2014). Thus, a strictly programmed multifactorial interplay between RA-producing and RA-metabolising enzymes, competing growth and development stimulating factors, and retinoid receptors and their time- and location-specific expression leads vertebrate embryogenesis from a fertilised egg to a morphologically recognisable vertebrate embryo. The central role of retinoid function in vertebrate embryogenesis provides opportunities for identifying biomarkers of abnormal development that may allow detection of a large proportion of developmental toxicants. Many teratogens and embryotoxicants may be assumed to interfere at some level with retinoid homeostasis, be it through direct interaction with, for example, its production, metabolism or receptor binding, or as a secondary consequence of initiating events occurring in pathways that interact with the retinoid effect, such as the expression of Hox genes or FGF. An AOP framework describing RA homeostasis and its functional interactions with other morphogenetic factors in embryogenesis could help identify such biomarkers. A first attempt towards such a framework was published recently (Tonk et al., 2015) and is depicted in Figure 5 below. This study also reviews data showing that retinaldehyde dehydrogenases, CYP26 members, and a host of RA-regulated patterning genes can be readily detected and shown to be regulated in alternative assays such as whole embryo culture, zebrafish embryo test and embryonic stem cell tests. Furthermore, in silico developmental models (Knudsen et al., 2015), such as exist for eye and limb development, also show direct connections with retinoid regulation
Figure 5: Proposed AOP framework for RA-neural tube/axial patterning pathway
Reproduced with permission from Tonk et al. (2015).
The importance of RA homeostasis is exemplified by human teratogens as well as by knockout mouse studies. The production of RA from beta-carotene is an important rate-limiting mechanism for systemic exposure in man. It is well known that pregnant women who consume high amounts of carrots during pregnancy may acquire an orange skin through extensive beta-carotene deposition, but this does not affect their babies due to limited metabolism to the active form of vitamin A, which is RA. Synthetic retinoids used as pharmaceuticals against persistent acne caused severe facial, limb and heart malformations (Lammer et al., 1985). However, oral human exposure in pregnancy to RA via multivitamin preparations marketed in the 1980s resulted in children with similar abnormalities (Werler et al., 1990; Rothman et al., 1995) Retinaldehyde dehydrogenase deficient mice show uncontrolled growth of undifferentiated tissue in the facial area (Rhinn and Dollé, 2012). CYP26-deficient mice show caudal regression syndrome due to precocious cell differentiation limiting caudal growth (Rhinn and Dollé, 2012). Because of the regional specification of CYP26 subtype expression, the specificity of malformations in CYP26-deficient mice depends on the CYP26 subtype being knocked out (Pennimpede et al., 2010). In humans, vitamin A deficiency has recently been related to ear malformations (Emmett and West, 2014).
It will be of great interest to investigate all areas of chemical space for their interactions with the retinoid system during embryogenesis in order to determine its predictive value and to define sensitive biomarkers for abnormal development in alternative test systems. Existing databases can be searched specifically for retinoid-related mediators of development, be it at the level of gene expression, proteomics, metabolomics, or whatever level of biology that provides practical tools for monitoring possible adverse effects of chemicals and drugs on vertebrate (and especially human) development. As an example, in the zebrafish embryo model, developmentally toxic triazole antifungals have been shown to upregulate CYP26 enzymes and downregulate retinaldehyde dehydrogenase (Hermsen et al., 2012). The use of azole compounds as fungicides is based on their greater affinity for the fungal sterol 14α-demethylase (CYP51) than for the mammalian or plant enzymes. In fungi they block the synthesis of the essential membrane component ergosterol. However, inhibition of CYP51 is not specific and other CYPs can also be affected, including CYP19 (the aromatase) and CYP26, which metabolises RA. Consequently application of RA or ketoconazole to pregnant rats (Mineshima et al., 2012) or itraconazole to pregnant mice (Tiboni et al., 2006) induced cleft palates and other skeletal effects. Inhibition of aromatase by azole compounds leads to post-implantation loss due to inhibition of 17β-oestradiol synthesis. Multiple additional examples of retinoid pathway modulation by embryotoxicants have been identified.