SEARCH

SEARCH BY CITATION

Keywords:

  • P450;
  • Cyp26C1;
  • retinoic acid;
  • chick embryo

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

We have cloned a novel retinoic acid (RA) catabolizing enzyme, Cyp26C1, in the chick and describe here its distribution during early stages of chick embryogenesis. It is expressed from stage 4 in the presumptive anterior (cephalic) mesoderm, in a subset of cephalic neural crest cells, the ventral otic vesicle, mesenchyme adjacent to the otic vesicle, the branchial pouches and grooves, a part of the neural retina, and the anterior telencephalon, and shows a dynamic expression in the hindbrain rhombomeres and neuronal populations within them. By examining the distribution of Cyp26C1 in the RA-free quail embryo, we can determine which of these expression domains is dependent on RA, and it is only the rhombomeric sites that do not appear, suggesting a role for RA in this location. The most striking domain of Cyp26C1 distribution is in the anterior cephalic mesoderm, which is adjacent to the domain of Raldh2 in the trunk mesoderm, but separated from it by a gap dorsal to which the posterior hindbrain will develop. We suggest that a gradient of RA within the mesoderm generated by Raldh2 and catabolized by Cyp26C1 could be responsible for patterning the hindbrain. We have compared this distribution of Cyp26C1 with that of Cyp26A1 and Cyp26B1 in the chick and shown that they generally occupy nonoverlapping sites of expression in the embryo, and as a result, we suggest individual roles for each of the Cyp enzymes in the developing embryo. Developmental Dynamics 230:509–517, 2004. © 2004 Wiley-Liss, Inc.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

Retinoic acid (RA), the major active derivative of vitamin A (retinol), is an important signaling molecule that regulates pattern formation during vertebrate embryonic development. Tissues such as the hindbrain (Maden et al., 1996; Gavalas and Krumlauf, 2000; Niederreither et al., 2000), spinal cord (Sockanathan and Jessell, 1998; Pierani et al., 1999; Liu et al., 2001), and eye (Hyatt et al., 1996; Dickman et al., 1997; Wagner et al., 2000) depend on a temporally and spatially regulated supply of RA.

Without this supply of RA, for example under conditions of maternal dietary deprivation (Kalter and Warkany, 1959; Heine et al., 1985; Dersch and Zile, 1993; Maden et al., 1996; Gale et al., 1999; Maden et al., 2000) or a null mutation in one of the enzymes that synthesizes RA (Niederreither et al., 1999), a battery of congenital defects result. Paradoxically, flooding this carefully regulated system by administering excess RA is also teratogenic, with many of the defects similar to those generated by a lack of vitamin A.

To become biologically active, retinol must first be enzymatically converted to RA by means of a two-step metabolic process. Retinol is oxidized to retinaldehyde by the retinol or alcohol dehydrogenases (RoDHs or ADHs) and then retinaldehyde is oxidized to either all-trans-RA or its isomer 9-cis-RA by the retinaldehyde dehydrogenases (RALDHs; Duester, 2000). Many distinct RALDHs exist, but RALDH1, RALDH2, and RALDH3 have been demonstrated to play an essential role in retinoid signaling and show discrete domains of expression during embryonic development (Ang and Duester, 1999; Haselbeck et al., 1999; Schneider et al., 2001). It is the spatial and temporal regulation of these RALDHs that generates a local RA supply to the embryo (the on switch or source), but the other component of the system (the off switch or sink) are the CYP26 enzymes, which breakdown RA.

The CYP26 enzymes are members of the P450 superfamily and catabolize RA into the inactive metabolites 4-OH-RA, 18-OH-RA, and 5,8-epoxy-RA (White et al., 1996, 2000; Fujii et al., 1997). Another product is 4-oxo-RA, but this compound shows considerable biological activity (Pijnappel et al., 1993; Sonneveld et al., 1999). Evidence that the cytochrome P450s play important roles in embryogenesis has been provided by generating knockout mice for the electron donor to all microsomal P450 enzymes, cytochrome P450 reductase (Otto et al., 2003). At embryonic day (E) 9.5, these embryos had increased levels of RA, reduced levels of retinol, and many of the observed abnormalities are consistent with disruption in RA homeostasis.

The first member of the CYP26 family to be identified, Cyp26A1, has been described in the Xenopus, chick, and mouse embryo (Fujii et al., 1997; Hollemann et al., 1998; de Roos et al., 1999; Swindell et al., 1999; Blentic et al., 2003). It has a dynamic expression pattern in the anterior epiblast, the tail bud, and the dorsal neural tube. In Cyp26A1 knockout mice, the observed phenotypes are consistent with excess RA administration, including caudal regression and spina bifida (Abu-Abed et al., 2001; Sakai et al., 2001). These defects suggest a role for CYP26A1 in regulating RA levels at the caudal end of the embryo. A second member of the family, Cyp26B1, has been described in the mouse (MacLean et al., 2001; Abu-Abed et al., 2002) and avian embryo where its regulation by RA has been demonstrated (Reijntjes et al., 2003). But a null mutant phenotype has not yet been reported. The expression pattern of Cyp26B1 in the chick embryo would suggest a role for this gene in vasculogenesis (Reijntjes et al., 2003). A novel murine cytochrome P450, the third member of the family, Cyp26C1, has very recently been described (Tahayato et al., 2003). Its expression in the early developmental stages of the mouse is in prospective rhombomeres 2 and 4 and first branchial arch. During late gestation, expression is in the inner ear epithelium and inner dental epithelium of the developing teeth.

Studies on the developing eye have revealed the important relationship between adjacent domains of the RA-generating enzymes, the RALDHs, and the breakdown enzymes, the CYPs. There are clear on/off borders where RA levels are high (the dorsal and ventral retina) and an equatorial region where RA is absent, and these regions correspond to sites of Raldh expression and Cyp26A1 expression, respectively (McCaffery et al., 1999). Similarly in the rostrocaudal axis of the embryo, there is a sharp on/off border of RA synthesis at the level of the first somite (Maden et al., 1998), which corresponds to the border of Raldh2 expression (Swindell et al., 1999) and undetectable levels of RA rostral to this border, which corresponds to the expression domain of Cyp26A1. This finding has led to the idea that a gradient of RA could be generated between these two domains across which the posterior hindbrain will develop and be patterned (Swindell et al., 1999; Maden, 1999). However, more detailed analyses have revealed that these two enzymes are expressed in completely different germ layers, Raldh2 is expressed in the mesoderm (subsequently the paraxial mesoderm), whereas Cyp26A1 is expressed in the epiblast (subsequently the anterior neuroepithelium; Blentic et al., 2003). This difference in tissue type is an unlikely scenario for generating a gradient. Indeed, the Cyp26A1 null mutant mouse in which the gradient should be abolished shows no abnormalities in anterior central nervous system (CNS) development (Abu-Abed et al., 2001).

Following on from the cloning of the novel murine CYP, Cyp26C1 (Tahayato et al., 2003), we have used the murine sequence to identify a chick expresses sequence tag (EST) from the chicken EST database (Boardman et al., 2002). We have cloned this enzyme from the chick, describe its distribution in the chick embryo, and compare this with the distributions of Cyp26A1 and Cyp26B1. We show that chick Cyp26C1 is expressed in the anterior mesenchyme and that a double in situ with Raldh2 exhibits a gap dorsal to which the posterior hindbrain will develop. We suggest that this is the CYP that is responsible for generating the RA gradient in the rostrocaudal axis of the mesoderm and predict that a Cyp26C1 null mutant mouse will show a strong anterior CNS phenotype.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

Expression in Stages 4–9

Cyp26C1 first appears in the chick embryo at stage 4 in the anterior-most part of the embryo (Fig. 1A), and expression moves caudally to surround Hensen's node (Fig. 1B) such that, by stage 6, it is intense throughout the rostral part of the embryo, except for a central domain where the prechordal mesoderm has differentiated (Fig. 1C). A close-up of this expression domain reveals a mosaic pattern (Fig. 1E), which on sectioning is revealed to be mesodermal (Fig. 1F). The neuroepithelium, ectoderm, and endoderm are devoid of expression. This expression domain of Cyp26C1 in the mesoderm is remarkably similar to that of Cyp26A1, which is expressed in the neural plate (Swindell et al., 1999; Blentic et al., 2003), and to determine whether these two Cyps do in fact overlap, we performed a double in situ with these two genes. Figure 1D shows that Cyp26C1 expression (in turquoise) extends more laterally than that of Cyp26A1 (in purple). The white arrows mark the lateral extent of Cyp26A1, and the black arrowheads mark the region where Cyp26C1 is expressed alone. However, the posterior boundary is the same (red arrow in Fig. 1D) and a sagittal section through this embryo (Fig. 1G) confirms the impression. Cyp26A1 in the neural plate (purple) has the same posterior border as Cyp26C1 in the mesoderm (turquoise).

thumbnail image

Figure 1. Expression of Cyp26C1 from stages 4 to 9. A: Expression begins at stage 4 at the anterior end of the embryo. B: A short while later (stage 4+), expression is extending caudally. C: By stage 6, expression is present throughout the anterior end of the embryo, except for the midline where the notochord is. D: A double in situ hybridisation at stage 6 for Cyp26C1 (in turquoise) and Cyp26A1 (in purple), showing that these two genes extend to the same caudal border (red arrow) but that Cyp26C1 expression extends more laterally (black arrowheads) than that of Cyp26A1. The lateral borders of Cyp26A1 are marked with white arrows. E: A close-up of the expression in C shows that it is not uniform but mottled in appearance. F: A section of the embryo in C shows that the expression is in the mesoderm (mes) not in the endoderm (en) or neural plate, which explains the mottled appearance from above, because the mesoderm is not a continuous sheet at this stage. ne, neuroepithelium; ep, epithelium. G: A section through the double in situ shown in D, showing Cyp26A1 (purple) in the neural plate with the caudal border (black arrowhead) and Cyp26C1 (turquoise) in the mesoderm with the same caudal border (red arrowhead). H: A double in situ hybridisation at stage 6 with Cyp26C1 (turquoise) in the anterior mesoderm and Raldh2 (purple) in the posterior mesoderm. Between them is a gap (red arrowheads) above which, within the neural plate, the posterior hindbrain will develop. I,J: Double in situ hybridizations to show the caudal spread of Cyp26C1 and its relationship to Hoxb-1 expression, which has a fixed anterior border at the rhombomere 3/4 border. I: At stage 6, there is a gap between the two domains. J: At stage 9, the gap has disappeared as Cyp26C1 spreads caudally. K: A close-up of the hindbrain of two stage 9 embryos (as in J), but with single gene hybridizations to show that the two domains overlap. On the left is the expression of Hoxb-1 in the neural tube (red arrowhead), which spreads caudally to the level of the rhombomere 3/4 border, just anterior to the first somite. On the right is the expression of Cyp26C1 in the mesoderm (black arrowhead), which has spread posterior to the first somite border and, thus, overlaps the rostral border of Hoxb-1.

Download figure to PowerPoint

Because Cyp26C1 catabolizes RA into mostly inactive compounds (see introduction section) and it is expressed in the mesoderm, it is of great interest to determine its spatial relationship to Raldh2, the enzyme that generates RA in the mesoderm. The double in situ in Figure 1H reveals that there is a gap between the anterior border of Raldh2 (in purple) at the level of the presumptive first somite (Blentic et al., 2003) and Cyp26C1 (in turquoise). The red arrowheads in Figure 1H mark this region, which is where the posterior hindbrain will develop in the neural plate above.

We also asked whether the posterior expression border of Cyp26C1 continued to move any further posteriorly after stage 6. To do this, we performed double in situs between Cyp26C1 and Hoxb-1, the latter having a stable anterior expression border at the rhombomere 3/4 border from stage 7 onward (Gale et al., 1999). At stage 6, there is a gap between these two gene domains (Fig. 1I – Cyp26C1 in purple, Hoxb-1 in turquoise) marked by the red arrowheads in Figure 1I. By stage 9 (Fig. 1J,K), the posterior border of Cyp26C1 has moved caudally within the mesoderm to reach past the level of the first somite (black arrowhead in Fig. 1K, right side) and, hence, overlaps the expression of Hoxb-1 at the r3/4 border (red arrowhead in Fig. 1K, left side), which is anterior to the level of the first somite. Thus, the posterior border of Cyp26C1 ultimately extends more caudally than that of Cyp26A1 in the overlying neuroepithelium, because in the latter case, Cyp26A1 and Hoxb-1 meet but never overlap (Blentic et al., 2003). Thus, up to stage 9, the expression of Cyp26C1 can be summarized as head mesoderm.

Expression in Stages 10–13

When neurulation has been completed, the formation of rhombomeres commences and neural crest emigration begins, there is an abrupt change of Cyp26C1 expression from head mesoderm to anterior rhombomeres and neural crest. Thus, at stage 10−, when only an enlarged rhombomere A (proto-rhombomeres 1 and 2) is present, Cyp26C1 expression is in a discrete region of this rhombomere (Fig. 2A,B). The left and right streams of crest move rostrally and ventrally and meet in a dense group of cells below the developing forebrain (red arrowhead in Fig. 2A). A section through this rhombomere A region (Fig. 2C) reveals that the whole thickness of the neuroepithelium expresses Cyp26C1 except the ventral floor plate and that in addition to the stream of neural crest the ectoderm of that region also expresses the gene. Strikingly, there is no expression in any tissue in the trunk (Fig. 2A). At stage 10+, the rhombomeric expression has expanded to include rhombomere 3 and by stage 11, Cyp26C1 expression is in rhombomeres 2, 3, and 5 (Fig. 2D). The lateral view of the head of a stage 11 embryo shown in Figure 2E emphasizes that Cyp26C1 is only expressed in a subset of cranial neural crest, in a stream descending ventrally to below the developing eye. The remaining crest surrounding the developing brain does not express this gene.

thumbnail image

Figure 2. Expression of Cyp26C1 from stages 10 to 14. A: Expression at stage (st) 10− is only found in the head; the rest of the embryo is completely devoid of expression. Within the head, expression is in the rostral hindbrain, in two streams of neural crest emanating from the rostral hindbrain and meeting in a knot ventral to the developing forebrain (red arrowhead). B: A close-up of the embryo in A, showing that the caudal part of this anterior rhombomere is expressing Cyp26C1. The red arrowhead marks the position of the midbrain/hindbrain border. C: Section through the expressing part of the rhombomere showing that the whole of the neuroepithelium, except the ventral floor plate, is expressing Cyp26C1 as well as the neural crest and the ectoderm of that region. D: At stage 11, the rhombomeres are more fully individualized and Cyp26C1 is now localized to rhombomeres 2, 3, and 5 and the neural crest emanating from them. E: Side view of a stage 12 embryo, showing that the Cyp26C1-expressing neural crest is a subset of the cephalic neural crest and is in a stream that spreads ventrally to meet under the forebrain. F: By stage 12, Cyp26C1 expression is in rhombomeres 2, 3, and 5, in the anterior neural crest, and in a patch of mesoderm adjacent to the otic placode (red arrowhead). G: By stage 14, the rhombomeric expression has changed to rhombomeres 5 and 6. H: A transverse section through the head of a stage 14 embryo to show expression in Rathke's pouch. I: A lateral view of a stage 14 embryo, showing Cyp26C1 expression in the mesenchyme adjacent to the otic vesicle (ov), in Rathke's pouch (red arrow), and the rest of the branchial endoderm (white arrow).

Download figure to PowerPoint

At stage 12, the rhombomeric expression of Cyp26C1 in 2, 3, and 5 is very clear (Fig. 2F). The expression in rhombomere 2 is particularly striking as it exactly divides the common 1/2 rhombomere in half, and to our knowledge, this is the only gene expressed in this manner. Expression continues at this stage in the rostral neural crest stream and a new site appears, in a small domain of mesoderm below the otic placode (red arrowhead in Fig. 2F).

Expression in Stages 14–20

Cy26C1 expression in these stages is found exclusively in the hindbrain and pharyngeal region. Thus, at stage 14, Cyp26C1 is expressed in the neural crest-derived cephalic mesenchyme adjacent to the hindbrain, both post-otic and pre-otic (Fig. 2I). Expression within the hindbrain has changed from rhombomeres 2, 3, and 5 (Fig. 2F) to rhombomeres 5 and 6 (Fig. 2G). It is also expressed in Rathke's pouch (Fig. 2H red arrowhead, Fig. 2I, red arrow) and in the pharyngeal endoderm (Fig. 2I, white arrow). By stage 15, the pre-otic mesenchymal expression domain has started to contract caudally and become concentrated around the otic vesicle. In the hindbrain, expression in rhombomeres 5 and 6 remains and both the endoderm of the 1st branchial pouch and the epidermis of the 1st branchial groove begin to express Cyp26C1.

At stage 18, rhombomeres 5 and 6 continue to express Cyp26C1 and the roof plate now begins expression (Fig. 3A–C, red arrowheads). The hindbrain mesenchyme expression has become concentrated into two discrete regions rostral and caudal to the otic vesicle (Fig. 3A). Sections through the hindbrain reveal how localized the expression domains are (Fig. 3B,C, black arrows). Also revealed in the sections is a new domain of Cyp26C1, which is the ventral quadrant of the otic vesicle (Fig. 3B, white arrowhead). A final region of expression at stage 18 is the endoderm of the pharyngeal pouches of arches 1, 2, and 3 and the epidermis of the corresponding pharyngeal grooves (Fig. 3A).

thumbnail image

Figure 3. Expression of Cyp26C1 from stages 18 to 22. A: Lateral view of a stage (st) 18 embryo showing expression in the hindbrain roof plate (red arrowhead), in the mesenchyme adjacent to the otic vesicle (ov) and in the branchial pouches and grooves. B: A section through a stage 18 hindbrain at the level of the otic vesicle, showing the roof plate expression (red arrowhead), the expression in the ventral part of the otic vesicle (white arrowheads in B and C), expression in discrete regions of mesenchyme beside the otic vesicle (black arrows), and within the hindbrain neuroepithelium in discrete domains. C: A section rostral to that in B showing another patch of expression in the mesenchyme (black arrow) and in the roof plate (red arrowhead), but now there is no expression in the neuroepithelium. D: Expression in the head of a stage 22 embryo showing the roof plate (red arrowheads), patches of cephalic mesenchyme, the ventral neuronal domains in the anterior rhombomeres, the nasal neural retina (white arrowhead), and a region of the telencephalic vesicles (black arrow). E: A frontal view of the embryo in D, showing the telencephalic expression of Cyp26C1 (red arrowheads). F: A flat-mounted stage 22 hindbrain, showing expression in the ventral neuronal populations of rhombomeres 2–6 but no expression in rhombomeres 1 and 7.

Download figure to PowerPoint

Expression at Stages 21–24

There are three new domains of expression of Cyp26C1 in stages post-20. These domains are a small segment of the nasal retina (Fig. 3D), a medial region of each telencephalic hemisphere (Fig. 3E), and the hindbrain rhombomeres. Rhombomeric expression is now seen in rhombomeres 2–6 in discrete neuronal populations varying in their dorsoventral location (Fig. 3F). The other three regions of expression seen earlier continue in these later stages, namely two discrete areas adjacent to the otic vesicle, which now resemble areas where cranial ganglia will differentiate, the ventral otic vesicle and the pharyngeal pouches of arches 1, 2, and 3.

Regulation of Expression by RA

To determine whether any of these expression domains depend on RA for their induction, we examined Cyp26C1 expression in the vitamin A-deficient (RA-free) quail embryo. These embryos do not contain any detectable retinoids either in the yolk or in the embryo (Dong and Zile, 1995), and the absence of retinoids results in multiple system abnormalities (Heine et al., 1985; Dersch and Zile, 1993; Maden et al., 1996, 2000; Stratford et al., 1999; Gale et al., 1999). They are a valuable resource for asking questions not only about their retinoid-dependent anatomy, but also about retinoid-dependent gene expression domains.

We first confirmed that the expression of Cyp26C1 described above in the chick embryo was the same in the normal quail embryo. We could detect no difference between these two species.

In the RA-free quail embryo, expression of Cyp26C1 in the anterior mesoderm began as in normals at stage 4 and spreads caudally to the level of the first somite (Fig. 4A). The change in expression domain to the neural crest from rhombomeres 2 and 3 also occurred on time at around stage 10 (Fig. 4B). However, a closer examination of these embryos by sectioning revealed that, although the neural crest strongly expressed Cyp26C1, the neuroepithelium of rhombomeres 2 and 3 did not. Figure 4C shows a section just rostral to rhombomeres 2 and 3 in which the neuroepithelium is devoid of expression, whereas the neural crest and ectoderm is intense. In Figure 4D, the section is through rhombomeres 2 and 3 and the dorsal neural tube from which the neural crest arises expresses Cyp26C1, as does the crest itself and the ectoderm, but the remaining three quarters of the neuroepithelium does not. This section should be compared with Figure 2C, which is the equivalent from a normal embryo.

thumbnail image

Figure 4. Expression of Cyp26C1 in retinoic acid (RA) -free quail embryos. A: At stage (st) 8, the expression is identical to the normal embryo (compare with Fig. 1C). B: At stage 10−, expression looks identical to normal (compare with Fig. 2A) with rostral neural crest streams but, in fact, there is no expression in the neuroepithelium (see C,D). C: A section through the rostral rhombomere A of the embryo in B showing expression in the neural crest but nothing in the neuroepithelium. D: A section caudal to that in C, at the level of the posterior rhombomere A, showing expression in the neural crest and dorsal neuroepithelium from which the neural crest arose but no expression in the rest of the neuroepithelium. E: Lateral view of a stage 14 embryo showing that the mesenchymal and endodermal expression is normal (compare with Fig. 2I) but that the hindbrain fails to show any expression. F: A section through a stage 20 RA-free embryo showing expression in the mesenchyme adjacent to the hindbrain but no expression within the neuroepithelium itself. G: A stage 20 RA-free embryo with the cephalic mesenchyme dissected away to show the lack of expression of Cyp26C1 in the hindbrain.

Download figure to PowerPoint

In stage 14 RA-free embryos (Fig. 4E), the pre-otic mesenchyme and the pharyngeal endoderm expresses Cyp26C1 but not the neuroepithelium of the hindbrain (compare with Fig. 2I). This situation continues throughout the remaining stages, and by stage 20, sections reveal Cyp26C1 expression in the mesenchyme but not the neuroepithelium (Fig. 4F) and an embryo from which the mesenchyme has been dissected away (Fig. 4G) shows clearly the absence of expression in the neuroepithelium compared with the normal expression in rhombomeres 2–6 (Fig. 3F).

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

We have cloned a novel cytochrome P450 enzyme of the CYP26 family, called Cyp26C1, in the chick embryo following on from its identification in the mouse (Tahayato et al., 2003). It has a different expression pattern from the mouse embryo. The latter can be summarized as rhombomeres 2 and 4, and the mesenchyme lateral to these rhombomeres, the 1st pharyngeal arch ectoderm, a portion of the otic vesicle, and the mesenchyme caudal to the otic vesicle. Cyp26C1 expression is far more extensive in the chick embryo and starts much earlier. It includes anterior (future cephalic) mesenchyme beginning at early gastrulation stages; head neural crest; rhombomeres 2, 3, 5, and subsequently neuronal populations in rhombomeres 2–6; the hindbrain roof plate; mesenchyme adjacent to the otic vesicle; the ventral otic vesicle; neural retina; and anterior telencephalon.

Comparisons Between the Cyps

The other CYP26s also differ in the details of their expression patterns between the chick and the mouse. For example, Cyp26A1 is expressed in the lateral vascular networks, rhombomere 2, the neural retina, and the maxillomandibular cleft of the mouse embryo (Fujii et al., 1997; McCaffery et al., 1999; MacLean et al., 2001), but in the chick embryo, it is expressed in rhombomere 3 (not r2), the dorsal lens (not neural retina), the 1st pharyngeal groove (not the maxillomandibular cleft), and the lateral vascular networks express Cyp26B1 not A1 (Swindell et al., 1999; Blentic et al., 2003; Reijntjes et al., 2003). Similarly with regard to Cyp26B1, it is expressed, amongst other places, in rhombomeres 3 and 5 and in the ectoderm and mesoderm of the limb bud in the mouse (MacLean et al., 2001), whereas in the chick, it is expressed in rhombomeres 4, 6, and the rhombic lip, in the heart and vasculature, the neural retina, and the distal mesenchyme of the limb bud (Reijntjes et al., 2003).

It is important, therefore, to compare CypA1, B1, and C1 expression patterns within one species and, to this end, we summarize their domains in the early chick embryo in Figure 5. It is clear from this figure that there are very few overlapping domains as each chick CYP has an almost unique expression pattern. The domains of Cyp26A1 can be summarized as neural tube, tail bud, lens, trunk endoderm, limb bud ectoderm; that of Cyp26B1 as placodal mesoderm, early tail bud, trunk vasculature, limb bud mesoderm, hindbrain, neural retina; and that of Cyp26C1 as cephalic mesoderm, neural crest, placodal mesoderm, hindbrain, pharyngeal pouches, and grooves. There are, therefore, only very brief periods of some overlap—in the tail bud, the mesenchyme lateral to the hindbrain, the 1st pharyngeal pouch and some neuronal populations in the hindbrain.

thumbnail image

Figure 5. Drawings of embryos at representative stages (st) to summarize the expression domains of Cyp26A1 (red), Cyp26B1 (green), and Cyp26C1 (blue).

Download figure to PowerPoint

From these distributions, we can make hypotheses as to the functions of the individual CYPs. Cyp26A1 could function in cephalic neuroepithelium differentiation, dorsal neural tube patterning, or tail bud formation. These possibilities have been tested (albeit in mouse) by generating null mutant embryos (Abu-Abed et al., 2001; Sakai et al., 2001). In the hindbrain, there are slight expansions of some gene expression such as Hoxb-1, EphA2, slight constrictions of others such as Krox-20, Meis2, and minor abnormalities in the trigeminal nerve, but anatomically the head is normal. Cyp26A1, being expressed transiently in rhombomere 3 (in the chick embryo) could, therefore, function to keep RA levels low there because the effects on the trigeminal nerve are reminiscent of (although somewhat weaker than) the effects of excess RA on late hindbrain development (Gale et al., 1996). No observations have been made on the dorsoventrality of the neural tube in these null mutants, but there are major deficits in the tail bud ranging from sirenomelia, loss of hindlimbs, posterior gut, and urinogenital abnormalities to less severe tail abnormalities (Abu-Abed et al., 2001; Sakai et al., 2001). These posterior defects are very similar to the effects of excess RA; therefore, it is most likely that the major function of Cyp26A1 is to keep the RA concentration low in the tail bud and that it serves only a minor role in head development. We propose below that the latter function is performed by Cyp26C1.

For Cyp26B1, we would propose a function in vasculogenesis, limb bud patterning, and the differentiation of individual hindbrain neuronal populations, including cerebellum development. The null mutant mouse phenotype has not yet been reported, and we are currently testing these hypotheses in the chick embryo in overexpression and siRNA studies. For Cyp26C1, we would propose a function in head development (see below) because its major expression site is in the early cephalic mesoderm, in a subset of cephalic neural crest, in pharyngeal arch patterning and in hindbrain neuronal populations. Again, the null mutant mouse phenotype is not yet available.

Regulation of the Cyps by RA

Because the substrate for these CYP enzymes is RA, it is interesting to consider whether their expression is regulated by the substrate. This we have considered by examining their expression pattern in the absence of RA, that is in the RA-free quail embryo. Some Cyp domains are unaffected in this situation, and some fail to appear. For Cyp26A1,the anterior neural expression is unaffected by the absence of RA, but the trunk expression (dorsal neural tube, tail bud) fails to appear (Blentic et al., unpublished data). For Cyp26B1 again, the head expression is normal (lateral hindbrain mesenchyme, rhombomeres), but the trunk expression in the vasculature and heart is completely absent (Reijntjes et al., 2003). For Cyp26C1, we showed here that the neural crest, lateral hindbrain mesenchyme was perfectly normal but that the rhombomeric expression was absent in the absence of RA (Fig. 4). Thus, individual expression domains of these enzymes are regulated by RA: dorsal neural tube, tail bud (Cyp26A1), trunk vasculature (Cyp26B1), rhombomeres (Cyp26C1), suggesting both vital and individual roles for these genes in these locations.

Cyp26C1 and Head Development

It is surprising that the Cyp26A1 null mutant mouse (Abu-Abed et al., 2001; Sakai et al., 2001) had minimal hindbrain patterning defects, because the main hypothesis for the function of this CYP was to generate a gradient of RA across the hindbrain neuroepithelium (Swindell et al., 1999; Maden, 1999). The RA synthesizing enzyme Raldh2 is expressed in the mesoderm with a rostral border at the level of presumptive somite 1, and the RA catabolizing enzyme Cyp26A1 is expressed rostrally with a caudal border at the level of the presumptive rhombomere 3/4 (Blentic et al., 2003). It has been proposed that the apposition of such a source (RALDH2) and a sink (CYP26A1) could generate a gradient of RA between the two domains, a gradient which would provide patterning information for the development of the posterior hindbrain. In support of this idea, when the source of RA is removed either by removing the RA itself (Maden et al., 1996; Gale et al., 1999) or by removing Raldh2 (Niederreither et al., 1999), then the posterior hindbrain fails to develop. However, as emphasized in Figure 6A, Cyp26A1 is expressed in the anterior epiblast (presumptive forebrain, midbrain, and anterior hindbrain of the neural plate) and so the proposed gradient would have to form between the mesoderm and the overlying neuroepithelium (Fig. 6A, blue arrows). This seems an unlikely scenario and may explain why the head of the Cyp26A1 null mutant mouse is essentially normal (Abu-Abed et al., 2001; Sakai et al., 2001). We would expect that, in the absence of the sink, RA would flood through the anterior neuroepithelium and generate a phenotype equivalent to excess RA administration, i.e., missing forebrain and midbrain (Simeone et al., 1995; Avantaggiato et al., 1996) or loss of anterior hindbrain (Morriss, 1972; Morriss-Kay et al., 1991; Cunningham et al., 1994; Leonard et al., 1995). This clearly does not happen.

thumbnail image

Figure 6. A,B: Representation of the neural plate and mesenchyme below of an early embryo to show the relative expression domains of Cyp26A1 and Raldh2 (A) and Cyp26C1 and Raldh2 (B). The neural plate is divided into three sections, represented by forebrain (fb), midbrain (mb), hindbrain (hb), and spinal cord. A: The expression domain of Cyp26A1 is shown in yellow in the neural plate and that of Raldh2 in the posterior mesoderm below is shown in red. If a gradient of retinoic acid (RA) is generated between these two domains, then it would have to turn at right angles as shown by the blue arrows. B: The expression domain of Cyp26C1 in the anterior mesoderm is shown in green and that of Raldh2 in the posterior mesoderm is shown in red. It is much more conceivable that a gradient of RA could be generated between these two domains, which could then signal to the overlying neural plate (blue arrows). In this region of the neural plate the posterior hindbrain will develop. This is the part of the embryo that is missing in the absence of RA or Raldh2 (see text). C: Representation of the RA gradient, which could form within the mesoderm. D: A double in situ hybridisation of a stage 6 embryo to show the relationship between the anterior mesodermal Cyp26C1 expression (turquoise) and the posterior mesodermal Raldh2 expression (purple) with a gap in between. Across this gap, a gradient of RA could be generated as elaborated in C.

Download figure to PowerPoint

However, on the basis of the results described above where Cyp26C1 is expressed in the same anterior domain as A1 but in the mesoderm rather than the neural plate, it is much more likely that Cyp26C1 is the important CYP in head development. It could clearly be the sink of the hypothesized RA gradient, because it is expressed in the mesoderm with a gap between itself and RALDH2 (Fig. 6B,D). The RA gradient (Fig. 6C), therefore, would be generated within the mesoderm, which would then send patterning signals to the overlying neuroepithelium (Fig. 6B, blue arrows). If this hypothesis is valid, then it predicts that the Cyp26C1 null mutant mouse will have a forebrain or hindbrain phenotype equivalent to administering excess RA to the embryo. We await the publication of these results.

EXPERIMENTAL PROCEDURES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

Embryos

Fertilized hens' eggs (mixed flock) and fertilized quail's eggs were obtained from a local farm supplier. RA-free quails eggs were obtained from a colony kept at King's College London. The adults in this colony were fed a diet completely devoid of any vitamin A except for the addition of 10 mg/kg all-trans-retinoic acid to the diet. All embryos were incubated at 37°C and staged according to Hamburger and Hamilton (1951).

Isolation of the Chicken Homologue of Mouse Cyp26C1

RNA from Hamburger and Hamilton stage 4, 9, 12, 16, and 28 (st4, st9, st12, st16, and st28) chick embryos was isolated by the guanidinium thiocyanate method of Chomczynski and Sacchi (1987). A cDNA pool from each stage was prepared by using M-MuLV reverse transcriptase (Amersham Pharmacia Biotech, Buckinghamshire, UK) per the manufacturer's instructions.

The mouse Cyp26C1 nucleotide sequence (Tahayato et al., 2003) was used to search the BBSRC Chicken EST Database (Boardman et al., 2002), and clone ChEST1007a3 was identified. The sequence homology between these two was 84%. The sequence of ChEST1007a3 was used to design primers to isolate Cyp26C1. The following primers were used for PCR: sense, 5′-TGCTGCTTTTTTCACCACGG-3′; antisense, 5′-TGTGTCACGGATGCTGTACA-3′ to generate a 514-bp product.

The obtained PCR products from st28 were subcloned into pGEM-T Easy vector (Promega, Madison, WI). DNA sequencing was performed by the dideoxynucleotide chain termination method (Sanger et al., 1977) with ABI PRISM 377 DNA sequencer.

Whole-Mount In Situ Hybridization

Whole-mount in situ hybridization was carried out by using standard procedures. Double in situ hybridization was carried out by using cohybridization of a digoxigenin-labeled probe with a fluorescein-labeled probe. First, the fluorescein probe was detected with an anti-fluorescein antibody conjugated to alkaline phosphatase and was visualized by exposure to BCIP producing turquoise labeling. The alkaline phosphatase from the first reaction was quenched with 4% paraformaldehyde, and the embryos were then exposed to a digoxigenin antibody and colour reacted with BM Purple producing the purple labeling. Digoxigenin- (Roche) and fluorescein- (Roche) labeled antisense RNAs were prepared by in vitro transcription of the linearized plasmid for Cyp26C1. Other probes were kindly given by E. Swindell (Cyp26A1) and V. Prince (Hoxb-1). After in situ hybridization, embryos were photographed and then embedded in gelatin type A for sectioning. Vibratome sections were cut at 80 μm.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

We thank Drs. Martin Petkovich and Ali Tahayato for the mouse Cyp26C1 sequence data before publication, the MRC for the provision of a studentship (S.R.) and the BBSRC for research funding (E.G., M.M.).

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES