Retinoic acid-induced caudal regression syndrome in the mouse fetus

Reproductive Toxicology, Vol. 12, No. 2, pp. 139-151, 1998 0 1998 Elsevier Science Inc. Printed in the USA. All rights reserved 0890-6238/98 $19.00 + .tM ELSEVIER

PI1 SOS90-6238(97)00153-6

RETINOIC

ACID-INDUCED

CAUDAL REGRESSION MOUSE FETUS

SYNDROME

IN THE

R. PADMANABHAN Department of Anatomy, Faculty of Medicine and Health Sciences, UAE University, Al Ain, United Arab Emirates Abstract - Caudal regression syndrome (CRS) comprises developmental anomalies of the caudal vertebrae, neural tube, urogenital and digestive organs, and bind limbs, the precursors of all of which are derived from the caudal eminence. Although the syndrome is well recognized, the etiology and pathogenetic mechanisms are poorly understood Genetic and experimental models may provide some important clues to the early events that precede the dysmorphogenesis in CRS. The objectives of this study were to determine the susceptible stages for induction of CR!? and to ascertain the early events that precede the development of this syndrome in a mouse model. Single orad doses of 100, 150, or 200 mg/kg retinoic acid (RA) were administered to TO mice on one of Gestation Days (GD) 8 to 12, and fetuses were observed on GD 18. All doses administered on GD 8 or 9 resulted in CRS in a large number of survivors. Agenesis of the tail, caudal vertebral defects, spina bifida occulta/aperta, imperforate anus, rectovesicle or rectourethral fistula, renal malformations, cryptorchidism, gastroschisis, and limb malformations, including the classical mermaid syndrome (sirenomelia), were characteristic features of this animal model. Several craniofacial malformations accompanied CRS in the GD 8 treatment group. Chronologic examination of treated embryos at early stages revealed pronounced cell death in the caudal median axis, hindgut, and neu!ral tube and consequently, failure of development of the tail bud in the high-dose groups. In the 100 mgkg RA group, patches of hemorrhage occurred initially that subsequently coalesced into large hematomas and the tail progressively regressed. Histologic examination revealed the onset and progression of hemorrhage, edema, and cell death in these embryos. Transillumination and histologic preparations also revealed dilation of the caudal neural tube in the prospective CRS embryos. Thus, a combination of cell death, vascular disruption, and tissue deficiency appears to be the highlight of caudal regression in this model. Symmelia appeared to be due to failure of fission or due to the merger of limb fields rather than a result of fusion of two limb buds. The data are also indicative of caudal agenesis in the high-dose RA groups and caudal regression due to a combination of 0 1998 Elsevier vascular disruption, edema, and cell death in the lower dose groups of TO mouse embryos. Science Inc. Key Words: rn~use retuses; TO strain; caudal regression syndrome; retinoic acid; limb and tail defects; vertebral defects; spina bifida; urogenital anomalies; vascular disruption.

malformations;

mythologic monsters (1). This complex affliction presents itself in variable forms determined largely by the organs affected; the visceral malformations are consistent and contribute to mortality, whereas sympodia is the least grave and most variable (1). The incidence is 1 in 60,000 births (2,3). The frequency is high (8 to 15%) in twinning and it is 100 to 150 times more common in monozygotic twins than in dizygotic twins or singletons (4). A vascular etiology has been suggested (5) but no definitive pathogenetic mechanism has been established. Several clinical and experimental studies have attempted to clarify some of the issues of mechanisms but there appears to be no agreement on any of them (2,6-12). Both genetic and experimentally induced animal models are of great help in identifying the pathogenetic mechanisms by circumventing extraneous variables. Kochhar (13) and Yasuda et al. (14) made a passing reference to the mermaid appearance of one of their retinoid-treated rat embryos. Hashimoto et al. (15) de-

INTRODUCTION

of caudal regression” Writing about the “syndrome Duhamel (1) lists (a) flexion and rotational defects of the lower limbs, (b) anomalies of the lumbosacral spine, (c) imperforate anus, (d) agenesis of kidneys and urinary tract, and (e) agenesis of internal genital organs with the exception of the gonads as constituting the developmental disorder now variously called caudal dysgenesis, caudal regression, sacral agenesis, and mermaid syndrome. Merging of th.e lower limbs (sympodia or symmelia) in some cases of caudal dysgenesis syndrome has led to the use of the fanciful term mermaid syndrome, a concept based on the resemblance to Greek and Roman

Address correspondence Anatomy, Faculty of Medicine PO Box 1766, Al Ain, United Received 4 August 1997; Accepted 29 November 1997.

visceral

to R. Padmanabhan, Department of and Health Sciences, UAE University, Arab Emirates. Revision received 28 November 1997;

139

Reproductive

140

Toxicology

scribed anorectal malformations in similarly treated mouse embryos but made no mention of the mermaid or caudal regression syndrome. Caudal regression has been reported recently in isotretinoin-treated mouse embryos (16) and ochratoxin A-treated chick embryos (17). Gardner (18) hypothesized that overdistention of the caudal neural tube might lead to spina bifida and several non-neural malformations, including mermaid syndrome. We had established previously an animal model of overdistention of the closed cranial neural tube and studied the pathogenesis of postclosure exencephaly and associated non-neural malformations (19 -2 1). In our preliminary attempts to induce similar neural tube closure defects in the TO mouse, we observed serendipitously a consistent incidence of agenesis of the tail and caudal regression, including spina bifida when mothers were treated with single doses of all trans-retinoic acid (RA) on Gestation Days (GD) 8 or 9. The caudal defects appeared to conform to Duhamel’s (1) description of the syndrome of caudal dysgenesis. The objectives of the present study were to (1) determine the susceptible stages for induction of caudal regression; (2) to characterize the internal and external anomalies associated with caudal regression; and (3) to ascertain the early events that precede the regression in our mouse model maternally exposed to RA.

MATERIALS

AND METHODS

The TO strain of mice used in this study were originally obtained from Harlan Olac, London and raised in our local facility. They were housed in light (12:12 h 1ight:dark cycle)- and temperature-controlled (2 1 ? 1“C) rooms and maintained on laboratory chow and tap water provided ad libitum. Adult virgin females (about 6 weeks of age and 30 g in weight) were mated overnight with males of the same stock. A vaginal plug observed on the following morning indicated Day 0 of pregnancy. Single doses of 100, 150, or 200 mg/kg of all trans-retinoic acid (a kind gift from Dr. H. Hummler and Dr. C. Dobson of La Roche, Basel, Switzerland) suspended in alcohol:com oil (1:9) mixture, were administered orally to groups of mice on one of GD 8 to 12. The controls were vehicle treated. The animals were killed on GD 18 by cervical dislocation. Fetuses were euthanized by ether, weighed, fixed in 95% ethanol, examined for external and internal malformations by a modified method of Sterz and Lehmann (22), and subsequently processed for alizarin red-S and/or alcian blue staining (23). Groups of mice treated with 100,150, or 200 mg/kg of RA on GD 8 were also sacrificed at 6-, 12-, and 24-h intervals, and the successive stages of regression of the caudal end of the embryos were examined. Three to five such embryos of the 200 mg/kg RA-treated group on GD 8 together with

Volume

12, Number 2, 1998

age matched vehicle-treated control embryos of each stage were fixed in a glutaraldehyde-paraformaldehyde mixture and embedded in araldite, and semithin sections cut with an ultramicrotome were stained with toluidine blue and examined with a microscope. Embryos of the 100 mg/kg group treated on CD 8 developed hematomas in the tail on GD 9. These embryos together with age-matched controls were fixed in Gender’s fluid, embedded in paraffin, sectioned serially at 7 pm thickness and stained with hematoxylin and eosin. Data were analyzed by ANOVA using litter as the unit. Multiple comparisons were performed by Scheffe’s test (24). Significance was assumed at P < 0.05. OBSERVATIONS Maternal effects No dose of RA employed in this investigation caused any overt maternal toxicity as evidenced from food and water consumption and maternal survival rate. Any failure to gain in body weight during gestation was found to be related to a proportionate increase in embryonic resorption, particularly in those mice treated on GD 8 or 9. Fetal effects Treatment with all three doses (100, 150, and 200 mg/kg body weight) of RA on GD 7 or earlier resulted in complete resorption of the embryos (Padmanabhan, 1997, unpublished data). Reduction in fetal body weight was both dose and developmental stage dependent (Table 1). Although the treated fetuses, particularly those of the 150 and 200 mg/kg RA groups, were growth retarded, ANOVA did not show the reduction in body weight to be significant. All doses were teratogenic from GD 8 through 12, and the severity of malformations appeared to increase with increase in the dose. Treatment on GD 8 and 9 caused a greater resorption than that on GD 10 to 12 (Table 1). Exencephaly, microcephaly, exophthalmia, astomia, microstomia, anotia, agnathia or micrognathia, aglossia/microglossia, agenesis or partial development of the palate, median clefts of the lips, and craniofacial skeletal defects were characteristic of GD 8 RA-treatment (Figure 1). Cleft palate occurred in all treatment groups; earlier stages of treatment seemed to result in severe intrauterine growth retardation and affect the development of the palatal shelves whereas the embryos treated at later stages exhibited delayed shelf elevation and failure of apposition rather than agenesis of the shelves. Characteristics

of caudal regression

syndrome

Tail and urogenital defects. The incidence of caudal regression syndrome (CRS) in near-term fetuses was 100% in all the three dose groups treated on GD 8 or 9; it was characterized by either a total absence or extreme

Pathogenesis

Table

Treatment Vehicle

Retinoic acid 100 mg/kg

150 mgikg

200 mgkg

1. Effect

Day of treatment (No. of litters) 8 (22) 9 (24) 10 (21) 11 (20) 12 (25) 8 (12) 9 (10) 10 (8) 11 (12) 12 (12) 8 (10) 9(11) 10 (9) 11 (10) 12 (10) 8 (12) 9 (10) 10 (10) 11 (12) 12 (10)

of maternal

exposure

Implants 12.53 13.53 12.50 12.333 12.466 12.0 12.0 13.1 11.5 11.25 13.0 13.1 13.3 13.8 12.5 12.41 12.00 11.10 11.5 11.7

* 1.11 2 0.97 2 1.24 + 1.093 2 1.320 ? 1.34 ? 1.56 2 1.24 2 1.24 2 1.28 2 0.94 k 0.98 c 1.00 -c 1.03 i 1.50 2 1.88 2 1.49 2 1.10 -+ 1.62 r 1.25

of caudal regression

to a single

Resorptions* 0.7 1.43 1.10 0.266 0.80 4.0 2.0 1.12 1.0 1.7 5.1 2.7 2.5 2.5 1.6 7.0 3.3 2.3 2.0 1.9

+ 0.85 rfr 1.09 k 0.98 2 0.403 k 0.803 2 1.20 2 1.15 + 0.99 + 0.85 k 0.45 t 0.99 2 1.27 ? 0.52 k 1.35 + 0.96 5 0.95 + 1.25 + 0.82 -c 1.34 + 1.10

Live fetuses* 11.80 12.10 11.36 11.80 11.623 6.0 10.0 11.87 10.5 9.5 7.9 10.4 10.6 11.1 10.9 5.4 8.7 8.8 9.5 9.5

ik t -c ? k -c ? ?I 2 5 t 5 2 + ? ? ?I 2 *

141

R. PADMANABHAN

dose of RA on one of GD 8-12

All values are mean + SD. *P < 0.001 when treated group means are compared with the control. The vehicle control data of different dose groups did not differ significantly combined control data and those of the experimental groups.

deformation of the tail depending on the dose and time of administration of RA (Figures 1 and 2). For example, in the GD 8-100 mg/kg RA group, the tail was absent in 76%, thin and withered in 14.%, and rudimentary (=pig-tailed) in 10% of the embryos. In the 150 mg/kg RA group, agenesis of the tail occurred in about 95%, and the remainder had thin and withered tails. In the 200 mg/kg RA group, no fetus developed any trace ol; a tail at all. The lower dose of RA administered on GD 9 produced rudimentary and withered tail whereas the higher doses resulted in near complete regression of the tail. A large hematoma occupied the tip of the stump (Figure 2F)‘. Treatment on GD 10, 11, and 12 resulted in a significant reduction in the tail length but not in obvious tail malfonmations (Table 1). Pig-tail/rudimentary tail, duplication, and agenesis of the tail observed in GD 8 treatment groups were almost invariably associated with anal agenesis. Razor blade sections of CRS fetuses revealed continuation of the lower rectum with the urethra and the absence of the anal canal (Figure 3E). In the female, the vagina was absent and the uterine horns were attached to the lateral or dorsolateral aspect of the rectourethral junction. Cryptorchidism was very common. Extreme degrees of renal hypoplasia, large cysts and agenesis, both uni- and bilateral nonascent, and occasionally partial or complete fusion of the kidneys were observed in most of the fetuses of GD 8 treatment groups. GD 11 and 12 were not susceptible to gross urogenital anomalies. The bladder appeared to be of normal size in most of the GD 8 and 9 groups, but megabladder was commonly

l

0.9166 0.79 0.716 1.030 0.930 0.85 1.56 1.12 0.79 1.16 0.3 1 0.82 1.11 0.56 0.87 1.50 1.15 1.13 0.79 1.08

Fetal weight (g) 1.47 1.45 1.49 1.45 1.47 1.32 1.10 1.46 1.44 1.41 0.96 0.92 1.25 1.20

2 0.10 + 0.15 + 0.126 + 0.086 + 0.110 2 0.15 5 0.28 rt 0.14 ? 0.12 2 0.07 2 0.22 k 0.24 ? 0.15 k 0.12 1.30+- 0.08 1.06 + 0.11 1.08 k 0.12 1.34 2 0.15 1.32 2 0.12 1.36 2 0.10

from one another

in the TO mouse

CR length (mm)* 24.15 23.59 23.48 24.276 24.203 20.13 23.59 22.59 21.27 22.10 21.41 22.74 21.87 21.25 21.88 15.54 21.79 22.87 24.00 22.62

and therefore

k k k k 2 ? 2 k 2 I? ? ? t k 2 -c t i t r

1.13 1.31 1.46 0.806 1.040 1.38 1.27 1.62 1.62 0.76 2.11 2.36 1.87 1.41 0.63 1.92 1.55 1.42 1.56 1.55

comparisons

Tail length (mm)* 11.80 11.22 11.36 10.96 11.01

8.24 10.45 10.53

4.66 9.27 8.92

5.06 9.04 9.20

? t k k 2 0 0 2 2 k 0 0 -t -c + 0 0 t ? +

0.53 0.60 0.54 0.593 0.453

2.08 1.07 0.36

1.57 0.67 0.35

1.05 1.17 0.62

were made between

observed in groups treated on GD 10. The urethral atresia described often to coexist with megabladder could not be demonstrated by razor blade sectioning. Neural tube defects. All fetuses with tail defects also had their lower spinal cord abnormally developed. At times both the cranial and caudal neuropores remained open. Both spina bifida aperta (8%) and spina bifida occulta (92%) were observed in “tailless” fetuses (Figure 4). In the aperta cases, the neural tube was open over a varying extent, everted, and overgrown. The lower lumbar and or the entire sacral regions were involved. The neural tissue was often visibly hemorrhagic. Overgrowth and hemorrhage also extended into the abdominal cavity on one or both sides of the vertebral column. In the occult type, the cord was cystic, hemorrhagic, and remained unprotected by the open vertebral arches and only covered by skin (Figure 4F). In these cases, the vertebral bodies ended high at the lower or mid lumbar region. Below this level, there was very little skeletal protection and the vertebral laminae, when present, were everted and fused longitudinally. In the GD 9 treatment group (200 mg/kg RA), whereas most of the tail was absent, the stump was bulbous with a hematoma and the caudal neural tube lacked skeletal protection. Limb-body wall defects. In the embryos of the 100 mg/kg of RA group treated on GD 9, both the fore- and hind limbs were externally normal; in the 150 and 200 mg/kg RA groups treated at the same stage, there was a low incidence of meromelia and reduction deformities of

142

Reproductive Toxicology

Volume 12, Number 2, 1998

Fig. 1. Mouse fetuses of CD 18 (200 mgkg of RA, treated on GD 8) showing total agenesis of the tail (A-F), craniofacial malformations such as exophthalmia (A,B,E), astomia (A,D,E), hypoplasia of the lower jaw, and reduction deformities of the limbs (A-E). The hind limbs are only represented by the feet and often a single foot (B,C,D). The roots of such limbs are merged with the genital tubercle which is located at the caudal tip (arrows in A,D,F). The umbilicus (open arrows in C and E) appears to be situated near the genital tubercle due to the regression of the caudal trunk. The abdomen of fetus in D was cut open for internal examination

before photographing. the digits (RDD) of both limbs. Treatment on GD 10 gave rise to predominantly forelimb meromelia and RDD. Both fore- and hindlimb meromelia and RDD occurred in the GD 11-RA-treated fetuses. GD 12 was sensitive to RA-induced meromelia of moderate degree, RDD (affecting the marginal digits mainly), and nail hypoplasia. The limb defects in the GD 8-RA group were a part of the syndrome of caudal regression and involved predominantly the hind limbs (Figures 1 and 2). The severity of the defects appeared to be dose dependent. In the 200 mg/kg RA group, the hind limb roots were found to be located almost invariably in the perineum, close to and sometimes on the genital swelling or genital tubercle (Figure 1). The genital tubercle was situated at the caudal

tip rather than in the infraumbilical position between the limb buds. This approximation was accompanied by extreme meromelia, varying degrees of fusion of the proximal segments, malrotation, cleft foot, oligodactyly, and polydactyly. Malrotation involved one or both sides. In some cases, the rotation was completely reversed so that the feet faced backwards (Figures 1,2, and 4). In 6% (4/65) of cases, the classical mermaid syndrome of fusion of the entire hind limbs was observed (Figure 2, D and E). Gastroschisis and prominent hematomas were also present in some CRS fetuses. In the lower dose groups, the body wall defects were small and only some loops of small intestine were protruding. In the 200 mg/kg group, in addition to the intestine, the liver,

Pathogenesis of caudal regression

l

R.

PADMANABHAN

143

Fig. 2. Mouse fetuses of GD 18 (200 mg/kg of RA treated on GD 8) showing caudal regression characterized by complete absence of the tail (B and C), classical sirenomelia (fusion of the entire hind limbs as in D or of the feet as in E), and unipodia with absence of the thigh and leg and the foot facing backwards (C). Microcephaly and micrognathia are also observed in such fetuses. F is an example of stubby tail with hematoma of the GD 9-200 mg/kg RA group. A, control.

stomach, and spleen also protruded through the large median fissure that extended to the caudal tip in the infraumbilical part of the ventral abdominal wall. Apodia and unipodia (lateral and median) with RDD, often with

malrotation, were also observed. All the embryos with CRS also exhibited a unique .pattem of axial and limb skeletal malformations, characteristic of their external appearance.

Fig. 3. Razor blade sections of fetuses of GD 18. C, control with well-developed urethra (U) issuing from the bladder (B) and rectum (R) situated dorsally. In the experimental fetus (200 mg/kg of RA treated on GD 8), the rectum and urethra have a common passage externally (E).

144

Reproductive Toxicology

Volume 12, Number 2, 1998

Fig. 4. Neural tube defects in CR.5 fetuses of GD 18 (200 mg/kg of RA treated on GD 8). Note that varying lengths c,f the luml Josalcral segment of the neural tube are open, everted, and overgrown (A-E). In some, the neural tube is open at both encis (D and Eh and the exposed neural tissue is haemorrhagic (arrows in E). F is an example of spina bifida occulta, most commonly obse rved in (ZRS fetuses. Fetus A and D are apodic, and B, C, and F are unipodic. Note that the digits are dorsally turned in E.

Skeletal defects associated with CRS. Alizarin red (with or without alcian blue)-stained skeletal preparations revealed the well-formed craniofacial skeleton, the vertebral column, and the pelvic girdle and limb bones in the control embryos. About 55% of them had 7 to 12 tail vertebrae and the remainder had 4 to 6; there were 7 cervical, 13 thoracic, 5 (26%) or 6 (74%) lumbar vertebrae, and 6 sacral vertebrae (Figure 5). In fetuses with caudal regression, the vertebral column was remarkably short due to complete agenesis of the coccygeal and most of the sacral vertebrae; the remaining lumboscaral vertebrae exhibited longitudinal fusion of the arches and bodies, malposition of the centra, uni- and double hemivertebrae, and gross dilation of the lower vertebral canal in keeping with the spina bifida occulta. In many fetuses, even the lower lumbar vertebral bodies were absent. These effects were generally dose dependent. For exam-

ple, in the 100 mg/kg RA group treated on GD 8, the coccygeal vertebral arches and bodies were absent and the lower lumbar and sacral arches were fused longitudinally. There were occasional lumbar and sacral hemivertebrae. As the dose was increased, agenesis involved more proximal vertebrae and the lower thoracic, lumbar, and sacral arches tended to fuse longitudinally. The incidence of uni- and double hemivertebrae exhibited a similar trend. In all treatment groups, the lower vertebral canal was generally dilated, and the longitudinally fused arches appeared to be ever-ted, thus giving the canal the shallow appearance of an empty table spoon (Figure 5, I3 and C). The centra were wide apart in this region. The dysgenesis/agenesis of the lower vertebral column had allowed the constituents of the hip bones to come close to each other; the ilium was extremely hypoplastic or absent. The ischium and pubis, also hypoplastic, were

Pathogenesis of caudal regression

l

R.

PADMANABHAN

145

Fig. 5. Alizarin-stained :skeletons of CRS fetuses of GD 18. A and E, dorsal and lateral view of control fetuses, respectively, showing well-formed ribs, pelvic bones (i, ilium; is, is&urn; p, pubis), lumbosacral and tail vertebrae, and hind limb bones. B-D and F, fetuses of 200 mg/kg-RA group treated on GD 8. Observe a gross reduction in the length of the vertebral column, fusion of caudal ribs, longitudinal fusion of the lumbosacral vertebral arches (B-D), absence of tail vertebrae (B-D and F), dilation of the lumbosacral spinal canal (B,C), and double hemivertebrae (arrow in B). Bones of hind limbs are extremely hypoplastic in B and F, and the pelvic girdle is absent in F. positioned

in the paramedian

position.

The femur

and

fibula were often absent. When present, these bones had moved caudally and ventrally to the bottom of the embryo in keeping with the external appearance of the limbs. Reduction in the total number of ribs (from the normal 13 pairs to 10 to 12 pairs), fusion of lower thoracic ribs, a high frequency of lumbar ribs, and eight sternal ribs instead of the usual seven were associated with caudal regression (Figure 5B, C, D). Chronologic ob:iervation. In the control embryos of GD 10, the hind lirnb buds appeared as oval surface

elevations in the lumbosacral region. The tails extended significantly beyond this level (Figure 6A). At this stage, the hind limb buds of the embryos of the 100 mg/kg RA group treated on GD 8 formed two fine surface thickenings on either side of the tail bud. In fact, the lateral margin of the tail bud and the caudal margins of the limb buds of both sides were continuous with each other. In about 95% of them, there were one or two constrictions in the tail; focal hemorrhages were located bilaterally on the ventral aspect of the tail (Figure 6 B and C). Hematomas were also prominent in the frontonasal and maxillary swellings. By GD 11, the patches of hemor-

146

Reproductive

Toxicology

Volume 12, Number 2, 1998

Fig. 6. Mouse embryos

treated with RA on GD 8 and photographed on GD 10. A and E, control. The tail is of considerable length, and the hind limb bud is situated in the lumbosacral segment (A). The hematoma in the tail of the embryo of the 100 mg/kg RA group is prominent (B), the rudimentary tail of the embryo of the 15O/kg RA group presents a hematoma distal to the constriction (C), and the tail bud is absent and the hind limb bud is projecting distally in the 200 g/kg RA group (D). The caudal spinal neural tube appears dilated when transilluminated (arrow in F) in contrast to the age-matched control (E). rhage in the tail had coalesced The proximal

segments

into a flame-shaped

of the tail appeared

rather

pool. pale

in vascularity. By GD 12, most of the distal segment was occupied by hematomas and the proximal segment seemed to be withered. The regression continued and by GD 16, and except for occasional skin tags, the tail had completely degenerated. The hind limb roots were now situated almost at the caudal tip and nearer to the genital swelling. In those embryos with spina bifida aperta, the open neural tube extended to the caudal tip and there was no remnant of the tail at all. The regression of the tail and approximation of the hind limb roots appeared to progress together. At this stage of development, the tail and the limbs of the control embryos were richly vascularised, and the tip of the tail reached as far as the umbilicus. In many embryos of the higher dose groups, the tail buds were not discernible, and the remainder had short withered tail buds with hematomas. By GD 11, in those without the tail buds, the hind limb buds formed the transverse arm of the letter T at the caudal narrow tip of and poor

the embryonic axis. The caudal end of neural tube appeared dilated in many (Figure 6F) and was widely open in a few. Hemorrhagic masses, possibly representing the spina bifida occulta, occupied the dorsal median area of the lumbosacral segment. The limb roots appeared to be separated only by the genital tubercle. As gestation advanced, no tail could be traced in any of these embryos. In the 200 mg/kg RA group, no tail bud was found to have developed on GD 9. The caudal margin of the limb bud was medial rather than caudal in orientation. Consequently, the hind limb roots were adjoining each other as well as the genital tubercle at the caudal tip. The approximation of the limb roots varied in extent and in direction. As a result, varying degrees of malrotation followed. There were several cases of a single median limb, often rotated aberrantly. The proximal segments developed incompletely, and the feet developed oligodactyly, syndactyly, and or clefts. Histologic examination of early embryos (GD 8,6 h post-treatment) revealed cell death in the caudal median area, particularly in the neural tube, hindgut, and paraxial

Pathogenesis of caudal regression 0 R.

PADMANABHAN

147

Fig. 7. Sagittal sections of the tail region of embryos of GD 10, stained with hematoxylin and eosin. A and B, control. Observe the well-defined segmented tail mesoderm (S, somites) and caudal neural tube (T). In the experimental embryos (C-F; 100 mg/kg RA treated on GD 8), large pools of hemorrhage occupy the core of the tail (bar in C and hm in D and F), and somites are small and disrupted. Also note the edema (ed; E) and pronounced cell death in the treated embryos (arrows in F). A and C X6.5; B, X 10; D, X1.5; E and F, X20.

mesoderm. With timja, cell death became more intense and extensive and involved the somites and adjoining lateral mesoderm so that the tail bud could not be identified at subsequent stages. The embryos treated on GD 8 with 100 mgkg of RA, revealed on GD 9, 10, and 11 unequivocal evidence for vascular disruption characterized by initial vascular dilation, patchy hemorrhages, and pronounced cell death in the caudal eminence. It was then followed by extensive disruption of the tail somites and subsequently progressive regression of the entire tail (Figure 7). In some embryos, the caudal neural tube was duplicated. The supernumerary cords were blind ending

and situated between the primary cord and the notochord. The dorsal aortae appeared dilated and there was extensive hemorrhage in the coelomic cavity. The tail gut, connected to the cloaca, was obvious in the control embryos whereas in the RA-treated ones, it was either absent or rudimentary. The anal pit failed to develop, and the cloaca was rudimentary and incompletely divided, thus maintaining direct continuity with the rectum. In many of the treated embryos, the caudal end of the neural tube was dilated. The undivided cloaca was persistent. There was pronounced cell death in the mesenchyme of the caudal axial region, urogenital ridge, and mesone-

148

Reproductive

phros. The ureteric bud either did not develop retarded in development.

Toxicology

or was

DISCUSSION The embryonic median plane is an area of intense developmental activities such as cell ingression, cell migration, cell differentiation, and tissue organization and induction. It is also considered to be a special kind of developmental field, particularly prone to complex malformations, the type of the anomalies reflecting the timing of the insults (11,25-27). These anomalies include atelencephaly, holoprosencephaly, cebocephaly, agenesis of the corpus callosum, cylopia, and oral and facial clefts at the rostra1 end and symmelia and other limb defects, spina bifida, imperforate anus, hypospadias, and rectovaginal, rectovesicle, and rectourethral fistulae at the caudal end. The precursors of these caudal organs arise in the caudal eminence. This elevation is first identified at Stage 9 between the cloaca1 membrane and neurenteric canal (Figure IA in Reference 11). The caudal eminence of the human embryo is different from the end bud of the chick embryos in that its mesoderm becomes segmented (11). Recent studies have shown that this is an area of active cell proliferation, migration, and differentiation, and of morphogenetic tissue interaction (28-30). This region provides mesenchyme for the formation of the notochord, somites (including caudal vertebrae), hindgut, neural tube, hind limbs, and blood vessels. The portion of the body derived from it meets with that derived from the primitive streak and node at the site of closure of the caudal neuropore in the upper sacral region (11,31). The cluster of developmental abnormalities possibly arising from faulty differentiation or regression of this region constitutes CRS. The visceral malformations of CRS are consistent and incompatible with life. Sirenomelia is the least common and most variable of the malformations that constitute CRS (1). A mutant mouse model shows that single gene abnormalities can lead to sirenomelia (32), although no Mendelian pattern of inheritance has been established in humans. Kampmeier (6), Stevenson et al (lo), and Stocker and Heifetz (2) have extensively reviewed the reported cases. The syndrome is now well recognized, but there is considerable confusion about the etiology. Maternal diabetic tendency and drug use have been implicated but not substantiated (9,33-35). Recently, Alles and Sulik (12) have highlighted the association of Potter sequence, VATER or VACTERL, and OEIS complex with CRS. The pathogenesis is poorly understood. There are as many as six theories postulated to explain the pathogenetic mechanisms: 1) lateral compression by amniotic folds (mechanical) (8,lO); 2) defective and or deficient caudal mesoderm (11); 3) defec-

Volume

12, Number 2, 1998

tive tissue interaction (36); 4) vascular steal (nutritional; References 3, 10, 37, 38); 5) overdistention of the neural tube (mechanical; Reference 18); and 6) a combination of vascular disruption, mesodermal injury, and defective microperfusion (38). A close look at the anatomy of the structures affected in CRS and the various theories postulated to explain the pathogenetic mechanisms reveal that regression in CRS is not confined to the caudal end of the embryonic axis (11) and that a host of median and paramedian structures are malformed and therefore in many instances, CRS should be addressed as “axial dysgenesis syndrome.” It is also apparent that no single theory could accommodate all these malformations. Therefore, the whole issue of pathogenesis is more complex than commonly understood. Experimental research might throw some light onto the complex processes of normal and abnormal development of these organs. Different teratogens appear to disrupt different aspects of the development of the caudal eminence and lead to CRS. Wolff (7) produced CRS, including sympodia, in chick embryos by irradiation of an area around the primitive node and discovered that symmelia could only occur if intervention occurred before the formation of the limb buds. Muecke (39) implanted plastic grafts in chick embryos and prevented the migration of primitive streak-derived mesoderm to the infraumbilical abdominal wall and successfully produced cloaca1 exstrophy but failed to induce other malformations seen in OEIS complex. Landauer (40) produced rumplessness (the phenotype equivalent of human CRS) in chick by exogenous insulin. This observation, because of occasional association of maternal diabetes with sacral hypoplasia or sacral agenesis, gave rise to the consideration that CRS in the human may be linked to the presence of insulin or insulin antagonists (41). Surprisingly, this possibility has not been strengthened by clinical observations (33). In a comprehensive review paper, Kalter (35) concluded that the association of maternal diabetes and sirenomelia appears to be virtually nonexistent. Gale’s (42) hamster embryos subjected maternally to hyperglycemia developed hypoplasia of the sacrum, pelvic bone, and femur. Kalter and Warkany (43) confirmed for the first time experimentally the occurrence of vascular steal in renal agenesis and imperforate anus. Stubby tail, sacral vertebral defects, and anorectal malformations have been observed in mouse embryos treated with RA (13,15). Myelocystocele, spina bifida aperta, sympodia, and skeletal anomalies in hamsters and axial skeletal defects associated with vascular lesions in the mouse and chick are reported to be induced by exposure to exogenous RA (44-46). Yasuda et al (14) reported a 5% incidence of “taillessness” in RA-treated ICR mouse embryos. Alles

Patiogenesis

of caudal regression

and Sulik (47) observed a high incidence of spina bifida and pronounced cell death in the neural tube and hindgut endoderm following three successive doses of RA in the mouse. Recent data from Sulik’s lab (12,16,48) show that etretinate augments programmed cell death and induces cell death at ‘ectopic sites when administered on GD 8 and diminution of the caudal cell population when given on GD 9 and thus leads to caudal regression in mouse embryos. However, these investigators did not observe vascular lesions or sympodia. Their ochratoxin A-treated chick embryo model, however, exhibits a 30% incidence of sirenomelia 6 d after treatment (17). Our TO mouse model presents all features of CRS (I) in a very high percentage of fetuses the mothers of which were exposed to a single dose of RA on Day 8 or 9 of gestation. The severity differed according to the dose and the developmental stage at the time of RA administration. This study also has identified the characteristic features of all these anomalies. Most obvious was a dose-dependent intrauterine growth retardation accompanied by regression to agenesis of the tail, meromelia, malrotation of the limbs, and in a few cases, sirenomelia. The fetuses also had spina bifida, agenesis of caudal vertebrae, urogenital malformations, anal agenesis, and rectourethral fistula. The development of tail regression in the GD 8-treatment group is an interesting story. In the IOO-mg group, most embryos developed obvious hematomas within 24 h of treatment. This was accompanied simultaneously by cell death, edema, tissue disruption, and finally total regression of the tail, complete in the following 3 to 4 d of gestation. In the 150 mg/kg group, most o-Fthe embryos failed to develop a tail bud whereas others developed hemorrhagic or avascular tail buds that subslequently degenerated. In the 200 mg/kg group, extensive cell death was observed as early as 6 h post-treatment involving the tissues of the caudal median axis. Later, cell death also affected the sclerotome-myotome junctional area, hindgut, and neural tube but not the notochord. The extreme degree of tissue loss in the median area appears to have led to agenesis of the tail and caudal vertebrae. The dilated vertebral canals of the spina bifida fetuses was also associated with vertebral arch fusion, hemivertebrae, reduction in number of ribs, and/or fusion of lower ribs indicating loss of precursor tissues of this region due to cell death. In the GD 9 RA-treated group, one or more constrictions divided the tail into a stubby proximal segment and an avascular withering distal segment; the later subsequently degenerated. One of the important new findings of this study is the role of vascular disruption in caudal regression. The histologic sequence of events of hemorrhagic degeneration of the tail bud in the 100 mg/kg RA-treated embryos is very much reminiscent of the

l

R. PADMANABHAN

149

vascular disruption sequence described in the tail short variable mutant (49). Vascular disruption/interruption and consequent hypoxia, edema, and nutritional deficiency is now fairly well established as a teratogenic mechanism (550-56). However, it should be stressed that vascular disruption and subsequent hemorrhage are probably secondary pathogenetic mechanisms in the production of developmental deformities and that what triggers the process is not known at present. Hemorrhage, edema, and tissue necrosis that follow uterine vascular clamping (57), amniotic sac puncture (51), and administration of chemicals or drugs (20,52,56-60) are known to lead to fetal malformations. It has now been well established that following vasoconstriction and ischemia, reperfusion occurs, that ischemia-reperfusion results in the production of oxygen-free radicals, and tissue damage follows (61). Such free radicals have been implicated in cocaineinduced neural tube defects in mice; antioxidants prevented cocaine-induced hemorrhage but failed to rescue the embryos from being malformed (60). In the 200 mg/kg group treated on GD 8, every living embryo exhibited total agenesis of the tail bud. The hind limb buds, which in the mouse emanate on GD 9.5, appeared to be situated at the caudal tip close to each other, possibly because of a severe deficiency of caudal mesoderm. Therefore, the unipodia or sympodia observed in this group perhaps results from unilateral failure of the limb bud to form (unipodia) and/or failure of fission of the precursor limb field (sympodia) due to either a primary mesodermal tissue deficiency in the area or inadequate growth of the hindgut and caudal neural tube or both. There appears to be no evidence for progressive fusion of separately developed limb buds. This observation is in agreement with those of Feller and Stemberg (62), Sperber and Marchin (63), and O’Rahilly and Muller (11). An alternate theory of merger is also found in the literature (64,65). In all of our embryos with CRS, the caudal vertebral column was shorter than the cord; spina bifida occulta appeared to be associated with overdistension of the caudal neural tube. Histologically demonstrable intense cell death in the caudal axial mesenchyme and tissue deficiency in the caudal embryonic axis appears to permit the overdistension of the unprotected caudal neural tube and reopening in some cases and affect the development of the adjacent nonneural organs, namely the cloaca and the limb buds (18). Partial to complete absence of the tail gut extension towards the tail root could result from such tissue deficiency in the caudal eminence from which the tail gut is derived (11). Arrested development of the tail bud or disruption of the formed tail bud either as a result of the direct effect of RA or via events involving vascular disruption, edema, and cell death may explain the tissue

150

Reproductive

Toxicology

deficiency facilitating expansion of the caudal neural tube and consequent displacement of visceral and skeletal primordia. Ureteric buds were small, and the metanephric tissue mass was significantly smaller than that in the control. Renal agenesis/hypoplasia observed in this study may be due to such primordial tissue deficiency or absence of an inductive interaction. The cellular mechanisms of retinoids have been the subject of investigation for some time. Perturbations in cell cycle parameters, interference with cell differentiation, cell migration, cell-cell interaction, and stagedependent cell death mediated via RA-induced activation or suppression of specific genes have been reported in susceptible embryonic tissues (66-68). The molecular mechanisms of action of retinoids in teratogenesis are beginning to be clarified in some detail (69-71). The HOX genes that play a vital role in pattern formation appear to be the target for the action of RA (72). Kessel and Gruss (73) and Kessel (74) have recently produced evidence that exogenous RA interferes with the establishment of Hox code and induces homeotic transformations in mice. There is some evidence that RA is involved in controlling the expression of Hox genes in the anteroposterior and proximodistal pattern formation and tissue specification of the limbs in murine embryos (75). In these embryos, the prospective paraxial mesoderm of the anterior body ingresses on GD 7 when Hox genes known to respond positively to RA begin to be expressed; administration of exogenous RA at this stage leads to posterior transformation. On GD 8.5, paraxial mesoderm for the posterior half of the axis ingresses and exogenous RA given at this stage causes anterior homeotic transformations. RARy-‘- embryos exposed to RA at 8.5 d postcoitus have been shown to develop no spina b&la, truncation of the lumbosacral vertebrae, fused ribs, disorganized vertebral centra, or fusion of lumbosacral vertebral arches as do the wild type suggesting that the presence RARy- is essential to produce at least part of the caudal regression syndrome (76). We are currently trying to find out what alterations are induced by exogenous RA in Hox gene expression in the caudal axis of TO mouse embryos and how such changes would correlate with development of caudal regression.

Volume 12, Number 2, 1998

I. 8.

9.

10.

11.

12.

13. 14.

15.

16. 17. 18.

19.

20.

21.

22.

23. Acknowledgments -The author is grateful to Mohamad Kasim, Ponet-y Samad, and Ijaz Ahmed for technical assistance, Dr Abdulbari Bener for statistical advice, and Ashok Prasad for photography. Janet C. Powell carefully read the manuscript and made many useful suggestions. This project was generously supported by the UAE University.

24.

25. 26.

REFERENCES 1. Duhamel B. From mermaid to anal imperforation:

the syndrome of caudal regression. Arch Dis Child. 1961;36:152-5. 2. Stocker JT, Heifetz SA. Sirenomelia: a morphological study of 33

27.

cases and review of the literature. Perspect Pediatr Pathol. 1987; 10:7-50. Kapur RP, Mahony BS, Nyberg DA, Resta RG, Shepard TH. Sirenomelia with a “vanishing twin.” Teratology. 1991;43:103-8. Davies J, Chazen E, Nance WE. Symmelia in one of monozygotic twins. Teratology. 1971;4:367-78. Van Allen MI. Fetal vascular disruptions: mechanisms and some resulting birth defects. Pediatr Ann. 1981;10:219-33. Kampmeier OF. On sireniform monsters, with a consideration of the causation and the predominance of the male sex among them. Anat Rec. 1927;34:365-89. Wolff E. La Science des Monstres, Paris: Galimard; 1948. Orr BY, Long SY, Steffek AI. Craniofacial, caudal and visceral anomalies associated with mutant sirenomelic mice. Teratology. 1982;26:31 l-7. Welch IP, Alterman K. The syndrome of caudal dysplasia: a review including etiologic considerations and evidence of heterogeneity. Pediatr Pathol. 1984;2:313-27. Stevenson RE, Jones KL, Phelan MC, Jones MC, Barr M, Jr, Clericuzio C, Harley RA, Benirschke K. Vascular Steal: the pathogenetic mechanism producing sirenomelia and associated defects of the viscera and soft tissues, Pediatrics. 1986;78:451-7. O’Rahilly R, Muller F. Interpretation of some median anomalies as illustrated by cyclopia and symmelia. Teratology. 1989;40:40921. Alles AJ, Sulik KK. A review of caudal dysgenesis and its pathogenesis as illustrated in an animal model. Birth Defects Orig Art Ser. 1993;29:83-102. Kochhar DM. Teratogenic activity of retinoic acid. Acta Path Microbial Stand. 1967;70:398-401. Yasuda Y, Konishi H, Kihara T, Tanimura T. Discontinuity of primary and secondary neural tube in spina bifida induced by retinoic acid in mice. Teratology. 1990;41:257-74. Hashimoto R, Oda S, Inouye M, Yamamura H. The pathogenesis of anorectal malformation induced by all-trans retinoic acid in mice. Cong Anom. 1993;32:133-42. Sessions RP, Lloyd RA, Sulik KK, Mesrobian HGJ. Imperforate anus induced by etretinate in mice. Teratology. 1993;47:435A, Wei X, Sulik KK. Pathogenesis of caudal dysgenesis induced by ochratoxin A in chick embryos. Teratology. 1996;53:378-91. Gardner WJ. Hypothesis: overdistension of the neural tube may cause anomalies of nonneural organs. Teratology. 1980;22:22938. Padmanabhan R. Experimental induction of cranioschisis aperta and exencephaly after neural tube closure: a rat model. J Neural Sci. 1984;66:235-43. Padmanabhan R. Light microscopic studies on the pathogenesis of exencephaly induced in the rat after neural tube closure. Teratology. 1988;37:29-36. Padmanabhan R. Is exencephaly the forerunner of anencephaly? An experimental study on the effect of prolongation of gestation on the exencephaly induced after neural tube closure in the rat. Acta Anat. 1991;141:182-92. Sterz H, Lehmann H. A critical comparison of the free hand razor blade dissection method according to Wilson with an in situ sectioning method for rat fetuses. Teratogen Carcinogen Mutagen. 1985;5:347-54. Inouye M. Differential staining of cartilage and bone in fetal mouse skeleton by alcian blue and alizarin red-S. Cong Anom. 1976;16:171-3. Elzey FF. Introductory statistics: a microcomputer approach. Adapted for IBM PC by Cloward RD. Monterry, CA: Brooks/Cole Publishing Co.; 1987. Opitz JM, Gilbert EF. Pathogenetic analysis of congenital anomalies in humans. Path01 Ann. 1982;12:301-49. Lubinski M, Moeschler J. Different clusters within the VATER association distinguished through cardiac defects: possible effects of teratogenic timing. Dysmorphol Clin Genet. 1987; 1:80-2. Torfs CP, Curry CJR, Bateson TF. Population-based study of tracheoesophageal fistula and esophageal atresia. Teratology. 1994;52:220-30.

Pathogenesis

of caudal regression

V, Dias MS. Mesoderm move28. Schoenwolf GC, Garcia-Martinez ment and fate during avian gastrulation and neurulation. Dev Dyn. 1992;193:235-48. 29. Selleck MAJ, Stem CD. Fate mapping and cell lineage analysis of Hensen’s node in the chick embryo. Development. 1991;112:61526. 30. Smith JL, Gestland KM, Schoenwolf GC. Prospective fate map of the mouse primitive streak at 7.5 days of gestation. Dev Dyn. 1994;201:279-89. 31. Muller F and O’Rahilly R. Development of the human brain, the closure of the caudal neuropore, and the beginning of the secondary neurulation at stage 12. Anat Embryo1 (Berl). 1987;176:41330. 32. Hoombcek EK. A gene producing symmelia in the mouse. Teratology. 1970;3:1-10. 33. Mills JL. Malformations in infants of diabetic mothers. Teratology. 1982;25:385-94. 34. Young ID, O’Reilly KM, Kendall CH. Etiological heterogeneity in sirenomelia. Pediatr Pathol. 1986;5:31-43. 35. Kalter H. Case reportlr of malformations associated with maternal diabetes: history and ‘critique. Clin. Genet. 1993;43: 174-9. R, Brunet C. Origin of abnormality in a human 36. Chandebois symmelian foetus as elucidated by our knowledge of vertebrate development. Teratology. 1987;36: 1 l-22. case of the 37. Murphy JJ, Fraser IGC, Blair GK. Sirenomelia: surviving mermaid. J Pediatr Surg. 1992;27:1265-8. 38. McCoy MC, Chescheir NC, Kuller JA, Altman GA, Flannagan LM. A fetus with sirenomelia, omphalocele, and meningomyleocele, but with normal kidneys. Teratology. 1994;50: 168-71. 39. Muecke EC. The role of cloacal membrane in extrophy: the first successful experimental study. J Ural. 1964;92:659-67. 40. Landauer W. Rumplessness of chicken embryos produced by injection of insulin and other chemicals. J Exp Zool. 1945;98:6577. 41. Wilson JSP, Valiance-Owen J. Congenital deformities and insulin antagonism. Lancet. 1966;2:940-2. 42. Gale TF. Effects of in vivo exposure of pregnant hamster to glucose. 1. Abnormalities in LVG strain fetuses following intermittent multiple treatments with two isomers. Teratology. 1991; 44:193-202. 43. Kalter H, Warkany J. Experimental production of congenital malformations in strams of inbred mice by maternal treatment with hypervitaminosos A. Am J Pathol. 1961;38:1-20. 44. Shenefelt RD. Morphogenesis of malformations in hamsters caused by retinoic acid: relation to dose and stage of treatment. Teratology. 1972;5:1133-18. 45. Tibbles L, Wiley MJ. A comparative study of the effects of retinoic acid given during the critical period for induction of spina bifida in mice and hamsters. Teratology. 1988;37:113-25. 46. Griffith CM, Wiley MJ. Direct effects of retinoic acid on the development of the tail in chick embryos. Teratology. 1989;39: 261-75. 47. Alles AJ, Sulik KK. Retinoic acid induced spina bifida: evidence for a pathogenetic mechanism. Development. 1990;408:73- 8 1. 48. Mesrobian HGJ, Sessions RP. Lloyd RA, Sulik KK. Cloaca1 and urogenital abnormalities induced by etretinate in mice. J Urol. 1994;152:675-8. 49. Seller MJ, Wallace ME. Tail short variable: characterization of a new mouse mutant and its possible analogy to certain human vascular disruption defects. Teratology. 1993;48:383-91. 50. Grabowski CT. Eml>ryonic oxygen deficiency: a physiological approach to teratological mechanisms. In Woollam DHM, ed. Advances in Teratology. New York: Academic Press; 1970:12566. de51. Kennedy LA, Persuad TVN. Pathogenesis of developmental fects induced in the rat by amniotic sac puncture. Acta Anat. 1977;97:23-35. 52. Rajala GM, Kaplan S. The formation of caudal haematoma in trypan blue-treated chick embryos as a function of morphological stage at treatment. Teratology. 1980;21:265-9. 53. Russel LJ, Weaver DD, Bull MJ, Weinbaum M. In utero brain

54.

55.

56.

57.

58.

59.

60.

61.

62.

63. 64.

65. 66.

67.

68.

69.

70.

71. 72. 73.

74. 75.

76.

0 R. PADMANABHAN

151

destruction resulting in collapse of the fetal skull, microcephaly, scalp rugae and neurologic impairment: The fetal brain disruption sequence. Am J Med Genet. 1984;17:509-21. Hoyme HE, Higginbottom MC, Jones KL. The vascular pathogenesis of gastroschisis: intrauterine interruption of the omphalomesenteric artery. J Pediatr. 1981;98:228-31. Bouwes JNB, Weaver DD. Subclavian artery supply disruption sequence: hypothesis of a vascular etiology for Poland, KlippelFeil, and Mobius. Am J Med Genet. 1986;23:903-18. Padmanabhan R, Pallot DJ. Aspirin-alcohol interaction in the production of cleft palate and limb malformations in the TO mouse. Teratology. 1995;51:404-17. Liest KH, Grauwiler J. Fetal pathology in rats following uterine vessel clamping on day 14 of gestation. Teratology. 1974;10:5568. Webster WS, Brown Woodman PDC. Cocaine as a cause of congenital malformations of vascular origin: experimental evidence in the rat. Teratology. 1990;41:689-97. Danielson MK, Danielsson BRG, Marchner H, Lundin M, Rundqvist E, Reiland S. Histopathological and hemodynamic studies supporting hypoxia and vascular disruption as explanation to phenytoin teratogenicity. Teratology. 1992;46:485-97. Zimmerman EF, Raju PB, Resnick E, Fisher KE. Role of oxygen free radicals in cocaine-induced vascular disruption in mice. Teratology. 1994;49:192-201. Halliwell B, Gutteridge JMC. Role of free radicals and catalytic metal ions in human disease: An Overview. In: Packer L, Glazer GN, eds. Methods in enzymology, vol. 186. San Diego, CA: Academic Press; 1990: 1- 85. Feller A, Stemberg H. Zur kenntnis der fehlbildungen der wirbelsaule. Ill. Mitteilung uber den vollandigen mange1 unteran. Virchows Arch (A). 1931;280:649-92. Sperber GH, Machin GA. Microscopic study of midline determinants in Janiceps twins. Birth Defects. 1987;23:243-75. Barr M Jr. Comments on “Origin of abnormality in a human symmelian foetus as elucidated by our knowledge of vertebrate development.” Teratology. 1988;38:487-8. Hoyme HE. The pathogenesis of sirenomelia: an editorial comment. Teratology. 1988;38:485-6. Kochhar DM, Satre MA. Retinoids and Fetal Malformations. In: Sharma RP, ed. Dietary factors and birth defects. California: Pacific Division, A A A S; 1993;134-227. Agnish ND, Kochhar DM. Developmental toxicology of retinoids. In: Koren G, ed. Retinoids in clinical practice. New York: Marcel Dekker Inc.; 1993:47-76. Jiang H, Kochhar DM. Induction of tissue transglutaminase and apoptosis by retinoic acid in the limb bud. Teratology. 1992;46: 333-40. Maden M. Distribution of cellular retinoic acid-binding proteins I and II in the chick embryo and their relationship to teratogenesis. Teratology. 1994;50:294-301. Jiang H, Gyda II M, Hamish DC, Chandraratna R, Soprano KJ, Kochhar DM, Soprano DR. Teratogenesis by retinoic acid analogs positively correlates with elevation of retinoic acid receptor-P2 mRNA levels in treated embryos. Teratology. 1994;50:38-43. Kochhar DM. Retinoids and retinoid receptors in teratogenesis. Cong Anom. 1995;35:55-71. Tabin C. The initiation of the limb bud: Growth factors, Hex genes, and retinoids. Cell. 1995;80:671-4. Kessel M, Gruss P. Homeotic transformations of murine vertebrae and concomitant alterations of Hox codes induced by retinoic acid. Cell. 1991;67:89-104. Kessel M. Respecification of vertebral identities by retinoic acid. Development. 1992;115:4487-501. Ruberte E, Friederich V, Morris-Kay GM, Chambon P. Differential distribution pattern of CRABP-1 and CRABP-II transcripts during mouse embryogenesis. Development. 1992;115:973-87. Lohnes D, Kastner P, Dierich A, Mark M, LeMeur M, Chambon P. Function of retinoic acid receptor y in the mouse. Cell. 1993;73: 643-58.