In higher vertebrates, the formation of the body axis proceeds in a craniocaudal direction during gastrulation. Cell biological evidence suggests that mesoderm formation and specification of axial positions occur simultaneously. Exposure of gastrulating embryos to retinoic acid induces changes in axial patterns, e.g. anterior and posterior homeotic transformations of vertebrae. These morphological changes are accompanied by changes in the nonidentical, overlapping expression domains of Hox genes.

In this report the influence of retinoic acid, administered at the end of and after gastrulation, on vertebral patterns is described. Anterior transformations and truncations affecting the caudal part of the vertebral column characterize animals exposed on day 8 and 9. 4 hours after retinoic acid administration on day 8 + 5 hours, Hox-1.8, Hox-1.9, and Hox-4.5 transcripts were not detected in their usual posterior expression domains, whereas transcripts of the anterior Hox-1.5 gene remained unaffected. 4 days after RA exposure on day 8 + 5 hours, Hox-1.8 expression was shifted posteriorly by an effectively low dose of RA, which induced the formation of supernumerary ribs. Hox-1.8 expression was limited to posterior, disorganized mesenchyme, bulging out neural tube, some intestinal loops and the hindlimb in truncated embryos exposed to a high dose of RA. A causal relation between the delayed activation of posterior Hox genes and anterior transformations or agenesis of vertebrae is discussed.

On day 10.5 posterior transformations begin to occur in the cervical region, while later exposures again affect more caudal structures. The distribution of the transformations along the vertebral column indicates an influence of RA on migrating sclerotome cells before they are finally fixed in the cartilagenous vertebrae. The findings show that the mesodermal segments originally specified during gastrulation can be respecified in their second migratory phase, with effects spreading for a second time in a craniocaudal direction. The transformations are discussed with regard to a molecular specification of axial levels by Hox codes, defined as combinations of expressed Hox genes.

Formation of the anteroposterior body axis is tightly coupled to the formation of mesoderm during gastrulation. In higher vertebrates such as birds and mammals, the primitive streak, extending from the posterior embryonic margin is the first, still transient, structure to interrupt the radial symmetry of the early embryo (Bellairs, 1986). When the streak is fully extended, medial mesoderm (notochord) migrates through its tip (Hensen’s node, primitive node, archenteron or node), paraxial mesoderm ingresses through the anterior streak and intermediate and lateral mesoderm through successively more posterior regions of the anterior half of the streak (Selleck and Stern, 1991).

The specification of axial positions is a prerequisite for the formation of the body axis. Numerous examples from the animal kingdom indicate that segmentation has been used independently in different phyla to organize axis formation (Hogan et al., 1985; Akam et al., 1988; Keynes and Lumsden, 1990; Stern, 1990). Segmentation of the axis in tetrapod vertebrates is observed in the neuroectoderm of the head and the paraxial mesoderm of the trunk. Thus, the hindbrain is segmented into rhombomeres, while the neural tube is not segmented in higher vertebrates (Lumsden, 1990). In adults, head segmentation remains apparent in the order of the cranial nerves. Segmentation of the paraxial mesoderm is only transient in the head region; the occipital somites develop into the unsegmented basioccipital bone. In contrast mesodermal segmentation in the trunk region, which is first evident in the somites and sclerotomes, remains obvious in the vertebral column.

Transplantation experiments indicate that neural crest cells from a neuroectodermal segment, the first branchial arch, maintain their specification autonomously (Noden, 1983). Segments from the paraxial mesoderm also keep their axial specification, or their identity, upon transplantation to a different axial position. For example, mesoderm from the prospective thoracic level will develop into rib-bearing vertebrae also after transfer into the cervical region (Kieny et al., 1972).

Normal morphogenesis of vertebrate embryos is severely affected by externally administered doses of retinoic acid (RA). Typically it leads to craniofacial abnormalities and heart defects by influencing normal development of neural crest cells, and to limb deformities (Shenefelt, 1972; Kochhar, 1973; Lammer et al., 1985). Recently it was shown that RA interferes with the axial specification of ectoderm and mesoderm during gastrulation (Durston et al., 1989; Kessel and Gruss, 1991; Ruiz i Altaba and Jessel, 1991; Sive et al., 1990). Analysis of vertebral patterns in mice showed that exposure during the specification of the anterior part of the body axis (occipital to mid-thoracic region) led to posterior homeotic transformations of vertebrae along the complete length of the vertebral column (Kessel and Gruss, 1991). Transformations included the formation of a proatlas at the craniocervical transition, formation of ribs on C7, and rostral shifts of the os sacrum. Exposure to RA during the formation of the posterior body axis (caudal of the mid-thorax), however, led to anterior homeotic transformations. These included generation of supernumerary ribs and caudal shifts of the os sacrum.

The paraxial mesoderm, most significantly the somites and their derivatives, the sclerotomes, are characterized by nonidentical, overlapping expression domains of Hox genes (for review see Kessel and Gruss,1990). It was suggested that combinations of expressed Hox genes, “Hox codes”, specify the identity of vertebral segments (Kessel and Gruss, 1991). Anterior and posterior transformations induced by RA could be explained on the basis of the known responsiveness of Hox genes to RA (Simeone et al., 1990; Simeone,1991), leading to posteriorized or anteriorized Hox codes. It could be demonstrated that exposure to RA during gastrulation indeed induced more anterior expression domains of the Hox genes Hox-1.5, Hox-1.1, and Hox-3.1 (Kessel and Gruss, 1991).

In this report, I describe the consequences on vertebral column development of RA exposure at the end of gastrulation and following stages. The phenotypic analysis reveals that vertebral identities, initially specified upon mesoderm ingression, can be respecified during somite differentiation. The observed vertebral transformations are discussed with regard to axis definition by specific Hox codes in the paraxial mesoderm.

All experimental details are described in a preceding publication (Kessel and Gruss, 1991) and only briefly summarized here.

Mice

Female outbred NMRI mice and male C57B16/DBA-F1 mice were mated overnight, and 12 a.m. of the following day (plug day) was taken as day 0.5. Alternatively, for timed matings animals were mated for 2 hours, and fertilization was assumed after 1 hour. Stages of animals from timed matings are specified in days plus hours, e.g. day 8 + 5 hours. 100 mg per kilogram body weight all-trans RA (Sigma) or 450 mg/kg 13-cis RA (Sigma) in sesame oil was administered by a single oral gavage at the indicated time.

Skeletal preparations

Specimens were fixed, stained with alcian blue and alizarin red, and cleared in potassium hydroxide as described before.

In situ RNA analysis

The Hox-1.5 (Fainsod, 1987) and the S8 probe (Kongsuwan et al., 1988) were as described. Hox-1.8 and Hox-1.9 probes were provided prior to publication by H. Haack (Göttingen). Paraffin sections were analyzed by standard procedures.

Skeletal phenotypes

The experimental approach used for the present study has been outlined in a previous publication (Kessel and Gruss, 1991). RA was administered to pregnant mice at different stages of gestation by oral gavage. RA has a short half life in vivo (Creech-Kraft et al., 1987), therefore the embryos are exposed to RA for relatively short times, depending on the amount of RA given. Fetuses removed shortly before birth or newborns were used for whole-mount skeletal preparations by alizarin red-alcian blue staining with subsequent clearing in potassium hydroxide solution. Anatomical changes observed in RA-exposed animals are described below, with emphasis on the changes of vertebral patterns. Cervical, thoracic, lumbar, and sacral vertebrae are designated C, T, L, and S, respectively, followed by their number. For comparative purposes it was occasionally useful to number the vertebrae consecutively beginning with the atlas (no. 1), without referring to their regional identity. In anterior transformations a vertebra assumes morphological characteristics of a more anterior vertebra. Specific anterior transformations observed are the generation of eight or nine vertebrosternal ribs (transformation h), the formation of ribs on vertebra no. 21 or no. 22 (transformation i), the transformation of the first sacral to a lumbar identity (transformation k), and the transformation of the first caudal to a sacral identity (transformation 1). The posterior transformations observed are the propagation of tuberculi anterior on C5 (transformation b), cervical ribs or extensive ossifications on C7 (transformation c), transformation of the last lumbar vertebra (no. 26) to a sacral vertebra (transformation f) and transformation of the last sacral vertebra (no. 30) to a caudal identity (transformation g). They are described below for the different experimental groups in the craniocaudal order of their occurrence and are summarized in Table 1.

Table 1:

Vertebral patterns induced by exposure to retinoic acid

Vertebral patterns induced by exposure to retinoic acid
Vertebral patterns induced by exposure to retinoic acid

Control mice

The vertebral columns from control mice not exposed to RA show little variation (Table 1). Variables not scored as transformations were small ossifications on C7 and the fusion of either three or four vertebrae in the os sacrum. A thorough description is given by Kessel and Gruss (1991).

RA exposure on day 8 + 5 hours

Exposure of mice to 450 mg/kg 13-cis RA was described before (Kessel and Gruss, 1991). The effect of this treatment resembled the application of 10 mg/kg of the all-trans isomer of RA. Increasing the RA dose to 100 mg/kg all-trans RA led to numerous resorptions of embryos and small litters. The surviving animals displayed multiple severe skeletal defects. Most obvious were cranial malformations and truncations affecting the complete caudal body half (Fig. 1). A typical skull abnormality was the fusion of the basisphenoid and basioccipital bone (Fig. 2). The cervical region of the vertebral column was always well developed, for example, this is apparent in the propagation of the tuberculi anterior (Figs 1B,C, 3A, 4A). In 63% of the animals eight or nine ribs were contacting the sternum (transformation h, Table 1), instead of the normal seven. Vertebral centers were split posterior to T6 or T7, posterior thoracic vertebrae were deteriorated, and neural arches of more posterior vertebrae were not closed. Posterior ribs were fused, degenerate, and sometimes not unequivocally countable. In most cases 14 or 15 ends of ribs were scored (transformation i; Figs 1, 3, 4). Lumbar, sacral or caudal vertebrae were not formed in any animal. This phenotype was always associated with spina bifida (Tibbles and Wiley, 1988; Alles and Sulik, 1990). The tail was formed only as a cutaneous appendage with no vertebral content. Hindlimbs and pelvis were either completely absent (Fig. 1C) or properly formed, but without any contact to a vertebral structure (Fig. 1B). Generation of hindlimbs requires the presence of paraxial mesoderm (i.e. lumbosacral somites) and somatopleura. Therefore, the phenotype shown in Fig. 1B, in contrast to 1C, implicates the initial formation of lumbosacral precursor structures. In these animals, however, the differentiation pathway from somites to sclerotome, and finally to vertebrae was apparently not continued, while the development of the muscle lineage via the dermomyotome could proceed. This finding is of particular interest, considering that the Hox genes are typically expressed in the sclerotome, but not in the myotome.

Fig. 1.

Complete skeletons of mice exposed to RA on day 8 + 5 hours. Whole-mount cleared preparations in which the rib cage was opened and the sternum with portions from the anterior ribs was removed in order to allow inspection of the ventral side of the vertebral column. (A) Wild-type newborn. Note the normal vertebral pattern (C7/T13/L6) and propagation of limbs and pelvis. (B) Day-19 fetus exposed to RA on day 8 + 5 hours. Note the proper generation of vertebrae C1 to T6, and the complete absence of lumbar, sacral and caudal vertebrae. All vertebral centers beginning from T7 are split, the posterior ribs are fused and fifteen ends of ribs can be counted. Hindlimbs of near normal structure and size are formed. (C) Day-19 fetus exposed to RA on day 8 + 5 hours. The skeleton resembles in most regards the description under (B). Less ribs can be counted due to the chaotic state at the end of the vertebral column. Remarkable in this specimen is the complete lack of hindlimbs.

Fig. 1.

Complete skeletons of mice exposed to RA on day 8 + 5 hours. Whole-mount cleared preparations in which the rib cage was opened and the sternum with portions from the anterior ribs was removed in order to allow inspection of the ventral side of the vertebral column. (A) Wild-type newborn. Note the normal vertebral pattern (C7/T13/L6) and propagation of limbs and pelvis. (B) Day-19 fetus exposed to RA on day 8 + 5 hours. Note the proper generation of vertebrae C1 to T6, and the complete absence of lumbar, sacral and caudal vertebrae. All vertebral centers beginning from T7 are split, the posterior ribs are fused and fifteen ends of ribs can be counted. Hindlimbs of near normal structure and size are formed. (C) Day-19 fetus exposed to RA on day 8 + 5 hours. The skeleton resembles in most regards the description under (B). Less ribs can be counted due to the chaotic state at the end of the vertebral column. Remarkable in this specimen is the complete lack of hindlimbs.

Fig. 2.

Fusions involving the occipital bone. (A) Typical cranial abnormalities after exposure to RA on day 8 + 5 hours. The long arrow points to the fusion between basisphenoid (bs) and basioccipital bone (bo). The short arrow points to the lack of a tympanic ring and the arrowhead to the lack of a palatal bone. (B) Typical cranial abnormalities after exposure to RA on day 10.5. The long arrow points to the fusion between exooccipital (eo) and basioccipital bone (bo). Tympanic rings (tr) are present, the arrowhead points to the lack of a palatal bone. (C) Control wild-type mouse at birth. Note the cartilagenous, non-ossified connections between the basioccipital and basisphenoid or exooccipital bone and a normally formed palatal bone (pb).

Fig. 2.

Fusions involving the occipital bone. (A) Typical cranial abnormalities after exposure to RA on day 8 + 5 hours. The long arrow points to the fusion between basisphenoid (bs) and basioccipital bone (bo). The short arrow points to the lack of a tympanic ring and the arrowhead to the lack of a palatal bone. (B) Typical cranial abnormalities after exposure to RA on day 10.5. The long arrow points to the fusion between exooccipital (eo) and basioccipital bone (bo). Tympanic rings (tr) are present, the arrowhead points to the lack of a palatal bone. (C) Control wild-type mouse at birth. Note the cartilagenous, non-ossified connections between the basioccipital and basisphenoid or exooccipital bone and a normally formed palatal bone (pb).

Fig. 3.

Vertebral patterns induced by RA exposure. Embryos had been exposed to all-trans RA (100 mg/kg) at the indicated times. Vertebral columns were dissected from clearing preparations beginning with the atlas (C1) as the most rostral vertebra. Clavicula, forelimbs, hindlimbs and pelvic girdle were removed, the rib cage was opened and ribs were trimmed close to the vertebrae. Columns were photographed from the ventral side. Frequencies of vertebral patterns, as well as further anatomical descriptions are given in Table 1. The cervical vertebra with the tuberculum anterior, either C5 or C6, is indicated on the right of the column, also the numbers of the last thoracic vertebrae are given. Lumbar vertebrae are numbered on the left of the column. For a schematic representation and further explanations see Fig. 4 and text. (A) Exposure on day 8 + 5 hours. Note fusion of ribs, anlagen for 15 ribs (arrow), nonformation of lumbar, sacral and caudal vertebrae (arrow). (B) Exposure on day 9.5. Note seven lumbar vertebrae, nonformation of caudal vertebrae (arrow). (C) Exposure on day 9.5. Note fusion of five vertebrae in the sacral bone (arrowhead), nonformation of posterior caudal vertebrae (arrow). (D) Wild-type vertebral pattern. The same phenotype was observed in animals exposed on day 9.5 to 450 mg/kg 13-cis RA. (E) Exposure on day 10.5. Note fused tuberculi anterior on C5 and C6 (arrow). (F) Exposure on day 10.5. Note tuberculi anterior on C5 (arrow), and transformation of vertebra no. 21 to a thoracic vertebra. (G) Exposure on day 11. Note transformation of vertebra no.21 to a thoracic vertebra (T14). (H) Exposure on day 11.5. Note transformation of vertebrae no. 21 (T14) and no.22 (T15) to thoracic vertebrae. The upper arrow points to the minute rib anlage on vertebra T15/no. 22 (see also Fig. 8). Only 4 lumbar vertebrae are generated (lower arrow). (I) Exposure on day 11.5. Note only four lumbar vertebrae (arrow), causing a rostral shift of the os sacrum.

Fig. 3.

Vertebral patterns induced by RA exposure. Embryos had been exposed to all-trans RA (100 mg/kg) at the indicated times. Vertebral columns were dissected from clearing preparations beginning with the atlas (C1) as the most rostral vertebra. Clavicula, forelimbs, hindlimbs and pelvic girdle were removed, the rib cage was opened and ribs were trimmed close to the vertebrae. Columns were photographed from the ventral side. Frequencies of vertebral patterns, as well as further anatomical descriptions are given in Table 1. The cervical vertebra with the tuberculum anterior, either C5 or C6, is indicated on the right of the column, also the numbers of the last thoracic vertebrae are given. Lumbar vertebrae are numbered on the left of the column. For a schematic representation and further explanations see Fig. 4 and text. (A) Exposure on day 8 + 5 hours. Note fusion of ribs, anlagen for 15 ribs (arrow), nonformation of lumbar, sacral and caudal vertebrae (arrow). (B) Exposure on day 9.5. Note seven lumbar vertebrae, nonformation of caudal vertebrae (arrow). (C) Exposure on day 9.5. Note fusion of five vertebrae in the sacral bone (arrowhead), nonformation of posterior caudal vertebrae (arrow). (D) Wild-type vertebral pattern. The same phenotype was observed in animals exposed on day 9.5 to 450 mg/kg 13-cis RA. (E) Exposure on day 10.5. Note fused tuberculi anterior on C5 and C6 (arrow). (F) Exposure on day 10.5. Note tuberculi anterior on C5 (arrow), and transformation of vertebra no. 21 to a thoracic vertebra. (G) Exposure on day 11. Note transformation of vertebra no.21 to a thoracic vertebra (T14). (H) Exposure on day 11.5. Note transformation of vertebrae no. 21 (T14) and no.22 (T15) to thoracic vertebrae. The upper arrow points to the minute rib anlage on vertebra T15/no. 22 (see also Fig. 8). Only 4 lumbar vertebrae are generated (lower arrow). (I) Exposure on day 11.5. Note only four lumbar vertebrae (arrow), causing a rostral shift of the os sacrum.

Fig. 4.

Schematic explanation of the vertebral patterns A-I in Fig. 3. Open box: cervical vertebrae, tuberculi anterior are indicated as small black squares. Black box: thoracic vertebrae. Diagonally hatched box: lumbar vertebrae. Stippled rectangle: sacral bone. Darkly stippled boxes: caudal vertebrae. Consecutive numbering of segments is given on the right margin of the figure, vertebrae are counted alongside the columns. Homeotic transformations b, c, h, i, k, and 1 are indicated by arrows, their directions indicate anterior or posterior transformations. For further explanations see legend to Fig. 3 and text.

Fig. 4.

Schematic explanation of the vertebral patterns A-I in Fig. 3. Open box: cervical vertebrae, tuberculi anterior are indicated as small black squares. Black box: thoracic vertebrae. Diagonally hatched box: lumbar vertebrae. Stippled rectangle: sacral bone. Darkly stippled boxes: caudal vertebrae. Consecutive numbering of segments is given on the right margin of the figure, vertebrae are counted alongside the columns. Homeotic transformations b, c, h, i, k, and 1 are indicated by arrows, their directions indicate anterior or posterior transformations. For further explanations see legend to Fig. 3 and text.

In summary, exclusively anterior transformations were induced in the mid-thoracic and more caudal regions. Transformation h (eight vertebrosternal ribs) and i (supernumerary ribs) were induced by both the low (Kessel and Gruss, 1991) and the high RA. dose on day 8.5. However, instead of the transformations k and 1 (caudal shift of the os sacrum), the higher dose induced complete disruptions of vertebral structures.

RA exposure on day 9.5

Defects in animals exposed to 100 mg/kg all-trans RA on day 9.5 were restricted to the posterior lumbar and more caudal regions. No defects of cranial structures were observed. Animals were tailless or had a stump tail phenotype with zero to five caudal vertebrae (Figs 3B,C, 4B,C). Unique features, never observed in any other experimental group, were the formation of seven lumbar vertebrae (11%) and the fusion of five vertebrae to form the os sacrum (28%; Table 1). The eight different vertebral patterns were caused by combinations of anterior transformations at the thoracolumbar, the lumbosacral and the sacrocaudal transition (transformations i, k, and 1, Table 1).

Skeletal structures of animals treated on day 9.5 with 450 mg/kg 13-cis RA, .which represents effectively a lower RA dose and thus a shorter exposure time, resembled the wild-type specimens with normal vertebral patterns (C7/T13/L6/S4).

RA exposure on day 10.5

Development and ossification of the skull appeared normal, except for a typical fusion occurring between the basi- and the exooccipital bones in 48% of the animals (Fig. 2, Table 1). Several changes occurred in the anterior half of the vertebral column (Figs 3E,F, 4E,F). 44% of the animals had one or two tuberculi anterior on cervical vertebra C5, while normally they are generated only on C6 (transformation b, Figs 5, 6). In these cases either additional tuberculi were found on C6, or C6 completely lacked a foramen transversarium, a feature normally unique to C7. The morphological changes of vertebrae C5 and C6 had consequences for other structures in the cervical region. Histological sections revealed the arteria vertebralis entering the vertebral column at C5, instead of the normal C6 (Fig. 7). Extensive ossifications on the processus transversi of C7 led to the formation of rib heads or cervical ribs in 62% of the animals (transformation c; Fig. 6). However, the cervical ribs fused to the first thoracic rib in each case and never made an independent contact to the sternum. Complete transformations with a sternal rib on C7 were exclusively observed after RA exposure on day 7.3 (Kessel and Gruss, 1991). The first of the normal six lumbar vertebrae was converted to a thoracic vertebra, as defined by a fourteenth rib in 21% of the animals (transformation i; Figs 3E,F, 4E,F). Interestingly, the posterior transformations b and c and the anterior transformation i were also found in the same animals.

Fig. 5.

Cervical vertebral columns with anterior transformations. Vertebrae no. 2 (axis with dens axis) to no. 7 are numbered, the pattern of the tuberculi anterior (t.a.) is indicated by open circles for the lack of tuberculi, and closed circles for their presence. Short arrows point to the head of a rib, long arrows a cervical rib on no. 7. (A) Wild-type phenotype. Note two t.a. on no. 6. B, C and D show phenotypes after RA exposure on day 10.5. (B) Note unilateral transformation with a large t.a. on no. 5, and a small t.a. on no. 6. (C) Note two t.a. on no. 5 and one on no. 6. (D) Note the complete transformation of no. 5 with two t.a., and not t.a. or foramina transversaria on no. 6. See Fig. 6 for dissection of this column.

Fig. 5.

Cervical vertebral columns with anterior transformations. Vertebrae no. 2 (axis with dens axis) to no. 7 are numbered, the pattern of the tuberculi anterior (t.a.) is indicated by open circles for the lack of tuberculi, and closed circles for their presence. Short arrows point to the head of a rib, long arrows a cervical rib on no. 7. (A) Wild-type phenotype. Note two t.a. on no. 6. B, C and D show phenotypes after RA exposure on day 10.5. (B) Note unilateral transformation with a large t.a. on no. 5, and a small t.a. on no. 6. (C) Note two t.a. on no. 5 and one on no. 6. (D) Note the complete transformation of no. 5 with two t.a., and not t.a. or foramina transversaria on no. 6. See Fig. 6 for dissection of this column.

Fig. 6.

Transformation of three cervical vertebrae to posterior identities. Vertebrae were prepared from an animal exposed to RA on day 10.5 (A, see Fig. 5D) and a control animal (B). Atlas (At, no. 1, C1), axis (Ax, no. 2, C2), C3, and C4 are unaffected by RA. Tuberculi anterior are formed on C5 instead of C6. The typical C7 morphology with no foramina transversaria is found for C6. A cervical rib fused to the first thoracic rib and an extensive rib anlage (caput costae) is generated on C7. Arrows indicate the three transformations of vertebral identities, which are defined further in the text as transformations b and c.

Fig. 6.

Transformation of three cervical vertebrae to posterior identities. Vertebrae were prepared from an animal exposed to RA on day 10.5 (A, see Fig. 5D) and a control animal (B). Atlas (At, no. 1, C1), axis (Ax, no. 2, C2), C3, and C4 are unaffected by RA. Tuberculi anterior are formed on C5 instead of C6. The typical C7 morphology with no foramina transversaria is found for C6. A cervical rib fused to the first thoracic rib and an extensive rib anlage (caput costae) is generated on C7. Arrows indicate the three transformations of vertebral identities, which are defined further in the text as transformations b and c.

Fig. 7.

Entering of arteria vertebralis. Embryos were dissected out on day 12.5 and routine histology was performed using trichrome staining. Indicated are the anlage for the basioccipital bone (bo), thoracic ribs (tr) beginning from prevertebra no. 8, a cervical rib (cr) anlage on no. 7, spinal ganglia (sg), and the neural tube (nt). The prevertebrae are numbered. (A) Wild-type embryo. Note the entering of the arteria vertebralis into the foramen transversarium of vertebra no. 6 (arrow). (B) Embryo exposed to RA on day 10.5. Note entering of the arteria vertebralis into vertebra no. 5 (arrow), presumably also indicating a shift of the tuberculum anterior to no. 5. A cervical rib is associated with no. 7.

Fig. 7.

Entering of arteria vertebralis. Embryos were dissected out on day 12.5 and routine histology was performed using trichrome staining. Indicated are the anlage for the basioccipital bone (bo), thoracic ribs (tr) beginning from prevertebra no. 8, a cervical rib (cr) anlage on no. 7, spinal ganglia (sg), and the neural tube (nt). The prevertebrae are numbered. (A) Wild-type embryo. Note the entering of the arteria vertebralis into the foramen transversarium of vertebra no. 6 (arrow). (B) Embryo exposed to RA on day 10.5. Note entering of the arteria vertebralis into vertebra no. 5 (arrow), presumably also indicating a shift of the tuberculum anterior to no. 5. A cervical rib is associated with no. 7.

Fig. 8.

Anterior transformations at the thoracolumbar transition. (A) Wild-type embryo. No ribs are formed on vertebra no. 21 and no. 22, they show the typical lumbar morphology.(B) Embryo exposed to RA on day 11.5. Note formation of ribs on vertebra no. 21.(C) Embryo exposed to RA on day 11.5. Note formation of ribs on vertebra no. 21, and of minute rib anlagen on no. 22 (arrow).

Fig. 8.

Anterior transformations at the thoracolumbar transition. (A) Wild-type embryo. No ribs are formed on vertebra no. 21 and no. 22, they show the typical lumbar morphology.(B) Embryo exposed to RA on day 11.5. Note formation of ribs on vertebra no. 21.(C) Embryo exposed to RA on day 11.5. Note formation of ribs on vertebra no. 21, and of minute rib anlagen on no. 22 (arrow).

RA exposure on day 11 + 0 hours

Exposure of animals on day 11+0 hours led to posterior transformations of C5, evident as shifts of the tuberculum anterior (16%). In this group a higher percentage of animals (80%) had extensive ossifications or cervical ribs on C7 (Table 1). Most of these animals (91%) exhibited a fourteenth rib on segment no. 21 (Table 1, Figs 3G and 4G).

RA exposure on day 11.5

Cervical ribs were observed in 33% of the animals, that is less frequently than in animals treated on day 10.5 or day 11 + 0 h (Table 1). At least fourteen ribs were present in all animals, four animals bearing minute anlagen of a fifteenth pair of ribs were obtained (transformation i; Table 1, Figs 3H,I; 4H,I; and 8). However, this fifteenth rib never reached the length of the fifteenth ribs generated in the animals exposed to 13-cis RA on day 8.5 (Kessel and Gruss, 1991). The number of lumbar vertebrae was reduced to four in 33% of the animals (Table 1). This led to an anterior shift of the os sacrum in those specimens, where only 25 presacral vertebrae were generated (transformations f and g). In summary, posterior transformations (c, f, g) as well as one anterior transformation (i) were observed in animals of this group.

Vertebral patterns

The vertebral patterns of animals exposed to high doses of retinoic acid at the end and after gastrulation varied dramatically from the highly constant wild-type pattern (C7/T13/L6/S4; Table 1). Ten distinct vertebral patterns induced by RA have been described previously (Kessel and Gruss, 1991). Ten unique types of homeotic transformations (a-1) were identified that cause these changes. The exposures described in the present communication add seven new vertebral patterns. They involved transformations b, c, f, g, h, i, k, and 1, which occurred in unique constellations not observed with early RA exposure (e.g. the propagation of seven lumbar vertebrae, four lumbar vertebrae, and five sacral vertebrae; Table 1).

The distribution of the transformations along the vertebral column suggests two temporal events occurring in a craniocaudal direction (Fig. 9). First, exposure during gastrulation (day 7 and 8) results in transformations along the length of the vertebral column, beginning at progressively more caudal levels with later RA exposures. The direction of transformation is dependent on the most anterior transformation, which directs the subsequent transformations to be exclusively posterior (day 7.3) or anterior (day 8 + 5 hours, day 8.5. day 9.5) transformations. Secondly, exposure during prevertebra formation (day 10-11) leads to transformations affecting first cervical and then also more caudal regions of the vertebral column. In this phase both anterior and posterior transformations occur in one animal.

Fig. 9.

Distribution of transformations in a twofold craniocaudal progression. The body axis is represented vertically, with consecutively numbered vertebral segments (no. l=atlas). Approximate expression boundaries for Hox genes of the different paralog groups (roman numbers) in the paraxial mesoderm (mostly sclerotomes) are given on the right. Transformations are indicated by an arrow pointing upwards for posterior, and downwards for anterior transformations. Definition of transformations a-1 is given in the text and by Kessel and Gruss (1991). The frequencies of the transformations were calculated from Table 1. The data for day 7.3 (10 mg/kg all-trans RA) and day 8.5 (450 mg/kg 13-cis RA) exposures are from Kessel and Gruss (1991), exposures on day 8 + 5 hours, day 9.5, day 10.5, day 11 + 0 hours, and day 11.5 were to 100 mg/kg all-trans RA. The dotted line separates the specification phase (gastrulation) from the respecification phase (somite differentiation).

Fig. 9.

Distribution of transformations in a twofold craniocaudal progression. The body axis is represented vertically, with consecutively numbered vertebral segments (no. l=atlas). Approximate expression boundaries for Hox genes of the different paralog groups (roman numbers) in the paraxial mesoderm (mostly sclerotomes) are given on the right. Transformations are indicated by an arrow pointing upwards for posterior, and downwards for anterior transformations. Definition of transformations a-1 is given in the text and by Kessel and Gruss (1991). The frequencies of the transformations were calculated from Table 1. The data for day 7.3 (10 mg/kg all-trans RA) and day 8.5 (450 mg/kg 13-cis RA) exposures are from Kessel and Gruss (1991), exposures on day 8 + 5 hours, day 9.5, day 10.5, day 11 + 0 hours, and day 11.5 were to 100 mg/kg all-trans RA. The dotted line separates the specification phase (gastrulation) from the respecification phase (somite differentiation).

Transformation h (8 vertebrosternal ribs) occurred in more than 50% of the animals after exposure on day 8 to the low dose (Kessel and Gruss, 1991) or the high dose of RA, but did not occur after exposure on day 9.5. Typical for the day 9.5 exposure was transformation 1 (transformation of the first caudal to a sacral identity) observed in 46% of the animals. On day 10.5 transformation b (shift of tuberculi) in the anterior vertebral column was again observed (44%). It occurred less frequently on day 11.0 (16%) and was not induced on day 11.5 (Fig. 9). Transformation c (cervical ribs) was present in 62% of the animals on day 10.5, rising to 80% on day 11.0 and decreasing to 33% on day 11.5. The frequency of supemumery ribs increased from 21%, to 91% and 100% between days 10.5 and 11.5 Finally a relatively rare effect on the posterior part of the vertebral column was induced by exposure on day 11.5, the os sacrum being shifted rostrally by one segment in 11% of the animals.

Hox gene expression

A major intention of this study was to establish a correlation between RA induced transformations and changes in the expression of Hox genes. The previously described transformations induced on day 7.3, that is relatively early during gastrulation, involved major rearrangements of vertebral patterns and could be correlated with changes in the expression patterns of Hox genes (Kessel and Gruss, 1991). Based on the model derived from these data the anterior transformations h, i, k and 1 should correlate with negative effects on posterior Hox genes, namely genes from the four or five paralog groups in the 5′ third of the Hox clusters. The posterior transformations b, c, f and g on the other hand should correlate with positive effects (see also Figs 9 and 12). Conceptually, embryos can be analysed for a fast effect of RA shortly after administration when the exposure probably is still going on. Alternatively, expression patterns can be studied after the transformations have been established. These latter analyses are however problematic, if the presence of a specific transcripts is causally connected to the presence of a specific structure.

Fig. 10.

Hox gene expression in RA exposed embryos on day 8. A-F show adjacent sagittal sections of an embryo exposed to RA in utero for 4 hours on day 8.5. An untreated control embryo is shown in G-K. In situ analysis was performed as described. The probes were specific for (A) Hox-1.5, (B) S 8, (C,G) Hox-4.5, (D,H) Hox-1.8, (E,I) Hox-1.9. The bright-field photographs (F,K) show the allantois (all), somites (so), the neural fold (nf), the head fold (hf) and the branchial arch mesenchyme (bm). For further description see text.

Fig. 10.

Hox gene expression in RA exposed embryos on day 8. A-F show adjacent sagittal sections of an embryo exposed to RA in utero for 4 hours on day 8.5. An untreated control embryo is shown in G-K. In situ analysis was performed as described. The probes were specific for (A) Hox-1.5, (B) S 8, (C,G) Hox-4.5, (D,H) Hox-1.8, (E,I) Hox-1.9. The bright-field photographs (F,K) show the allantois (all), somites (so), the neural fold (nf), the head fold (hf) and the branchial arch mesenchyme (bm). For further description see text.

Fig. 11.

For legend see p. 498. Hox-1.8 expression in day-12.5 embryos exposed to RA on day 8 + 5 hours. A-D show sagittal sections (original magnification 25×) of day-12.5 embryos exposed to a high dose of RA (100 mg/kg all-trans RA) on day 8 + 5 hours. Note the proper development up to the lumbar level, and the abrupt truncation leaving the posterior part of the embryo in complete disorganization. A shows a bright-field and B a dark-field view of a near midsagittal section. Twenty one prevertebrae are properly formed, a few intestinal loops are discernable in the posterior part. Tongue (t), heart (h), aorta (ao), liver (li), allantoic stalk (al), intestine (in), spina bifida (sb, arrow indicates anterior boundary) and prevertebrae (numbered) are marked. A weak Hox-1.8 signal is present on very posterior cells, in particular of the bulging out neural tube, prevertebrae do not express Hox-1.8. C shows a bright-field and D a darkfield view of parasagittal sections. Note strong expression of Hox-1.8 on the hindlimb, the posterior neural tube, some intestinal loops and posterior mesenchyme. Ribs (r) are numbered in (C). E-H show parasagittal sections (200×) of a day-12.5 wild-type embryo (E,F) and an embryo exposed to an effectively low dose (see text) of RA (450 mg/kg 13-cis RA) (G,H). The typical Hox-1.8 expression of wild-type embryos (weak on no. 20, strong on no. 21 and following ones, H. Haack, personal communication; E,F) is shifted posteriorly by one segment in the exposed embryo. The last prevertebra generating a rib (r) is marked by an arrow, spinal ganglia by (sg), the neural tube by (nt), and prevertebrae are numbered.

Fig. 11.

For legend see p. 498. Hox-1.8 expression in day-12.5 embryos exposed to RA on day 8 + 5 hours. A-D show sagittal sections (original magnification 25×) of day-12.5 embryos exposed to a high dose of RA (100 mg/kg all-trans RA) on day 8 + 5 hours. Note the proper development up to the lumbar level, and the abrupt truncation leaving the posterior part of the embryo in complete disorganization. A shows a bright-field and B a dark-field view of a near midsagittal section. Twenty one prevertebrae are properly formed, a few intestinal loops are discernable in the posterior part. Tongue (t), heart (h), aorta (ao), liver (li), allantoic stalk (al), intestine (in), spina bifida (sb, arrow indicates anterior boundary) and prevertebrae (numbered) are marked. A weak Hox-1.8 signal is present on very posterior cells, in particular of the bulging out neural tube, prevertebrae do not express Hox-1.8. C shows a bright-field and D a darkfield view of parasagittal sections. Note strong expression of Hox-1.8 on the hindlimb, the posterior neural tube, some intestinal loops and posterior mesenchyme. Ribs (r) are numbered in (C). E-H show parasagittal sections (200×) of a day-12.5 wild-type embryo (E,F) and an embryo exposed to an effectively low dose (see text) of RA (450 mg/kg 13-cis RA) (G,H). The typical Hox-1.8 expression of wild-type embryos (weak on no. 20, strong on no. 21 and following ones, H. Haack, personal communication; E,F) is shifted posteriorly by one segment in the exposed embryo. The last prevertebra generating a rib (r) is marked by an arrow, spinal ganglia by (sg), the neural tube by (nt), and prevertebrae are numbered.

Fig. 12.

Hox codes of mesodermal segments and interpretation of RA-induced transformations. Expression domains are indicated (+). Anterior boundaries of expression are marked according to published in situ analyses, where posterior boundaries are in most cases not clearly defined. Here nine expressing segments are marked (for further discussion and references see Kessel and Gruss, 1991). Arrows indicate the posterior (a, b, c, d, e, f, g) or anterior (h, i, k, l) transformations discussed in the text, Figs 4 and 9, and by Kessel and Gruss (1991). Two classes of Hox genes become evident in this scheme: either correlating with posteriorizing (column of right-to-left arrows) or with anteriorizing effects (column of left-to-right arrows) along the vertebral column. In the first group genes from paralog groups VI-XIII (roman numbers) are represented, with the exception of Hox-2.5, the only member of paralog group V (*). All these genes respond positively to RA in tissue culture cells (Boncinelli et al., 1991). The second group comprises genes from paralog groups I-V, all with homeoboxes of the Abd-B type. Genes of this group were shown not to be activated by RA (Hox-3.2, Erselius et al., 1990; Hox-4.4 and Hox-4.5, Simeone et al., 1991; Hox-1.8, Hox-1.9, Hox-4.5, this paper), or to be downregulated by RA (Hox-4.5, Hox-4.6, Hox-4.7, and Hox-4.8, Simeone et al., 1991). An exception here is Hox-4.3 (*) belonging to paralog group VI, and being upregulated by RA in tissue culture cells (see text for discussion of positive RA effects on posterior Hox genes). Note the first loss-of-function effect (anterior transformation h) occurs in the mid-thoracic region posterior to the expression boundary of the last positively responding gene and anterior to the first gene not activated by RA.

Fig. 12.

Hox codes of mesodermal segments and interpretation of RA-induced transformations. Expression domains are indicated (+). Anterior boundaries of expression are marked according to published in situ analyses, where posterior boundaries are in most cases not clearly defined. Here nine expressing segments are marked (for further discussion and references see Kessel and Gruss, 1991). Arrows indicate the posterior (a, b, c, d, e, f, g) or anterior (h, i, k, l) transformations discussed in the text, Figs 4 and 9, and by Kessel and Gruss (1991). Two classes of Hox genes become evident in this scheme: either correlating with posteriorizing (column of right-to-left arrows) or with anteriorizing effects (column of left-to-right arrows) along the vertebral column. In the first group genes from paralog groups VI-XIII (roman numbers) are represented, with the exception of Hox-2.5, the only member of paralog group V (*). All these genes respond positively to RA in tissue culture cells (Boncinelli et al., 1991). The second group comprises genes from paralog groups I-V, all with homeoboxes of the Abd-B type. Genes of this group were shown not to be activated by RA (Hox-3.2, Erselius et al., 1990; Hox-4.4 and Hox-4.5, Simeone et al., 1991; Hox-1.8, Hox-1.9, Hox-4.5, this paper), or to be downregulated by RA (Hox-4.5, Hox-4.6, Hox-4.7, and Hox-4.8, Simeone et al., 1991). An exception here is Hox-4.3 (*) belonging to paralog group VI, and being upregulated by RA in tissue culture cells (see text for discussion of positive RA effects on posterior Hox genes). Note the first loss-of-function effect (anterior transformation h) occurs in the mid-thoracic region posterior to the expression boundary of the last positively responding gene and anterior to the first gene not activated by RA.

In order to study the effect of RA on day 8 + 5 hours, embryos were dissected out 4 hours after RA administration, fixed and prepared for in situ analysis. Antisense RNA probes derived from the S8, the Hox-1.5, Hox-4.5, Hox-1.8, and Hox-1.9 genes were applied. These genes represent a homeobox gene not belonging to the clustered Hox genes (S<S; Opstelten et al., 1991), a Hox gene with a very anterior expression boundary (Hox-1.5’, Gaunt, 1988), and three Hox genes expressed only in the posterior body region {Hox-4.5, Duboule and Dolle, 1989; Hox-1.8 and Hox-1.9, H. Haack et al., unpublished data). The in situ analysis shown in Fig. 10 was performed on adjacent sections of RA-exposed and control embryos. Expression of Hox-1.5 was not affected by RA exposure. As expected Hox-1.5 RNA was found from the posterior region up to the first somites (Fig. 10A). A normal, predominantly mesenchymal expression pattern (Opstelten et al., 1991) was found also for the S8 gene (Fig. 10B) in RA-exposed embryos. No expression was detected with the Hox-4.5, Hox-1.8 and Hox-1.9 probes (Figs 10C,D and E), even though the posterior body region including the allantois was present, it was morphologically indistinguishable from an untreated embryo (Fig. 10F) and positive for a Hox gene from the 3′ region of the cluster (Fig. 10A). In a parallel experiment, with the same probes being applied in the same hybridisation, exposure and development conditions, Hox-4.5, Hox-1.8 and Hox-1.9 expression was detected in an untreated control embryo in posterior mesoderm cells at the base of the allantois (Fig. 10G,H, and I).

In separate experiments embryos were exposed on day 8 + 5 hours and analyzed on day 12.5 by in situ hybridization. The phenotype of these embryos was as expected from the day-19 fetuses described above. The prevertebral column up to prevertebrae no. 19, no. 20 or no. 21 and all organs up to this level were properly developed (Fig. 11A,C). The more posterior region was truncated and severely disorganized. The neural tube bulged out in the prevertebrae less area (spina bifida) and ended in a small appendix. Several loops of the gut were discernable, but probably not connected to the surface (imperforate anus). The genital eminence was not properly formed. Sporadically disorganized pre vertebral condensations were recognizable. The majority of the cells were unordered and mesenchymal. Analysis of the Hox-1.8 expression in these embryos revealed signals on very posterior mesenchyme, on the bulging-out posterior neural tube, on intestinal loops, and on the hindlimb bud, whenever it was generated at all (Fig. 11B,D). No expression was observed on prevertebral cells, even when the normally Hox-1.5-positive pre vertebrae nos 20 and 21 where discernable. In order to produce a milder phenotype, embryos were exposed to 13-cis RA on day 8 + 5 hours. A significant homeotic transformation induced with this treatment is the generation of 14 ribs, reflecting an anterior transformation of vertebra no. 21 (Kessel and Gruss, 1991). In wild-type embryos a weak Hox-1.8 signal is observed on prevertebra no. 20, and stronger signals are seen on no. 21 and onwards (H. Haack, unpublished observations). In the exposed embryo a posteriorization of the Hox-1.8 expression domain by one segment was detected (Fig. 11,E-H), correlating with the one-segment-anter-iorization of morphology.

A potentially positive effect on thoracic Hox genes was studied four hours after or two days after RA administration on day 10.5 or day 11 + 0 hours. Emphasis was put on the Hox-3.1 and the Hox-1.1 gene, which have been thoroughly analysed through all stages of somite differentiation in this laboratory (Breier et al., 1988; Kessel et al., 1987; Mahon et al.,1988) . In wild-type mice Hox-1.1 is extremely weakly expressed in prevertebra no. 10, while only prevertebra no. 11 and the following show the full level of hybridisation (Kessel and Gruss, 1991). By silver grain counting, setting the grain number on prevertebra no. 9 as 0%, and no. 12 as 100%, about 5% of signal was normally found on prevertebra no. 10. Analysis of embryos exposed to RA on day 10.5 or day 11 + 0 hours never revealed a Hox-1.1 signal on prevertebra no. 9. In parallel also an anteriorization of Hox-3.1 expression domain could not be induced. This was in contrast to the embryos treated on day 7.3, which clearly revealed on day 12.5 anteriorized boundaries of expression for the Hox-1.5, Hox-1.1, and Hox-3.1 genes (Kessel and Gruss, 1991). In the absence of qualitative differences, attempts were made to detect quantitative differences, although the in situ technology clearly reached its limits in terms of sensitivity. Therefore, it is not clear if the higher Hox-1.1 signal (about 20% by grain counting) observed on prevertebra no. 10 in several cases of RA-exposed embryos represents a significant difference from the untreated controls (data not shown).

Two different phases of vertebral column development are analyzed in this communication: in the first phase (day 8 and 9) vertebral precursor cells are just being generated by ingression through the late primitive streak. In the second phase the precursors of vertebrae, the somites, differentiate and after complex cellular rearrangments the sclerotome cells begin to form vertebrae in a cranial to caudal succession (for review see Gruss and Kessel, 1991). It is noteworthy, that in both phases the relevant cells leave an epithelium, either the ectodermal sheet or the somite, and enter a migratory phase. In both phases the distribution of anterior and posterior transformations indicates a temporal window of RA sensitivity, which progresses craniocaudally along the vertebral axis. RA interferes with the establishment of mesodermal specification on days 7, 8 and 9. In addition, RA has the remarkable ability to respecify identities of prevertebral segments on days 10 and 11.

Specification

On day 8 + 5 hours a mouse embryo has about 11 somites (4 occipital, and 7 cervical somites) and an estimated length of unsegmented, presomitic mesoderm equivalent to 6 prospective somites (Theiler,1989) . Not counting the occipital segments it can be assumed that mesoderm of the axial level equivalent to vertebral segment no. 13 (T5) is ingressing, when RA is administered on day 8 + 5 hours. The exposed animals show their most rostral effect on segment no. 14 (split vertebral center of T7, Fig. 1B,C) and the first transformation h affects segment no. 15 (T8). Similar calculations reveal that RA treatment on day 9.5 (21-29 somites) also affects segments for which presomitic precursor cells are just ingressing. Transplantation experiments suggest that cells must receive their axial specification during this step (Kieny et al., 1972). Following on from previous results (Kessel and Gruss, 1991) the data presented here indicate that RA directly interferes with this specification process. Analysis of several Hox genes at various embryonic stages have indicated that cells begin to express a particular Hox gene upon leaving the primitive streak. Therefore, it was tempting to discuss Hox genes involved in the specification of axial positions. Corroborating experimental evidence came from ectopic expression in transgenic mice (Balling et al., 1989; Kessel et al.,1990), inactivation of Hox genes by antibodies (Wright et al., 1989) or gene targetting (Chisaka and Capecchi, 1991; Lufkin, 1991), transplantation experiments in the chick limb bud (Izpisua-Behnonte et al., 1991; Nohno et al., 1991) or vertebral reprogramming induced by retinoic acid (Kessel and Gruss, 1991). Such analyses were recently combined in a model suggesting a molecular coding of axial positions by Hox codes (Kessel and Gruss, 1991, see also Fig. 12).

Application of the rules established by Lewis for homeotic genes of Drosophila (Lewis, 1978) on the murine Hox genes would imply that anterior transformations result from negative or loss-of-function events, which in their extreme form may be represented in the observed truncations. The analysis of Hox genes from the third and the fourth paralog group in day-8 embryos has indeed shown that they are not expressed in the posterior part of the RA-exposed embryos, which later is truncated. Analysis of older embryos (day 12.5) showed a posterior shift of the Hox-1.8 expression domain by one prevertebral segment after exposure to 13-cis RA correlating with the generation of a rib on prevertebra no. 21. No Hox-1.8 expression in the normal domains posterior of prevertebra no. 19, but small domains at the end of these truncated embryos were observed after all-trans RA exposure. This expression pattern correlated with the agenesis of posterior structures. These observations suggest that the onset of expression was de novo repressed in the presence of RA either allowing a delayed activation or inhibiting activation completely, dependent on the RA concentration and the length of exposure. Less likely, but not formally excluded, is the possibility that already activated Hox genes were downregulated.

The presented data have indicated a correlation between the formation of specific structures and the expression of Hox genes relevant for a specific axial level. They reinforce the previous argument (Kessel and Gruss, 1991) that transformations induced by RA during gastrulation are homeotic changes of regional identities, correlating with the expression of murine homeotic genes, the Hox genes.

Respecification

Differentiation of the prevertebral column occurs in a craniocaudal direction. On day 10 the somites begin to differentiate into sclerotomes and dermomyotomes. Again progressing craniocaudally, the formation of cartilagenous prevertebrae is achieved around day 12. Coinciding with this second RA responsive phase of the paraxial mesoderm is the beginning of expression of the gamma RA receptor gene in somitic mesoderm, which is first seen in the sclerotomes on day 10.5 (Ruberte et al., 1990). Analysis of vertebral morphologies after RA exposure on day 10 and 11 indicated that the transformations b, c, f, g, and i, which were already observed after RA exposure on day 7.3, can also be induced at these later stages. The extent of the transformations c and i, however, was less pronounced than after exposure on day 7.3. Moreover, no progression of a transformation on all more posterior located segments was observed. As described in the result section, the distribution and frequency of these transformations indicated that each transformation can rather be seen as a specific local event, which does not necessarily evoke an effect in the same direction on all more posterior segments.

In situ analysis was performed in order to find experimental evidence for an anteriorization of Hox codes underlying the posterior transformations induced by RA exposures on day 10 or 11. Significantly and reproducibly these attempts were unsuccsessful. I have not been able to detect the postulated anterior extensions of expression domains. One explanation could be a qualitative switch in the expression pattern of Hox genes, which has been shown to occur in tissue culture cells upon exposure to RA. Differentially spliced transcripts may possess different coding capacities and could escape the in situ analysis. Alternatively, it is also possible that quantitative rather than qualitative differences are involved. RA might be able to modulate the expression of Hox genes only within the domains defined during gastrulation. In the already strongly expressing segments, this may be without morphological consequences. However, a weakly expressing segment at the anterior boundary of an expression domain, may be lifted over a threshold and develop a more posterior identity, as seen in transformations b and c. It is conceivable that this occurs either by inducing higher Hox levels in the already expressing cells or, alternatively, by inducing more cells to transcribe available Hox genes. The latter mechanism has been shown in Drosophila, where a weak transformation seems to correlate with a smaller number of reprogrammed cells, rather than a weaker transformation of individual cells (Postlethwaite and Schneiderman, 1971; Gibson et al., 1990). The analytical tool, RNA in situ hybridization, has not allowed this point to be settled so far for the mouse embryos exposed to RA during somite differentiation.

The Hox code

Fig. 12 summarizes the expression domains of Hox genes in the paraxial mesoderm. The distinct combinations of expressed Hox genes in distinct vertebral segments were recently defined as Hox codes and their functional significance in the specification of axial levels was suggested (Kessel and Gruss 1991). Transformations described in this report and the groups of positively or negatively responding Hox genes are indicated by arrows. The positive response of Hox genes from the 3’ paralog groups (6-13) to RA is well documented (for review see Boncinelli et al., 1991). It could be correlated with gain-of-function effects, i.e. posterior transformations of vertebrae during gastrulation (Kessel and Gruss, 1991). The regulation of the 5’ genes from paralog groups 1, 2, 3, 4, and possibly 5 for which negative as well as positive effects are reported is more complex. Thus, it has been shown in tissue culture cells (Simeone et al., 1991), and here also in vivo, that RA can downregulate several of the posterior Hox genes or inhibit their activation. If high concentrations (also equivalent to long exposure times) are applied to the anterior margin of the chick limb bud also negative effects on posterior Hox genes and formation of less digits are observed (Izpisua-Belmonte et al., 1991). The RA exposures on days 10.5, 11 + 0 hours, and 11.5 also induced limb deformations with reduced digit numbers in the animals described in this report (data not shown). These could result from downregulation of posterior Hox genes during the specification of the limb axes. In contrast, administration of low RA concentrations to the anterior chick limb bud has a positive influence on posterior Hox genes (Izpisua-Belmonte et al., 1991; Nohno et al., 1991). The observed activation of new expression domains in the anterior part of the chick limb bud correlates with induction of mirror image duplications of digits (Izpisua-Belmonte et al., 1991; Nohno et al., 1991). Furthermore, the posterior transformations in the caudal part of the vertebral column induced by exposure to RA on day 7.3 were explained by anterior shifts (i.e. positive responses) of posterior Hox gene expression domains (Kessel and Gruss, 1991). In these experiments the transformations in the posterior body region induced by RA occurred more than 24 hours after exposure, when the exogenous RA had certainly disappeared from the syStern. It appears that in the case of the vertebral transformations the cascade of gene activations proceeds all the way along the cluster, thus transferring a positive RA effect on 3′ genes also into the 5′ region.

It was suggested earlier that the establishment of Hox codes during gastrulation involves two aspects (Kessel and Gruss, 1991; Kessel, 1992): one aspect is the successive opening of a Hox cluster, making genes available for transcription in a 3’ to 5’ direction along the cluster (see also Pfeiffer et al., 1987). This is separable from a second aspect, namely the actual transcription of a Hox gene. While opening and transcription appear to coincide initially, in later stages transcription seems no longer obligatory. If a Hox code represents the molecular basis of axial specification, it must not be possible to change an established code easily. In the experiments described here I found no evidence for a changing of Hox expression domains induced by high doses of RA during somite differentiation. The analysis of the Hox-1.1 and the Hox-3.1 genes during somite differentiation clearly indicated that these particular Hox genes can not be induced by exposure to RA in a segment not open for transcription. In a completely different set of experiments a similar conclusion was reached for neurulating Xenopus embryos (Sharpe, 1991). While the timing and level of a Hox gene could be influenced, the normally nonexpressing, anterior region remained negative even after RA exposure. Although further evidence is necessary, it seems probable that after gastrulation RA can modulate expression of a Hox gene in those regions only where it is open for transcription. In this case an elevated level of a Hox transcript could involve gain-of-function effects at the boundaries of its expression domain, whereas downregulation would be expected to cause loss-of-function effects. Such modulations could be responsible for the observed respecifications of vertebral identities. Besides RA no other agent or procedure has been described up to now, which can respecify segmental identities during somitogenesis. Therefore, the findings reported here reinforce the observations establishing a fink between axial specification and retinoic acid.

I thank P. Gruss for support and comments, S. Gross for excellent technical assistance, M. Gross, P. Tremblay, and H. Haack for discussions, and R. Altschäffel for photography. Probes were generously provided by F. Ruddle (New Haven), F. Meijlinck (Utrecht), and, prior to publication, H. Haack (Göttingen). The project was supported by the Max-Planck-Gesellschaft.

Akam
,
M.
,
Dawson
,
I.
and
Tear
,
G.
(
1988
).
Homeotic genes and the control of segment diversity
.
Development
104
Supplement
,
123
133
.
Alles
,
A.
and
Sullk
,
K.
(
1990
).
Retinoic acid-induced spina bifida: evidence for a pathogenetic mechanism
.
Development
108
,
73
81
.
Balling
,
R.
,
Mutter
,
G.
,
Gruss
,
P.
and
Kessel
,
M.
(
1989
).
Craniofacial abnormalities induced by ectopic expression of the homeobox gene Hox-1.1 in transgenic mice
.
Cell
58
,
337
347
.
Bellairs
,
R.
(
1986
).
The primitive streak
.
Anat. Embryol
.
174
,
1
14
.
Boncinelli
,
E.
,
Simeone
,
A.
,
Acampora
,
D.
and
Mavillo
,
F.
(
1991
).
Hox gene activation by retinoic acid
.
Trends Genet
.
7
,
329
334
.
Breier
,
G.
,
Dressier
,
G. R.
and
Gruss
,
P.
(
1988
).
Primary structure and developmental expression pattern of Hox-3.1, a member of the munne Hox-3 homeobox gene cluster
.
EMBO J
.
7
,
1329
1336
.
Chisaka
,
O.
and
Capecchi
,
M. R.
(
1991
).
Regionally restricted developmental defects resulting from targeted disruption of the mouse homeobox gene Hox-1.5
.
Nature
350
,
473
479
.
Creech-Kraft
,
J.
,
Kochhar
,
D. M.
,
Scott
,
W. J.
and
Nau
,
H.
(
1987
).
Low teratogenicity of 13-cis retinoic acid (isoretinoin) in the mouse corresponds to low embryo concentrations during organogenesis: Comparison to the all-trans isomer
.
Toxicol. Appl. Pharmacol
.
87
,
474
482
.
Duboule
,
D.
and
Dollé
,
P.
(
1989
).
The structural and functional organisation of the murine HOX gene family resembles that of Drosophila homeotic genes
.
EMBO J
.
8
,
1497
1505
.
Durston
,
A. J.
,
Timmermanns
,
L. P. M.
,
Hage
,
W. J.
,
Hendriks
,
H. F. J.
,
De Vries
,
N. J.
,
Heineveld
,
M.
and
Nieuwkoop
,
P. D.
(
1989
).
Retinoic acid causes an anteroposterior transformation in the developing nervous syStern
.
Nature
340
,
140
144
.
Erselius
,
J. R.
,
Goulding
,
M. D.
and
Gruss
,
P.
(
1990
).
Structure and expression pattern of the murine Hox-3.2 gene
.
Development
110
,
629
642
.
Fainsod
,
A.
,
Awgulewitsch
,
A.
and
Ruddle
,
F. H.
(
1987
).
Expression of the murine homeobox gene Hox-1.5 during embryogenesis
.
Dev. Biol
.
124
,
125
133
.
Gaunt
,
S. J.
(
1988
).
Mouse homeobox gene transcripts occupy different but overlapping domains in embryonic germ layers and organs: a comparison of Hox-3.1 and Hox-1.5
.
Development
103
,
135
144
.
Gibson
,
G.
,
Schier
,
A.
,
LeMotte
,
P.
and
Gehring
,
W. J.
(
1990
).
The specificities of Sex combs reduced and Antennapedia are defined by a distinct portion of each protein that includes the homeodomain
.
Cell
62
,
1087
1103
.
Gruss
,
P.
and
Kessel
,
M.
(
1991
).
Axial specification in higher vertebrates
.
Cure. Opinion m Dev. Biol
.
1
,
204
210
.
Hogan
,
B. L. M.
,
Holland
,
P. W. H.
and
Schoffield
,
P.
(
1985
).
How is the mouse segmented?
Trends Genet
.
3
,
67
74
.
Izpisua-Belmonte
,
J.-C.
,
Tickle
,
C.
,
Dolló
,
P.
,
Wolpert
,
L.
and
Duboule
,
D.
(
1991
).
Expression of the homeobox Hox-4 genes and the specification of position in chick wing development
.
Nature
350
,
585
589
.
Kessel
,
M.
(
1991
).
Molecular coding of axial positions by Hox genes
.
Seminars in Dev. Biol
.
2
,
367
–-373.
Kessel
,
M.
,
Balling
,
R.
and
Gruss
,
P.
(
1990
).
Variations of cervical vertebrae after expression of a Hox-1.1 transgene in mice
.
Cell
61
,
301
308
.
Kessel
,
M.
and
Gruss
,
P.
(
1990
).
Murine developmental control genes
.
Science
249
,
374
379
.
Kessel
,
M.
and
Gruss
,
P.
(
1991
).
Homeotic transformations of murine vertebrae and concomitant alteration of Hox codes induced by retinoic acid
.
Cell
67
,
89
104
.
Kessel
,
M.
,
Schulze
,
F.
,
Fibi
,
M.
and
Gruss
,
P.
(
1987
).
Primary structure and nuclear localization of a murine homeodomain protein
.
Proc. Natl. Acad. Sci. USA
84
,
5306
5310
.
Keynes
,
R.
and
Lumsden
,
A.
(
1990
).
Segmentation and the origin of regional diversity in the vertebrate central nervous syStern
.
Neuron
2
,
1
9
.
Kieny
,
M.
,
Mauger
,
A.
and
Sengel
,
P.
(
1972
).
Early regionalization of the somitic mesoderm as studied by the development of the axial skeleton of the chick embryo
Dev. Biol
.
28
,
142
161
.
Kochhar
,
D. M.
(
1973
).
Limb development in mouse embryos. I. Analysis of teratogenic effects of retinoic acid
.
Teratology
,
7
,
289
298
.
Kongsuwan
,
K.
,
Webb
,
E.
,
Housiaux
,
P.
and
Adams
,
J. M.
(
1988
).
Expression of multiple homeobox genes within diverse mammalian haematopoetic lineages
EMBO J
.
7
,
2131
2138
.
Lammer
,
E. J.
,
Chen
,
D. T.
,
Hoar
,
R. M.
,
Agnish
,
A. D.
,
Benke
,
P. J.
,
Braun
,
J. T.
,
Curry
,
C. J.
,
Fernhoff
,
P. M.
,
Grix
,
A. W.
, Lott,I. T.,
Richard
,
J. M.
and
Sun
,
S. C.
(
1985
).
Retinoic acid embryopathy
.
New Eng. J. Med
.
313
,
837
841
.
Lewis
,
E. B.
(
1978
).
A gene complex controlling segmentation in Drosophila
.
Nature
276
,
565
570
.
Lufkin
,
T.
,
Dierich
,
A.
,
LeMeur
,
M.
,
Mark
,
M.
and
Chambon
,
P.
(
1991
).
Disruption of the Hox-1.6 Homeobox gene results in defects in a region corresponding to its rostral domain of expression
Cell
66
,
1105
1119
.
Lumsden
,
A.
(
1990
).
The cellular basis of segmentation in the developing hindbrain
.
Trends Neurosci
.
13
,
329
335
.
Mahon
,
K. A.
,
Westphal
,
H.
and
Gruss
,
P.
(
1988
).
Expression of homeobox gene Hox-1.1 during mouse embryonic development
.
Development
104
Supplement
,
187
195
.
Noden
,
D. M.
(
1983
).
The role of neural crest in patterning of avian cranial skeletal, connective, and muscle tissue
.
Dev. Biol
.
96
,
144
165
.
Nohno
,
T.
,
NoJi
,
S.
,
Koyama
,
E.
,
Ohyama
,
K.
,
Myokai
,
F.
,
Kuroiwa
,
A.
,
Saito
,
T.
and
Taniguchl
,
S.
(
1991
).
Involvement of the Chox-4 chicken homeobox genes in determination of anteroposterior axial polarity during limb development
.
Cell
64
,
1197
1205
.
Opstelten
,
D. J.
,
Vogels
,
R.
,
Robert
,
B.
,
Destree
,
O. H.
,
Deschamps
,
J.
,
Lawson
,
K. A.
and
Mejlinck
,
F.
(
1991
).
The mouse homeobox gene S8 is expressed during embryogenesis predominantly in mesenchyme
Meeh. Dev
.
34
,
29
43
.
Pfeiffer
,
M.
,
Karch
,
F.
and
Bender
,
W.
(
1987
).
The bithorax complex: Control of segmental identity
.
Genes Dev
.
1
,
891
898
.
Postlethwaite
,
J. H.
and
Schneiderman
,
H. A.
(
1971
).
Pattern formation and determination in the antenna of the homeotic mutant Antennapedia of Drosophila melanogaster
.
Dev. Biol
.
25
,
606
635
.
Ruberte
,
E.
,
Dollé
,
P.
,
Krust
,
A.
,
Zelent
,
A
,,
Morriss-Kay
,
G.
and
Chambon
,
P.
(
1990
).
Specific spatial and temporal distribution of retinoic acid receptor gamma transcripts during mouse embryogenesis
.
Development
108
,
213
222
.
Ruiz i Altaba
,
A.
and
Jessell
,
T.
(
1991
).
Retinoic acid modifies mesodermal patterning in early Xenopus embryos
.
Genes Dev
.
5
,
175
187
.
Selleck
,
M. A.
and
Stern
,
C. D.
(
1991
).
Fate mapping and cell lineage analysis of Hensen’s node in the chick embryo
.
Development
112
,
615
626
.
Sharpe
,
C. R.
(
1991
).
Retinoic acid can mimic endogenous signals involved in transformation of the Xenopus nervous syStern
.
Neuron
7
,
239
247
.
Shenefelt
,
R. E.
(
1972
).
Morphogenesis of malformations in hamsters caused by retinoic acid: Relation to dose and stage at treatment
.
Teratology
5
,
103
118
.
Simeone
,
A.
,
Acampora
,
D.
,
Arcionl
,
L.
,
Andrews
,
P.
,
Boncinelli
,
E.
and
Mavilio
,
F.
(
1990
).
Sequential activation of Hox2 homeobox genes by retinoic acid in human embryonal carcinoma cells
.
Nature
346
,
736
766
.
Simeone
,
A.
,
Acampora
,
D.
,
Nigro
,
V.
,
Faiella
,
A.
,
D’Esposito
,
M.
,
Stornaiuolu
,
A.
,
Mavilio
,
F.
and
Boncinelli
,
E.
(
1991
).
Differential regulation by retinoic acid of the homeobox genes of the four Hox loci in human embryonal carcinoma cells
.
Meeh. Dev
.
33
,
215
228
.
Sive
,
H. L.
,
Draper
,
B. W.
,
Harland
,
R. M.
and
Weintraub
,
H.
(
1990
).
Identification of a retinoic acid-sensitive period during primary axis formation in Xenopus laevts
.
Genes Dev
.
4
,
932
942
.
Stern
,
C. D.
(
1990
).
Two distinct mechanisms for segmentation?
Seminars in Dev Biol
.
1
,
109
116
.
Theiler
,
K.
(
1989
).
The House Mouse. Development and Normal Stages from Fertilisation to 4 Weeks of Age
.
Berlin
:
Springer Verlag
.
Tibbles
,
L.
and
Wiley
,
M. J.
(
1988
).
A comparative study of the effects of retinoic acid given during the critical period for inducing spina bifida in mice and hamsters
.
Teratology
37
,
113
125
.
Wright
,
C. V. E.
,
Cho
,
K. W. Y.
,
Hardwicke
,
J.
,
Collins
,
R. H.
and
De Robertis
,
E. M.
(
1989
).
Interference with function of a homeobox gene in Xenopus embryos produces malformations of the anterior spinal cord
.
Cell
59
,
81
93
.