Transcription of a mutant collagen I gene is a cell type and stage-specific marker for odontoblast and osteoblast differentiation

The Movl3 mouse mutant carries a Moloney leukemia (M-MuLV) provirus inserted in the first intron of the α1(I) collagen gene (Schnieke et al. 1983; Harbers et al. 1984). Studies on homozygous embryos and cell lines have shown that the insertion blocks expression of the collagen gene by preventing initiation of transcription (Hartung et al. 1986). Consequently, homozygous carriers of the mutation (Movl3/Movl3 or M/M) cannot synthesize collagen I, the major collagen type of the interstitial matrix, and die as embryos between 11 and 14 days (Jaenisch et al. 1983; Schnieke et al. 1983; Lohler el al. 1984; Kratochwil et al. 1986).

It was therefore unexpected to find that the mutation had no effect on tooth development. Tooth germs derived from homozygous embryos and grown as transplants produced a dentin layer with normal amounts of collagen I. It turned out that odontoblasts, in contrast to all other cell types studied before, are capable of transcribing the Movl3 allele of the crl(I) collagen gene including its proviral insert (Kratochwil etal. 1989). In addition, alveolar bone was seen to develop in these grafts, suggesting that osteoblasts may also be capable of expressing the mutant collagen gene.

The finding that the Movl3 mutation affects expression of the α(I) collagen gene in a cell-specific manner showed that the retroviral insert does not prevent transcription per se and suggested that it interferes with cell-type-specific regulation of the gene. This interpretation implies that odontoblasts, and presumably osteoblasts as well, control transcription of the α (I) collagen gene in a way different from other mesodermal cells (‘fibroblasts’), an assumption consistent with the characteristic differences in the pattern of collagen synthesis between these cell types (for discussion, see Kratochwil et al. 1989). The aim of the present investigation was to test this hypothesis by studying the developmental expression of the mutant collagen gene during the differentiation of odontoblasts and mandibular osteoblasts. Transcription of the Movl3 allele was investigated in heterozygous carriers (which develop normally), by in situ hybridization with a probe specific for the primary transcript of the mutant gene.

We show that the onset of transcription of the Movl3 allele is precisely correlated with the differentiation of odontoblasts in the developing incisor and molars of the lower jaw. In addition, Movl3 transcripts were also detected in osteoblasts of mandibular bone, and again at defined developmental stages. No other cell type of the lower jaw was seen to transcribe the Movl3 allele.

The origin and the characteristics of the Movl3 mouse strain (genetic background: C57BL/6) have been described (Jaenisch et al. 1981, 1983; Schnieke et al. 1983; Harbers et al. 1984). Embryos of dated pregnancies (vaginal plug=day 0) and newborns were genotyped by the length of the EcoRI restriction fragment containing the viral insertion site, as described (Jaenisch et al. 1983; Harbers et al. 1984).

The in situ hybridization procedure essentially follows the protocol of Hogan et al. (1986) and has been described in detail (Kratochwil et al. 1989). Briefly, ³⁵S-labelled RNA probes were obtained by transcribing linearized plasmids from SP6 promoters ([α-³⁵S]UTP with 1000 to 1500 Ci mmol^-1 was from NEN). The cloning of the fragments and the preparation of single-stranded RNA probes has been described (Kratochwil etal. 1989). In this investigation, we have used three different probes, recognizing different regions of the mouse a-l(I) collagen gene, and its Movl3 allele, respectively. These probes are shown in Fig. 1B.

The ‘α (I) collagen probe’ in fact consists of three cloned subfragments of the last exon spanning a total of 1.5 kb. With the exception of the last 38bp of the coding sequence, this probe represents only the’-untranslated region of the mRNA, minimizing the chance of cross-hybridization with transcripts of related collagen genes.

The ‘Movl3 probe’ was derived from the genome of the Moloney leukemia virus representing the mutagenic insert. A 1.65 kb Htndlll-BawHI fragment from the pol-env region was transcribed in a 5’ to 3’ direction with respect to the viral insert. This RNA probe, therefore, does not recognize viral transcripts but (due to the inverted orientation of the insert - see Harbers et al. 1984) hybridizes to transcripts starting from the collagen promoter and proceeding through the viral insert, i.e. it detects transcripts of the Movl3 allele of the α 1(I) collagen gene. Since the viral anti-sense RNA sequence is spliced out with the first intron, this probe can detect unprocessed pre-mRNA only.

The ‘intron probe’ was derived from a 550 bp Stul-Hpal fragment of the first intron. It was used to detect intronic sequences of unprocessed mRNAs of either allele.

All tissues used for in situ hybridization were frozen immediately after dissection without prior fixation. Rapid freezing appeared to be essential for the detection of unprocessed mRNAs. Subsequent treatment of 6 ;zm frozen sections followed the protocol of Hogan et al. (1986) with an additional 4h prehybridization step at 50°C to reduce entrapment of the probe by the hydroxyapatite of mineralized dentin or bone (see Snead et al. 1988). Photographic emulsion was Ilford’s K5 (diluted 1:1) and exposure time varied from 5 to 50 days. After fixation, tissues were stained with toluidine blue or haematoxylin and embedded in methacrylate.

All sections used for in situ hybridization were from dated heterozygous Movl3/+ (M/+) animals, which carry both the wild-type and the mutant allele of the al(I) collagen gene. In this material, our ‘αl(I) collagen probe’ recognized transcripts of both alleles, while the ‘Movl3 probe’ served as specific indicator for the activity of the mutant allele. Unlike the first probe, however, the ‘Movl3 probe’ could detect only the short-lived primary transcript of the mutant gene, and this should be borne in mind in estimating the transcriptional activity from the autoradiographs. For direct comparison, we have used the ‘intron probe’ which also recognized primary transcripts only, but of both alleles (see Methods and Fig. 1).

Because of the inverted orientation of the proviral insert, transcripts starting from the collagen promoter of the Movl3 allele can clearly be distinguished from viral transcripts by the use of strand-specific in situ probes. This is the more important as the leukemia virus of the Movl3 insertion becomes active during fetal life, with infection then spreading rapidly to virtually all tissues of the fetus and newborn (Jaenisch et al. 1981). We have detected viral transcripts from 16.5 days onwards, by in situ hybridization with the opposite strand of the ‘Movl3 probe’ (in preparation). Both strands of this sequence thus recognize ‘sense’ transcripts, though of different genes, and this situation severely compromises the use of +/- strand probes as controls. We have therefore used tissues of C57BL/6 ‘wild-type’ (with respect to the Movl3 mutation) animais as controls, with at least two animals per stage. In none of these sections did the ^lMovl3 probe’ give a signal, whereas the hybridization pattern obtained with the ‘αl(l) collagen probe’ was the same as in heterozygous material. Moreover, in a large number of paired sections, we have obtained signals with the ‘Movl3 probe’ only in stages and areas where the ‘ol(I) collagen probe’ indicated intensive transcription of the collagen gene (see Figs 3-5, and 7-12).

Tooth development in the lower jaw starts on day E12 with the ingrowth of the dental epithelium (to give rise to the enamel organ), followed by the formation of the mesenchymal dental papilla from which the odontoblasts arise (see Fig. 2). In situ hybridization with the cvl(I) collagen probe did not indicate elevated transcription of the gene in the prospective tooth mesenchyme of 12.5 or 13.5 day embryos (not shown). At the early cap stage of the incisor (day 14/15), the mesenchyme of the dental papilla does not show more silver grains than the loose mesenchyme of the jaw, in sharp contrast to the intense signal obtained from the osteogenic region around Meckel’s cartilage (see below - Fig. 10). Even at 15.5 days, when the anlage of the incisor has considerably grown in length, transcription of the crl(I) collagen gene remains undetectable in the papilla (Fig. 3). At this stage, its outermost cells have not yet formed the epithelial odontoblast layer.

On day 16, finally, transcripts of the αl(I) collagen gene appear in odontoblasts of the incisor. The auto-radiographs after in situ hybridization with the αl(I) collagen probe indicate that the gene is now expressed at a high rate (Fig. 4A). Transcription starts at the tip and is more intense at the labial (front) side of the tooth, which is consistent with the differentiation programme of the tooth and the characteristic asymmetry in dentin deposition. At the same developmental stage, the first transcripts of the MovlS allele became detectable in the odontoblasts at the tip of the incisor (Fig. 4B). The longer exposure time required and the dotted distribution of the signal indicate that the Movl3 probe in fact detects unstable sequences localized within nuclei, i.e. the bona fide primary transcript of the mutant allele. The localization of the signal precisely matched that obtained with the αl(I) collagen probe, except that it does not extend as far proximally along the tooth, which most likely reflects the lower sensitivity of a test for an intronic sequence.

We have followed the expression of both alleles of the ci’l(I) collagen gene in the incisor until 2 days after birth. Transcripts could be detected with the crl(I) collagen probe, as well as with the Movl3 probe during all stages examined (Figs 5, 8). Activity of the Movl3 allele was seen only in the odontoblast layer and in mandibular bone (below), whereas the cvl(I) collagen probe detected lower level expression of the collagen gene in other tissues of the jaw, as well. The distribution of silver grains suggests that most, if not all, odonto-blasts of the incisor transcribe both the wild-type and the Movl3 allele of the collagen gene.

Molars develop later than the incisor; the first post-mitotic and fully differentiated odontoblasts were reported to appear at the apex of the cusps in 18.5 day first molars (Ruch et al. 1976; Ruch, 1985). At earlier stages, from 15.5 days onwards, the al(I) collagen probe did not reveal significant transcription of the gene in cells of the papilla (Fig. 6, see also Fig. 5). As seen in the incisor, the papilla of the immature molar anlage appears to produce very little collagen I, a result consistent with immunohistochemical data (Thesleff et al. 1979; Lesot et al. 1981). The earliest and still very faint hybridization signal was obtained with the a-l(I) collagen probe in the cusp of one 18.5 day first molar; the same tooth was negative when hybridized with the Movl3 probe (not shown). On the following day (day 19, at term), however, intensive transcription of the 0-1(1) collagen gene was seen at the tips and along the posterior ridges of the cusps, and transcripts of the Movl3 allele also became detectable (Fig. 7). Two days after birth, the o-l(I) collagen probe revealed abundant transcripts in the entire odontoblast layer of the first molar, and first signs of transcription in the second molar. (The third molar was not studied as its development starts only in the first week after birth.) At all stages of molar development, transcripts of the Movl3 allele were found exclusively in the odontoblast layer (Fig-8)-

In several cases not all cells of the molar odontoblast layer were labelled with the Movl3 probe (see Fig. 8B). We have therefore hybridized adjacent sections with the intron probe, which detects primary transcripts of the mutant as well as the wild-type allele of the α(l) collagen gene. As seen in Fig. 8C, this probe apparently hybridized to all odontoblasts. Such a comparison suggests that the Movl3 allele may not be active in all odontoblasts at all stages.

As expected, in situ hybridization with the αl(I) collagen probe revealed intensive transcription in the developing mandibular bone. The earliest indication of elevated collagen I synthesis was seen in the osteogenic area around Meckel’s cartilage in 13.5 day embryos (Fig. 9A). At this stage, transcripts of the mutant allele could not yet be detected by in situ hybridization with the Movl3 probe (Fig. 9B). However, since some Movl3 homozygous embryos survive to 13.5 days, we could use them to assay for the expression of the mutant allele with the n-l(I) collagen probe, which offers the advantage of hybridizing to a stable sequence in the transcript (the 3’-exon). As shown in Fig. 9C, this probe revealed activity of the mutant gene in the osteogenic region of 13.5 day jaws. One day later, when bone tissue had become morphologically recognizable, signals were also obtained with the Movl3 probe (Fig. 10). Small clusters of silver grains, again suggesting nuclear localization of the transcripts detected, were found exclusively along the bony matrix. Positive results were obtained with the Movl3 probe at all later stages examined (to postnatal day 4) and the signals were always strictly associated with bone tissue (see Figs 3-5, 10, 11). Both the temporal and the spatial pattern of in situ hybridization strongly suggest that the Movl3 probe detects transcripts in osteoblasts.

It became quite apparent that the Movl3 probe labelled only few cells in each section, considerably less than what would have been expected from the results obtained with the αl(I) collagen probe. We have therefore assayed adjacent sections with the two probes recognizing intronic sequences. In direct comparison, such as the one shown in Fig. 12, it was consistently found that the ‘intron probe’ (hybridizing to primary transcripts of both alleles) labelled far more cells than the Movl3 probe. This result indicates that the scarcity of the Movl3 signal is not (only) due to the instability of the part of the transcript recognized, but rather suggests that the Movl3 allele is not expressed by all bone cells.

Our results show that the Movl3 allele of the αl(I) collagen gene is transcribed in a much more limited range of tissues than its wild-type allele. Expression of the mutant gene was followed in heterozygous animals throughout normal tooth development in vivo, by assaying for the transcription of the mutagenic proviral insert within the first intron. Because of the inverted orientation of the M-MuLV genome, our in situ probe was designed to recognize viral anti-sense RNA and thus allowed us to unequivocally discriminate transcripts of the mutant collagen gene from (possible) virus expression. The long exposure times required (low abundance) and the dotted distribution of silver grains, suggesting nuclear localization, indicated that the Movl3 probe detected an unprocessed mRNA precursor. Moreover, signals were obtained only in sections of carriers of the Movl3 mutation, and only in stages and from tissues where the αl(I) collagen probe indicated high expression of the collagen gene. Taken together, these data provide good evidence for our interpretation that the ^lMovl3 probe’ detects primary transcripts of the mutant collagen gene.

The results indicate that the onset of Movl3 allele transcription indeed coincides with the differentiation of the odontoblast, in the incisor as well as in the two anterior molars of the lower jaw. The temporal and spatial sequence of tissue differentiation and matrix deposition in teeth is known at some detail (see Thesleff and Hurmerinta, 1981; Kubler et al. 1988; Snead et al. 1988). In the rodent incisor, terminal differentiation of the odontoblasts proceeds from the tip down, with the front (labial) aspect leading, where more dentin is deposited than on the lingual side. Expression of the Movl3 allele precisely follows this pattern (Figs 4, 5). In the molars, cytodifferentiation and matrix deposition starts at the tips of the cusps, proceeding downward faster at the posterior edge in teeth of the lower jaw (Snead et al. 1988). Again, this pattern is precisely reflected in the expression of the Movl3 allele as shown by in situ hybridization (Fig. 7). The mutant collagen gene is thus transcribed in a cell-type- and stage-specific manner, and our results indicate that it becomes active in heterozygous odontoblasts as soon as these cells start producing collagen I for deposition in the dentin layer. The pattern of in situ hybridization suggests that the mutant allele is expressed by most or all odontoblasts of the incisor, but apparently not so in the molar. It is, however, conceivable that transcription at a low level escapes detection by the intronic Movl3 probe.

In addition to the odontoblast, we have found another cell type that can transcribe the Movl3 allele. From the distribution of the signal and its developmental appearance we assume that these are osteoblasts of mandibular and alveolar bone. This conclusion is also supported by our previous finding that transplants derived from Movl3 homozygous embryos were capable of forming alveolar bone (Kratochwil et al. 1989). Although the Movl3 probe detected the first transcripts only at 14.5 days, the more sensitive test with the 0-1(1) collagen probe in homozygous embryos showed that the mutant collagen gene is already expressed at low levels one day earlier. By this criterion, the first osteoblast in the mandible would thus appear at 13.5 days p.c. At all stages, however, the hybridization patterns obtained with the Movl3 probe, and with the intron probe, respectively, were different. While the strength of the. signal per cell was about the same for both probes, the intron probe labelled far more cells in bone than the Movl3 probe (despite the fact that the target sequence for the latter probe was three times longer). It is therefore unlikely that this difference, which was not seen in the odontoblasts of the incisor, reflects different sensitivity of the two probes. It rather suggests that not all cells in bone transcribe the mutant gene at the same high rate as its wild-type allele. A heterogeneity in the bone cell population had also been found with respect to the expression of the bone-characteristic gene Spp 1 (secreted phosphoprotein 1, or osteonectin, or SPARC; Holland et al. 1987; Nomura et al. 1988). In molar teeth and alveolar bone, Sppl is expressed in a pattern conspicuously similar to that of the Movl3 allele (cf. Figs 6 and 8 in Holland et al. 1988 with our autoradiographs). On the other hand, it is also possible that the mutant collagen allele is active only at some specific stage of osteoblast maturation.

Odontoblasts and osteoblasts have in common that they secrete large amounts of a collagenous and eventually mineralizing extracellular matrix. The collagen, which amounts to 90% of the organic matrix, is (almost) all type I, and there is no collagen III present in either dentin or bone (Veis, 1984); in the mamma), odontoblast and osteoblast appear to be the only two cell types that synthesize collagen I without producing any collagen HI. In the lower jaw, osteoblasts and odontoblasts are also related by origin as both are derived from head neural crest (see Lumsden, 1988; Noden, 1988). However, since we have also found transcription of the Movl3 allele in long bone derived from lateral plate mesoderm (in preparation), but not in non-skeletal derivatives of head mesectoderm, lineage relationship appears to be less important than common differentiative function. We have argued (Kratochwil et al. 1989) that the exclusive production of collagen I in odontoblasts and osteoblasts may necessitate transcriptional regulation that is different from the fibroblast, where the synthesis of collagens 1 and III is more or less coordinately regulated (Liau et al. 1985). This type of regulation may enable the cells to transcribe the mutant allele. Our finding that the expression of the Movl3 allele has all the features of a reliable and unique marker for odontoblast and osteoblast differentiation is consistent with such an hypothesis.

The transcriptional regulation of collagen genes is currently under active investigation, with special interest focusing on possible control elements in the first intron (Schmidt et al. 1986; Rossi and de Combrugghe, 1987; Rossouwe/a/. 1987; Horton et al. 1987; Bornstein et al. 1987, 1988; Bornstein and McKay, 1988; Burbelo et al. 1988; Rippe et al. 1989). The Movl3 mutant, with its 9 kb proviral insert close to the 5’-boundary of the first intron, may prove to be helpful in this analysis. Our finding that the insert does not inhibit transcription in two distinct cell types has revealed unexpected complexities in the regulation of the crl(I) collagen gene: the conditionally mutagenic effect of this insertion may indicate that odontoblasts and osteoblasts on one side, and ‘fibroblasts’ on the other, use different c/s-regulatory elements to control transcription of the same gene.

We thank Dr R. Jaenisch (Hamburg, now Whitehead Institute) for generously making the Movl3 mutant available to us. The excellent technical assistance of 1. Gmachl (Salzburg) and K. Haase and U. Müller (Hamburg) is gratefully acknowledged. This study was supported by the Austrian Fonds zur Fdrderung der wissenschaftlichen Forschung and by the Deutsche Forschungsgemeinschaft. The Heinrich-Pette-Institut is supported by the Freie und Hansestadt Hamburg and the Bundesministerium für Jugend, Familie, Frauen und Gesundheit.