Rat Maf Related Genes: Speci®C Expression in Chondrocytes, Lens and Spinal Cord

SHORT REPORT Rat maf related genes: speci®c expression in chondrocytes, lens and spinal cord

Masaharu Sakai1, Junko Imaki2, Kazuhiko Yoshida1, Akihiko Ogata1, Yuko Matsushima-Hibiya1, Yoshinori Kuboki3, Makoto Nishizawa4 and Shinzo Nishi1

1Department of Biochemistry, School of Medicine and 3Department of Biochemistry, School of Dentistry, Hokkaido University N15 W7, Kita-ku, Sapporo 060; 2Department of Anatomy, Nippon Medical College, Sendagi, Bunkyo-ku, Tokyo 113; 4Department of Molecular Oncology, Kyoto University, Faculty of Medicine, Yoshida-Konoecho, Sakyo-ku, Kyoto 606, Japan maf is a family of oncogenes originally identi®ed from tion domain is truncated. MafK forms a heterodimer avian oncogenic retrovirus, AS42, encoding a nuclear with NF-E2 p45, a transcription factor that regulates bZip transcription factor. We have isolated two maf the globin gene (Andrews et al., 1993; Igarashi et al., related cDNA clones, maf-1 and maf-2, from a rat liver 1994). Recently it was found that MafB interacts with cDNA library. Comparison of the sequence homologies Ets-1, a transcription factor containing a helix ± turn ± of the proteins encoded by maf-1 and maf-2 with those of helix DNA binding domain, inhibits Ets-1 activity, c-maf and chicken mafB indicated that maf-1 and maf-2 thus interrupting erythroid dierentiation (Sieweke et are the rat homologues of mafB and c-maf, respectively. al., 1996). Maf-family transcription factors heterodi- Both genes are expressed at low levels in a wide variety merize not only with bZip proteins, but also with of rat tissues, including spleen, kidney, muscle and liver. helix ± turn ± helix proteins. Thus, a wide variety of Immunohistochemical studies and in situ hybridization Maf-heterodimers are formed, which in turn could analyses show that maf-1 and maf-2 are strongly regulate a wide variety of target genes depending on expressed in the late stages of chondrocyte development which heterodimer is formed. The mouse mafB gene in the femur epiphysis and the rib and limb cartilage of was identi®ed as the gene responsible for the 15 day old (E15) embryo in rat. Cartilage cells, induced segmentation of hind brain during early development by subcutaneous implantation of bone morphogenic by analyses of kreislar (Kr) mutant mice (Cordes and protein, also expressed maf-1 and maf-2. In situ Barsh, 1994). These ®ndings indicate that Maf hybridization analyses of E15 embryos show that both oncogene products play an important role in morpho- genes are expressed in the eye lens and the spinal cord as genic processes and cellular dierentiation. well as the cartilage. However, the expression patterns of Determination of maf expression should facilitate maf-1 and maf-2 in lens and spinal cord are dierent. identi®cation of target genes and would thus further clarify maf and target gene functions. In this report, we Keywords: maf; transcription factor; chondrocyte; lens; isolated two maf related cDNA clones, maf-1 and maf- spinal cord 2, from a rat liver cDNA library and studied the expression of these genes. Approximately 16106 recombinant phages from a The maf oncogene (v-maf) was initially identi®ed from rat liver cDNA library were screened with a v-maf an avian oncogenic retrovirus, AS42, which induces probe, and two clones, maf-1 and maf-2, were obtained musculoaponeurotic ®brosarcoma in vivo and trans- (nucleotide sequences of both cDNAs were submitted forms chicken embryo ®broblast cells in vitro to GenBank/EMBL/DDBJ DNA database, accession (Nishizawa et al., 1989). The v-maf oncogene product numbers are U56241 and U56242 for maf-1 and maf-2, is a transcription factor containing a typical bZip DNA respectively). binding domain. The binding consensus sequence of The predicted amino acid sequences of Maf-1 and the v-maf product (Maf recognition element, MARE, - Maf-2 show great similarity with that of v-Maf. The TGCTGACTCAGCA-) overlaps with the TPA respon- overall amino acid sequence homologies of Maf-1 and sive element (TRE), and cyclic AMP responsive Maf-2 with those of chicken MafB and c-Maf element (CRE) (Kataoka et al., 1994b). The maf (Kataoka et al., 1994a) are 84% and 81%, respec- oncogene products can form homo- and hetero-dimers tively; thus rat Maf-1 and Maf-2 appear to be the rat with transcription factors having bZip structures, such homologues of MafB and c-Maf. The bZip structures as Jun and Fos (Kataoka et al., 1994a,b, 1996; of all Mafs are conserved at more than 80% homology. Kerppola and Curran, 1994). Furthermore, several This suggests that these Maf-related proteins recognize maf-related genes have been identi®ed (i.e. mafB, similar sequence elements. The DNA binding analyses mafK, mafG, mafF and Nrl) (Fujiwara et al., 1993; using Ecoliexpressed proteins show that both Maf-1 Kataoka et al., 1994a, 1995; Swaroop et al., 1992). and Maf-2 bind with almost equal anity to the Interestingly, mafK, mafG and mafF encode small MARE consensus sequence. However, when subtle proteins in which the N-terminal putative transactiva- variations of this sequence were used, Maf-1, Maf-2 and v-Maf showed signi®cant dierences in binding anity (manuscript in preparation). Maf-1 is almost identical (98% homology) with the Correspondence: M Sakai Received 23 May 1996; revised 7 October 1996; accepted 8 October Kr gene product, a mouse mafB homologue (Cordes 1996 and Barsh, 1994). Expression of Rat maf-related oncogenes MSakaiet al 746 The transactivation functions of the proteins hybridization studies using 15 day old (E15) rat encoded by maf-1 and maf-2 were con®rmed by a co- embryos clearly show that maf-2 is strongly expressed transfection assay using a fusion construct consisting of in the cartilage of ribs (Figure 3c) and limbs (Figure a MARE and luciferase gene. Figure 1 shows both 3d). Almost identical results were obtained using the maf-1 and maf-2 activate the gene containing MARE maf-1 probe. These ®ndings suggest that Mafs are with activities similar to those of v-maf. The activities closely associated with a chondrocyte dierentiation were much stronger than c-jun, although MARE process not only in the adult epiphysis but also in contained complete TRE (TPA responsive element, - embryonic cartilage tissue. To con®rm the expression TGACTCAG-) sequence. of Mafs in chondrocytes, we used a BMP (bone We investigated the tissue distribution of maf-1 and morphogenic protein) implantation system to induce maf-2 mRNAs by RNase protection analysis (Figure cartilage formation. BMP stimulates immature me- 2). maf-1 is expressed at low levels in a variety of senchymal cells to create a process similar to organs including spleen, muscle and liver. The tissue endochondral ossi®cation when they are subcuta- distribution of maf-2 mRNA is similar to that of maf- 1, but not identical. Immunohistochemical staining by speci®c antisera against Maf-1 and Maf-2 did not detect any signi®cant positive signals in liver, though mRNA is expressed there (Figure 2). However, anti-Maf antibodies strongly and speci®cally stained chondrocyte nuclei M located in the hypertrophic chondrocytes of the femur — epiphysis, and the immature proliferating chondrocytes — Brain — Heart — Lung — Spleen — Liver — Kidney — Stomach — Intestine — Uterus — Muscle — Bone marrow were stained very weakly (Figure 3a and b).

Hypertrophic chondrocytes are maturated non-divid- ing cells in chondrocyte dierentiation. The antiserum ¨ against Maf-1 yields almost the same results as maf1 obtained with anti-Maf-2 antiserum. RNase protection analyses detected mRNA of both mafs from femur epiphysis but signals were weak (data not shown). This is probably due to the fact that the expressed cells are low in number in the dissected samples of femur

epiphysis as shown in Figure 3a and b. In situ

¨ maf2 ¨ GAPDH

Figure 2 Expression of maf-1 and maf-2 mRNA in rat tissues, examined by RNase protection analysis. The RNAs from various tissues were prepared using an RNA extraction kit (Nippon Gene Figure 1 Transactivation of MARE-luciferase gene by Mafs. Co. Toyama, Japan). Nucleotides 829 ± 1037 and 1248 ± 1417 of MARE (-TGCTGACTCAGCA-) was ligated into the luciferase the maf-1 and maf-2 clones, respectively, were inserted into expression vector, pGV-P (Nippon Gene, Toyama, Japan), and pBluescript II vector (Stratagene, California, USA), and anti- co-transfected with indicated expression plasmids into HepG2 sense RNA probes were prepared (approximately 16109 c.p.m./ cells, a human hepatoma cell line. Transfection and determination mg). Ten micrograms of the total RNAs from indicated tissues of luciferase activity were done as described (Sakai et al., 1995). were used for RNase protection analysis as described previously The v-maf and c-jun expression plasmids were previously (Sakai et al., 1992). The same preparations of RNAs were described (Sakai et al., 1995). Expression plasmids of maf-1 and quantitated using an anti-sense probe for glyceraldehyde 3- maf-2 were constructed by inserting the cDNA fragments phosphate dehydrogenase (GAPDH). HpaII digested pUC18 corresponding to amino acids 18-C-terminus and 19-C-terminus DNA, terminally labeled by 32P, was used as a size marker for maf-1 and maf-2, respectively, into a human b-actin expression (M). Autoradiograms were exposed for 2 days and overnight for vector (Gunning et al., 1987). Vec indicates expression vector mafs and GAPDH, respectively. This experiment was done three without insert times with nearly identical results Expression of Rat maf-related oncogenes M Sakai et al 747

Figure 3 Maf expression in chondrocytes. (a and b): The speci®c antibodies of Maf-1 and Maf-2 were prepared as follows: Fragments of Maf-1 and Maf-2 proteins were produced in E coli as fusion proteins with maltose binding protein and glutathione S- transferase, respectively. maf-1 cDNA fragment corresponding to amino acids 49 ± 122 was inserted into pMal-c2 vector (New England Biolab, USA) and a Maf-2 cDNA sequence encoding amino acids 64 ± 158 was inserted into pGEX-T2 vector (Pharmacia LKB, Sweden). The fusion proteins were puri®ed according to the supplier's instructions, used to immunize rabbits, and antisera were anity puri®ed by the corresponding antigens. Rats were sacri®ced by perfusion ®xation via the aorta with 4% freshly prepared paraformaldehyde, and axial sections of knee joints were made. The sections were stained by anti-Maf-1 antibody using an avidin-biotin-peroxidase complex method. (c and d): Expression of maf-2 mRNA in E15 rat ribs (c) and limbs (d) analysed by in situ hybridization. The sections were dried and immersed in 1% Triton X-100 in 50 mM Tris/HCl, 25 mM EDTA (pH 8.0) and acetylated with acetic anhydride. The sections were dehydrated and hybridized with the probes (16106 d.p.m./ml) at 55 ± 608C overnight. Slides were rinsed in 46SSC, digested with RNase A (20 mg/ml) and washed sequentially in 26SSC, 16SSC, 0.56SSC, then for 30 min in 0.16SSC at 608C. The sections were exposed to X-ray ®lm (Kodak X-omat) for 3 days, then dipped in NTB2 nuclear emulsion, (1 : 1 with water, Kodak), and exposed for 2 weeks. Radioactive RNA probes of maf-1 and maf-2 (nucleotides 195 ± 410 and 1248 ± 14170, respectively) were synthesized with [a-35S]UTP by T7 RNA polymerase. As a control for nonspeci®c labeling, the adjoining series were treated with RNase A (20 mg/ml) for 30 min at 378C and then hybridized as above. In control experiments, adjacent sections were treated with RNase before hybridization or sense-strand cRNA probes were used for hybridization. Scale bar equals 480 mm for a, 100 mm for b,50mm for c and 250 mm for d neously implanted with an insoluble carrier (Urist et protection analyses together with RNAs from liver and al., 1984). When a ®brous glass membrane is used as a kidney as references. Figure 4c shows that maf-1 is BMP carrier, cartilage formation is induced. Induction expressed strongly in both 1 and 2 week old implants. of cartilage cells in this system is evidenced by type II maf-2 is also expressed, though not as strongly, and the collagen synthesis (Kuboki et al., 1995). The ®brous expression decreased at 2 weeks. glass implants were removed after 2 weeks and In situ hybridization studies of E15 rat embryos examined by Alcian blue staining and immunohisto- indicate that both mafs are strongly expressed not only chemical staining using anti-Maf-1 antiserum. Alcian in cartilage tissue but also in the eyes and spinal cord, blue stains proteoglycan is known to be speci®cally and have similar hybridization patterns. However, produced by chondrocytes. Figure 4a and b show that when examined in detail, the expression locations anti-Maf-1 positive cells correspond to the Alcian blue were dierent in the eyes and spinal cord. The maf-1 stained cells. Both antisera, anti-Maf-1 and anti-Maf-2 and maf-2 genes were expressed in the lens but not in gave apparently identical results. As in the epiphysis the retina of the E15 rat eyes (Figure 5). Expression of chondrocytes (Figure 3a and b), the signals of bigger, maf-1 was con®ned to the equator of the lens, while hypertrophic-like cells were stronger than those of maf-2 was distributed throughout the lens (Figure 5b small, immature cells. RNase protection analyses in the and c). Immunohistochemical staining of E16 rat lens BMP implanted tissues were carried out (Figure 4c). using speci®c antisera against Maf-2 protein strongly The implants, 1 and 2 weeks after implantation, were and speci®cally stained nuclei in the outer equatorial removed and total RNAs were analysed by RNase epithelium and lens ®bers (Figure 5d and e). The Expression of Rat maf-related oncogenes MSakaiet al 748 c BMP Implanted

— — Liver — Kidney —1W —2W

¨ maf-2

¨ maf-1 ¨ GAPDH

Figure 4 Maf expression in BMP implanted tissues. BMP (0.3 mg) was absorbed to a piece of ®brous glass membrane GA-100 (106561 mm, Advantec Co. Tokyo, Japan) and was implanted subcutaneously in the back of rats (Wistar, male, 4 weeks old) as described previously (Kuboki et al., 1995). After 1 or 2 weeks, the implants were removed and subjected to immunohistochemical staining and RNA preparation. Two week old BMP implanted tissue was (a) stained with Alcian blue and hematoxylin, and (b) was stained with anti-Maf-1 antiserum as described in the legend for Figure 3. (c) The total RNAs from implants were analysed by RNase protection analysis using maf-1 (lower panel) and maf-2 (upper panel) anti-sense RNA probes as described in the legend for Figure 2

developmental expression of maf-1 and maf-2 shows a Figure 6 shows the mafs expression in spinal cord. It number of interesting correlations with the changing is interesting that each gene is expressed in dierent pattern of cell proliferation and dierentiation in the regions of spinal cord. The maf-1 and maf-2 mRNA developing rat lens. The most striking of these is the positive cells in the spinal cord could be divided into observation that levels of both gene products rise in the three groups according to a previous report about nuclei of the dierentiating outer epithelial cells. It was somatostatin expression (Senba et al., 1982). The ®rst also interesting that maf-1 mRNA was con®ned to the two groups of cells are located in the dorsal part of the equator of the lens where the epithelial cells were just dorsal horn and the ventral part of the dorsal horn, dierentiating to ®ber cells, in contrast to maf-2 and the third is located in the ventral horn. For E15, a mRNA which is detectable after the dierentiation of strong signal of maf-1 mRNA was observed in the lens ®ber cells. These results suggest that both genes ventral part of the dorsal horn and the ventral horn of play a role in the dierentiation from epithelial cells to the spinal cord. For maf-2, strong mRNA signals were ®ber cells, whereas the maf-2 mRNA also participates observed in the dorsal and ventral parts of the dorsal in the further dierentiation of lens ®ber cells. horn and a much weaker signal was observed in the Expression of Rat maf-related oncogenes M Sakai et al 749 ventral horn. Although details of cell lineages and function of these regions in the spinal cord are not known, the dierent but related expression patterns of maf-1 and maf-2 suggest there are slight functional dierences. This result shows that the initial differ- entiation and development of maf-1 and maf-2 mRNA containing cells takes place during the prenatal period of rats, before the establishment of normal synaptic transmission. Recently the expression pattern of Nrl, a member of the maf-family transcription factor from mouse was reported (Liu et al., 1996). Nrl is expressed mainly in the retina of E14.5 embryo, but weak expression was seen in the equatorial region of the lens where both maf-1 and maf-2 are expressed. The maf-1 and maf-2 expression patterns are clearly dierent from Nrl, which is evenly expressed throughout the whole spinal cord in E15.5. Both maf genes are expressed predominantly in post-mitotic cells, suggesting that the Mafs play a role in the ®nal dierentiation process in these cells. Nrl is also expressed in post-mitotic cells of neurons (Liu et al., 1996). This gene family may play a similar role though in dierent cell types. Figure 5 mafs expression in the lens. Sagittal sections of E15 rat Study of the Kr mutant mouse indicates that mafBis lens were (a) Nissl-stained, and in situ hybridization of mRNA for expressed in the rhombomere, and has an important (b) maf-1 and (c) maf-2 was performed. Immunoreactivity for function in segmentation of the rhombomere in the Maf-2 protein in the horizontal section of E16 rat lens (d) low magni®cation and (e) high magni®cation. Bar indicates 250 mm very early stages of development (Cordes and Barsh, for a, b, c, d and 50 mm for e 1994). This ®nding, together with our observations,

Figure 6 Expression of mafs mRNA in E15 rat spinal cord. Dark-®eld photographs of sagittal section in which mRNA was detected by in situ hybridization (see Figure 3 legend) for (a) maf-1 and (b) maf-2. Autoradiographs of cross section of the spinal cord for (c) maf-1 and (d) maf-2. Cresylviolet was used to stain the adjacent section in panel (e). Sense probes for maf-1 and maf-2 showed no hybridization (data not shown). Scale bar equals 150 mm Expression of Rat maf-related oncogenes MSakaiet al 750 strongly suggests that the Maf proteins not only have dral ossi®cation at the epiphyseal plates (Reimold et an important role in hind brain morphogenesis in early al., 1996). Since distribution of ATF-2 is similar to that development, but also in the dierentiation processes of Mafs, and ATF-2 can heterodimerize with Maf-1 of the cartilage, lenses and spinal cord. However, in Kr and Maf-2 (manuscript in preparation), it is possible mouse (i.e., a mutant strain of mafB), no abnormalities that ATF-2/Maf(s) heterodimers function in chondro- of cartilage, lenses or spinal cord formation were cyte dierentiation. reported, except for a hyoid greater horn abnormality Further analyses on the stage speci®c expression of (Frohman et al., 1993). It is possible that a functional maf-1 and maf-2 in the cartilage, lens and spinal cord, redundancy of MafB may be responsible for those in conjunction with the identi®cation of speci®c genes results as, in this study, both maf-1 and maf-2 were that are targets for transcriptional regulation by this shown to have similar, but not exact, expression protein, should prove valuable in delineating the patterns. molecular pathways underlying the regulation of In a preliminary study, Maf-1 could heterodimerized transcription during development in these tissues. with Fos family proteins (c-Fos, FosB, Fra1 and Fra2) while Maf-2 could heterodimerized only c-Fos. These heterodimers have signi®cantly dierent DNA binding speci®cities. It is quite possible that the many dierent Acknowledgements heterodimers that can be formed are useful in enabling We thank Dr T Fujita, Tokyo Metropolitan Institute of Gerontology, for kindly providing the rat liver cDNA a wide range of regulation of target genes with various library and Dr John F Maune for critical reading of the heterodimers having varying DNA binding speci®cities manuscript. This work was supported by grants from the and heterodimer protein components having varying Ministry of Education, Science and Culture, Japan. expression patterns. It was recently shown that lack of EMBL/Genbank accession No. for maf-1 and maf-2 are the ATF-2 gene causes great abnormality in endochon- maf-1: U56241 and maf-2: U56242.

References

Andrews NC, Kotkow KJ, Ney PA, Erdjument BH, Tempst Liu Q, Ji X, Breitman ML, Hitchcock PF and Swaroop A. P and Orkin SH. (1993). Proc. Natl. Acad. Sci. USA, 90, (1996). Oncogene, 12, 207 ± 211. 11488 ± 11492. Nishizawa M, Kataoka K, Goto N, Fujiwara KT and Kawai Cordes SP and Barsh GS. (1994). Cell, 79, 1025 ± 1034. S. (1989). Proc.Natl.Acad.Sci.USA,86, 7711 ± 7715. Frohman MA, Martin GR, Cordes SP, Halamek LP and Reimold AM, Grusby MJ, Kosaras B, Fries JW, Mori R, Barsh GS. (1993). Development, 117, 925 ± 936. Maniwa S, Clauss IM, Collins T, Sidman RL, Glimcher Fujiwara KT, Kataoka K and Nishizawa M. (1993). MJ and Glimcher LH. (1996). Nature, 379, 262 ± 265. Oncogene, 8, 2371 ± 2380. Sakai M, Matsushima HY, Nishizawa M and Nishi S. (1995). Gunning P, Leavitt J, Muscat G, Ng SY and Kedes L. (1987). Cancer Res., 55, 5370 ± 5376. Proc. Natl. Acad. Sci. USA, 84, 4831 ± 4835. Sakai M, Muramatsu M and Nishi S. (1992). Biochem. Igarashi K, Kataoka K, Itoh K, Hayashi N, Nishizawa M Biophys. Res. Commun., 187, 976 ± 983. and Yamamoto M. (1994). Nature, 367, 568 ± 572. Senba E, Shiosaka S, Hara Y, Inagaki S, Sakanaka M, Kataoka K, Fujiwara KT, Noda M and Nishizawa M. Takatsuki K, Kawai Y and Tohyama M. (1982). J. Comp. (1994a). Mol. Cell. Biol., 14, 7581 ± 7591. Neurol., 208, 54 ± 66. Kataoka K, Noda M and Nishizawa M. (1994b). Mol. Cell. Sieweke MH, Tekotte H, Frampton J and Graf T. (1996). Biol., 14, 700 ± 712. Cell, 85, 49 ± 60. Kataoka K, Igarashi K, Itoh K, Fujiwara KT, Noda M, Swaroop A, Xu JZ, Pawar H, Jackson A, Skolnick C and Yamamoto M and Nishizawa M. (1995). Mol. Cell. Biol., Agarwal N. (1992). Proc. Natl. Acad. Sci. USA, 89, 266 ± 15, 2180 ± 2190. 270. Kataoka K, Noda M and Nishizawa M. (1996). Oncogene, Urist MR, Huo YK, Brownell AG, Hohl WM, Buyske J, 12, 53 ± 62. Lietze A, Tempst P, Hunkapiller M and DeLange RJ. Kerppola TK and Curran T. (1994). Oncogene, 9, 675 ± (1984). Proc. Natl. Acad. Sci. USA, 81, 371 ± 375. 684. Kuboki Y, Saito T, Murata M, Takita H, Mizuno M, Inoue M, Nagai N and Poole AR. (1995). Connective Tissue Res., 32, 219 ± 226.