Mitf and Tfe3, Two Members of the Mitf-Tfe Family of Bhlh-Zip Transcription Factors, Have Important but Functionally Redundant Roles in Osteoclast Development

Mitf and Tfe3, two members of the Mitf-Tfe family of bHLH-Zip transcription factors, have important but functionally redundant roles in osteoclast development

Eiri´kur Steingri´msson*†, Lino Tessarollo‡, Bhavani Pathak§¶, Ling Houʈ**, Heinz Arnheiterʈ, Neal G. Copeland§, and Nancy A. Jenkins§

*Department of Biochemistry and Molecular Biology, Faculty of Medicine, University of Iceland, 101 Reykjavik, Iceland; ‡Neural Development Group and §Mouse Cancer Genetics Program, National Cancer Institute, Frederick, MD 21702-1201; and ʈLaboratory of Developmental Neurogenetics, National Institute of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, MD 20892

Communicated by Liane B. Russell, Oak Ridge National Laboratory, Oak Ridge, TN, February 6, 2002 (received for review December 1, 2001) The Mitf-Tfe family of basic helix–loop–helix-leucine zipper (bHLH- neuroepithelial-derived retinal pigment epithelium of the eye, Zip) transcription factors encodes four family members: Mitf, Tfe3, ultimately resulting in microphthalmia, and some Mitf mutations Tfeb, and Tfec. In vitro, each protein in the family can bind DNA as result in a reduction in mast cell numbers. Thus, the Mitf a homo- or heterodimer with other family members. Mutational mutations are pleiotropic, affect many different cell types, and studies in mice have shown that Mitf is essential for melanocyte can be arranged in an allelic series (4). and eye development, whereas Tfeb is required for placental A mutation has been made in the Tfeb gene (TfebFcr) and vascularization. Here, we uncover a role for Tfe3 in osteoclast results in embryonic lethality because of defects in placental development, a role that is functionally redundant with Mitf. vascularization; the labyrinthine cells fail to express vascular Although osteoclasts seem normal in Mitf or Tfe3 null mice, the endothelial growth factor, the embryonic vasculature is unable combined loss of the two genes results in severe osteopetrosis. We to invade the placenta, and the embryos degenerate as a result also show that Tfec mutant mice are phenotypically normal, and of hypoxia (5). Unfortunately, the embryos die before melano- that the Tfec mutation does not alter the phenotype of Mitf, Tfeb, cytes or mast cells are formed thus it is unclear whether Tfeb is or Tfe3 mutant mice. Surprisingly, our studies failed to identify any important for the development of these cell types as well. phenotypic overlap between the different Mitf–Tfe mutations. Molecular analysis of the semidominant Mitf mutations shows These results suggest that heterodimeric interactions are not that they affect the basic or transcriptional activation domains of essential for Mitf-Tfe function in contrast to other bHLH-Zip fam- ͞ ͞ the protein (6–8). These mutant proteins cannot bind DNA; ilies like Myc Max Mad, where heterodimeric interactions seem to however, they can still dimerize with proteins such as Tfe3 and be essential. thereby interfere with DNA binding of the wild-type partner (9). The dominant negative behavior of these mutant proteins likely asic helix–loop–helix-leucine zipper (bHLH-Zip) proteins accounts for the phenotype seen in heterozygous mice. Consis- Bregulate gene expression by binding to the E-box tent with this hypothesis, the recessive Mitf mutations affect (CANNTG) as hetero- or homodimers; some dimers activate the dimerization domain of the protein or lack Mift expression gene expression whereas others repress it. The prototypic (6, 7, 10). ͞ ͞ bHLH-Zip family is Myc Max Mad. Work from many labora- Interestingly, a few of the strong semidominant Mitf mutations GENETICS tories suggests that these proteins function through a complex also produce osteopetrosis because of osteoclast defects, which network of interacting proteins (1). The ubiquitous Max protein is not seen in Mitf mutant mice carrying loss of function is at the heart of this network. When Myc concentration is high, mutations at the locus. This phenotype is limited to homozygous it complexes with Max, resulting in increased gene expression animals indicating that the mutations are recessive with respect with concomitant effects on the cell. In contrast, when the to the osteoclasts. Bone marrow transplantation experiments concentration of a Mad family member is high, Max complexes performed on the original Mitf microphthalmia (Mitf mi) mutation with Mad, resulting in reduced gene expression. Mad-mediated (one of the semidominant Mitf mutations that produces osteo- repression results from the interaction of Mad proteins with petrosis) showed that osteopetrosis can be rescued by trans- mSin3 proteins, which in turn interact with the histone deacety- planting cell suspensions from wild-type spleen and bone mar- lases to moderate gene expression (2). The emerging complexity row to the mutant animals (11). Similarly, transplantation of of this network suggests that the specificity of this protein family bone marrow from Mitf mi mice to lethally irradiated wild-type is because of protein–protein interactions as well as different mice induces osteopetrosis (12). Ultrastructural analysis shows target gene specificity (3). It has been difficult, however, to mi confirm these results in vivo, as a result of the complexity of this that osteoclasts from Mitf mice are smaller than normal, lack family and that many of the mutations are embryonic-lethal. ruffled borders (13), and exhibit fusion disability (14). These Another well studied bHLH-Zip protein family is Mitf-Tfe. strong dominant negative Mitf alleles therefore seem to induce Four Mitf-Tfe family members have been identified: microph-

thalmia (Mitf), Tfe3, Tfeb, and Tfec. Mutations in Mitf were Abbreviation: bHLH, basic helix–loop–helix. recognized as early as 1942 and since then over 20 spontaneous †To whom reprint requests should be addressed. E-mail: [email protected]. or induced mutations have been identified at the locus (4). ¶Present address: American Association for the Advancement of Science, 1200 New York Interestingly, about half of the mutations are semidominantly Avenue, NW, Washington, DC 20005. inherited (i.e., they show a partial phenotype in the heterozygous **Present address: Genetic Diseases Research Branch, National Human Genome Research condition) and about half are recessively inherited. All Mitf Institute, National Institutes of Health, Bethesda, MD 20892. mutations result in defects in neural crest-derived melanocytes, The publication costs of this article were defrayed in part by page charge payment. This manifested by reduction or lack of pigment in the coat and inner article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. ear. Many of the mutations also affect differentiation of the §1734 solely to indicate this fact.

www.pnas.org͞cgi͞doi͞10.1073͞pnas.072071099 PNAS ͉ April 2, 2002 ͉ vol. 99 ͉ no. 7 ͉ 4477–4482 Downloaded by guest on September 29, 2021 osteopetrosis by affecting osteoclast differentiation; the defect is Mitf mi-vga9͞Mitf mi-vga9 mice were maintained and propagated at cell-autonomous with respect to osteoclasts. the National Institutes of Health. Several observations suggest that Mitf-Tfe proteins mediate their effects as heterodimers, like Myc͞Max͞Mad. For example, Results the Mitf-Tfe family members are coexpressed in several cell Tfe3 and Tfec Mutations. To begin to analyze the functions of Tfe3 types, including cells affected by Mitf mutations such as oste- and Tfec in vivo and determine whether any of these functions oclasts (15) and melanocytes (16). Furthermore, all possible overlap with Mitf and Tfeb, we made germ-line null mutations in combinations of Mitf-Tfe homo- and heterodimers can form in both genes by means of gene targeting in mouse embryonic stem vitro (9), and Mitf-Tfe heterodimers have been shown to exist in cells. Part, or all, of the bHLH-Zip domain of each protein was cell cultures (15). Taken together, these data suggest that the replaced with pGKNeo (Fig. 1 A and D), and the mutations, absence of any one protein would produce a phenotype that is designated Tfe3Fcr and TfecFcr, respectively, were introduced into at least partially overlapping with one or more of the other family the mouse germ line by standard techniques (Fig. 1 B and E). As members. Here, we produce germ-line null mutations in Tfe3 expected, the Tfec message expressed in homozygous TfecFcr and Tfec and show that the osteopetrosis associated with the mice is missing 132 bp of the Tfec bHLH-Zip domain and dominant negative Mitf alleles is most likely a result of dominant contains a frameshift mutation, which deletes the remainder of negative interference with Tfe3. Surprisingly, none of the Mitf- the protein (Fig. 1F). In contrast, the Tfe3Fcr mutation destabi- Tfe phenotypes studied overlap, suggesting that the homodimer, lizes the Tfe3Fcr message, and no stable mRNA is made in not the heterodimer, is the essential functional unit of the homozygous Tfe3Fcr mice (Fig. 1C). Surprisingly, homozygous Mitf-Tfe family. Tfe3Fcr and TfecFcr mice are indistinguishable from their wild- type littermates; they are viable and fertile, normally pigmented, Materials and Methods have normal eyes and mast cells, and show no osteopetrosis (Fig. Targeted Disruption of Tfe3 and Tfec. Tfe3 and Tfec genomic clones 2F, Table 1, and data not shown). Furthermore, detailed analysis were isolated from a 129͞Sv mouse library (Stratagene) by using of homozygous Tfe3Fcr mice (C. Tunyaplin and K. Calame, human Tfe3 and rat Tfec cDNA probes, respectively (17, 18). The unpublished observations) shows that B-cell development is X-chromosome origin (19) of the most strongly hybridizing Tfe3 normal in contrast to what was observed in Tfe3:Rag2 chimeric clone (clone Tfe3-1) was confirmed by interspecific backcross mice (25). mapping. Similarly, the chromosomal origin of the most strongly hybridizing Tfec clone (clone Tfec4-1) was confirmed by map- Double-Mutant Combinations. To examine functional redundancy ping. To generate the Tfe3 targeting vector, two NotI fragments among the Mitf-Tfe family members, we generated mice that from clone Tfe3-1 were subcloned into pBluescript. A 5.5-kb were double homozygous for all possible combinations of Mitf ClaI-NotI fragment, containing sequences located upstream of (three alleles, described in Table 2), TfecFcr, and Tfe3Fcr, as well the known Tfe3 coding region, was excised and subcloned. Next, as mice that were triple homozygous for Mitf Mi-white (Mitf Mi-wh), a 1.8-kb fragment from the end of the gene was generated by TfecFcr, and Tfe3Fcr. In addition, we generated mice that were PCR, using primers P1 (5Ј-gcaacgctccaaagacctggggatccg- simultaneously heterozygous for TfebFcr and homozygous for gcagcgg) and P2 (5Ј-ccagtgccaggactagttggactggcaattccc) and Mitf Mi-wh, TfecFcr,orTfe3Fcr (the TfebFcr mutation was kept fused with the 5.5-kb clone in the correct orientation. A PgkNeo heterozygous because the TfebFcr mutation is lethal when cassette (20) was inserted between these fragments in the homozygous). The results are summarized in Table 1. Except for opposite transcriptional orientation, and a tk gene was placed at the double- and triple-mutant combinations involving Mitf and the 5Ј end of the targeting construct. To generate the Tfec Tfe3, no functional redundancy was observed. For example, targeting construct, a 1.4-kb EcoRI-EcoRV fragment from clone double homozygous Tfe3Fcr;TfecFcr mice are phenotypically nor- Tfec4-1 was cloned into pBluescript. A 6.0-kb EcoRV-SacI mal, double homozygous Mitf Mi-wh;TfecFcr mice are identical to fragment containing the 3Ј end of the gene was ligated onto the homozygous Mitf Mi-wh mice, homozygous Mitf mi;TfecFcr or 1.4-kb clone in the correct orientation, and a PgkNeo cassette Mitf mi;Tfe3Fcr mice are identical to Mitf mi mice, and triple was inserted between these fragments in the same transcriptional homozygous Mitf Mi-wh;TfecFcr;Tfe3Fc mice are identical to Mitf mi orientation. Finally, the tk gene was placed at the 5Ј end of the mice (Table 1). Similarly, no genetic interactions were observed construct. The mutant constructs were linearized with NotI and between TfebFcr in the heterozygous condition and Tfe3Fcr, electroporated into CJ7 embryonic stem (ES) cells as described TfecFcr, and Mitf Mi-wh in the homozygous condition (Table 1). In (21). DNAs derived from G418͞FIAU-resistant ES clones were contrast, mice homozygous for Tfe3Fcr and Mitf mi-vga9 (an Mitf identified by hybridization. For Tfe3, recombinant clones con- null allele) developed severe osteopetrosis, a phenotype not taining the predicted rearranged bands were obtained at a produced by loss of function mutations in either gene alone frequency of 2͞179 or 1.1%. For Tfec, a single positive clone was (compare Fig. 2 J to E and F; Table 1). In the double mutants, obtained from 300 clones analyzed and its structure confirmed the bone marrow cavity is completely filled with bony material by using an internal probe. The Tfe3 and Tfec mutant clones were (Fig. 2 J), tooth eruption is delayed, and the animals die within injected into C57BL͞6J blastocysts, and the resulting chimeras 3 weeks of birth. were identified by agouti coat color. Chimeras that transmitted the Tfe3 and Tfec mutations through the germ line were mated Mitf-Associated Osteopetrosis. A number of the semidominant to both 129͞Sv and C57BL͞6J females to establish lines. For Mitf mutations, including Mitf mi, Mitf mi-eyeless white (Mitf mi-ew), and genotyping purposes, DNA was isolated from ES cells and from Mitf Mi-oak ridge (Mitf Mi-or), induce osteopetrosis in the homozy- mouse tails as described (22, 23) and processed for Southern gous condition (Table 1). In each case, these mutations produce analysis (24). The genotypes of Mitf Mi-wh and Mitf mi-ew animals dominant negative proteins, which can interfere with the func- were ascertained by PCR analysis. tion of other Mitf-Tfe proteins (9). Thus, it has been proposed that the osteopetrosis observed is because of dominant negative Mice. The mutant strains C57BL͞6J- Tfe3Fcr, C57BL͞6J- TfecFcr, interference with one or more of the Mitf-Tfe proteins in NAW-Mitf mi-ew͞Mitf mi-ew, C57BL͞6J-Mitf Mi-wh͞Mitf Mi-wh, and osteoclasts. The most likely target for down-regulation is Tfe3, C57BL͞6J-Mitf mi͞Mitf mi were maintained and propagated in because Tfe3 and Mitf are both expressed in osteoclasts where the Mammalian Genetics Laboratory (Advanced BioScience they are known to form stable heterodimers (15). Our results Laboratories–Basic Research Program, Frederick, MD), and the confirm this speculation and support the hypothesis that osteo-

4478 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.072071099 Steingrı´msson et al. Downloaded by guest on September 29, 2021 Fig. 1. Generation of Tfe3Fcr and TfecFcr mutant mice. (A) The Tfe3 knockout. The exon-intron organization of Tfe3 is shown at the top with exons indicated as black boxes. A 5.5-kb genomic fragment containing seven Tfe3 exons, the 5Ј end of Tfe3, and the bHLH-Zip domains was replaced with PgkNeo. The first exon corresponds to nucleotides 14–153 of mouse Tfe3 (19). (B) Identification of Tfe3Fcr mutant animals. Tail DNA from the progeny of a heterozygous Tfe3Fcr intercross was digested with BamHI and analyzed by Southern analysis with probe B, which detects 9.7- (wild-type) and 7.3-kb (mutant) alleles. Tfe3Fcr males only transmit the mutant allele to their daughters, confirming the X chromosome linkage of Tfe3 (19). (C) Tfe3 expression in mutant animals. Kidney RNA from Tfe3Fcr and wild-type males was reversed-transcribed and PCR-amplified by using Tfe3 primers 5Ј-ccaagctggcttcccaggctctcac and 5Ј-gttaatgttgaatcgcctgcgtcg. The wild-type 463-bp fragment corresponds to nucleotides 245–708 of Tfe3 (19). The same samples were also amplified with Tfeb primers 5Ј-ccctgtctagcagccacctgaacg and 5Ј-gccgctccttggccagggctctgctc as a control. Primer pairs spanned exon-intron boundaries to eliminate false positives. (D) The Tfec knockout. A 5-kb fragment containing two Tfec bHLH exons was replaced with PgkNeo. Deleted exons correspond to nucleotides 507–639 of rat Tfec (18). (E) Identification of TfecFcr mutant animals. Tail DNA from the progeny of a heterozygous mutant intercross was digested with HindIII and analyzed by Southern analysis with probe A, which detects 9.0- (wild-type) and 2.4-kb (mutant) alleles. (F) Sequence of the TfecFcr mutant transcript. Kidney RNA from wild-type and mutant animals was reverse-transcribed and PCR-amplified with Tfec primers 5Ј-cagtgatgctggctgtgc and 5Ј-gtagccacttgatgtagtcc. Sequencing of the mutant transcript showed that it lacked two conserved Tfec functional domains (indicated by brackets in the wild-type sequence) and is out of frame for the rest of the coding region. B, BamHI; C, ClaI; N, H, HindIII; RI, EcoRI; N, NotI; RV, EcoRV; S, SacI.

petrosis results from mutations in Mitf and the associated most severe osteopetrosis, with extensive accumulation of un-

down-regulation of Tfe3 activity. resorbed endochondral bone and no bone marrow cavity, as seen GENETICS Among the three semidominant mutations, Mitf mi has the in sections through femur (Fig. 2B), tibia, ribs, and head bone

Fig. 2. Osteopetrosis in Mitf and Tfe3Fcr mutant animals. (A–J) Sections through femurs of wild-type and mutant mice showing different levels of osteopetrosis. The genotypes of the different animals are shown at the top of each image. Brackets indicate the extent of osteopetrosis. gp, growth plate; bm, bone marrow.

Steingrı´msson et al. PNAS ͉ April 2, 2002 ͉ vol. 99 ͉ no. 7 ͉ 4479 Downloaded by guest on September 29, 2021 Table 1. A summary of phenotypes of Mitf and Tfe mutant mice Phenotype

Genotype Eyes Pigment Mast cells Osteoclasts Placenta

Single mutants Tfe3Fcr͞Tfe3Fcr —— — — — TfecFcr͞TfecFcr —— — — — TfebFcr͞TfebFcr ND ND ND ND ϩϩϩϩ Mitf mi-vga9͞Mitf mi-vga9 ϩϩϩϩ ϩϩϩϩ ϩ —— Mitf mi͞Mitf mi ϩϩϩϩ ϩϩϩϩ ϩ ϩϩϩϩ — Mitf Mi-wh͞Mitf Mi-wh ϩϩ ϩϩϩϩ ϩ —— Mitf mi-ew͞Mitf mi-ew ϩϩϩϩ ϩϩϩϩ ϩ ϩ — Double and triple mutants Mitf Mi-wh͞Mitf Mi-wh; TfecFcr͞TfecFcr ϩϩ ϩϩϩϩ ϩ —— Mitf Mi-wh͞Mitf Mi-wh; Tfe3Fcr͞Tfe3Fcr ϩϩ ϩϩϩϩ ϩ ϩϩϩ — Mitf mi͞Mitf mi; TfecFcr͞TfecFcr ϩϩϩϩ ϩϩϩϩ ϩ ϩϩϩϩ — Mitf mi͞Mitf mi; Tfe3Fcr͞Tfe3Fcr ϩϩϩϩ ϩϩϩϩ ϩ ϩϩϩϩ — Tfe3Fcr͞Tfe3Fcr; TfecFcr͞TfecFcr —— — — — Mitf mi-ew͞Mitf mi-ew; Tfe3Fcr͞ϩ ϩϩϩϩ ϩϩϩϩ ϩ ϩϩϩ — Mitf mi-ew͞Mitf mi-ew; Tfe3Fcr͞Tfe3Fcr ϩϩϩϩ ϩϩϩϩ ϩ ϩϩϩϩ — Mitf mi-vga9͞Mitf mi-vga9; Tfe3Fcr͞Tfe3Fcr ϩϩϩϩ ϩϩϩϩ ϩ ϩϩϩϩ — Mitf Mi-wh͞Mitf Mi-wh; TfecFcr͞TfecFcr, Tfe3Fcr͞Tfe3Fcr ϩϩ ϩϩϩϩ ϩ ϩϩϩ — Tfeb double mutants Tfe3Fcr͞Tfe3Fcr; TfebFcr͞ϩ —— — — — TfecFcr͞TfecFcr; TfebFcr͞ϩ —— — — — Mitf Mi-wh͞Mitf Mi-wh; TfebFcr͞ϩ ϩϩ ϩϩϩϩ ϩ ——

—, no change; ϩ, affected; ϩϩϩϩ, severely affected. ND, not determined because of embryonic lethality.

such as maxillae of homozygous 21-day-old mice (not shown). mal with respect to bone development (8). Although both of Homozygotes are half-normal size and do not live past 3 weeks these mutations behave in a dominant negative fashion in vitro, of age. Animals carrying the Mitf mi-ew mutation show the mildest they still retain DNA-binding activity under certain conditions bone phenotype. Homozygotes are white and severely microph- (9). This behavior is in sharp contrast to the Mitfmi, Mitfmi-ew, and thalmic yet show no signs of growth retardation at 3 weeks of age. MitfMi-or proteins, which cannot bind DNA under any conditions. However, detailed histopathological analysis of bones from 21-day-old Mitf mi-ew͞Mitf mi-ew mice shows clear signs of hyper- Mitf and Tfe3 Interaction Is Allele-Specific. To determine whether osteosis, with extensions of bony trabeculae that reach further the Mitf–Tfe3 interaction is allele-specific, we crossed Tfe3Fcr into the bone marrow cavity than in wild type (Fig. 2C). mice to Mitf mi-ew and Mitf Mi-wh mice. In contrast to the single This phenotype resembles the hyperosteosis associated with homozygotes, Mitf mi-ew;Tfe3Fcr double-mutant mice show severe Camurati–Engelmann disease (26). Although the Mitfmi-ew osteopetrosis (Fig. 2H, Table 1). The double-mutant animals are protein acts in a strong dominant negative fashion in vitro (9), smaller than their single-mutant littermates [Mitf mi-ew;Tfe3Fcr only a small fraction of the mutant protein translocates to the 6.1 Ϯ 0.8g (n ϭ 6); Mitf mi-ew 10.4 Ϯ 1.3g (n ϭ 9); Tfe3Fcr 11.3 Ϯ nucleus (27). Thus, the mild phenotype is consistent with 1.3g (n ϭ 8)], tooth eruption is delayed, and the animals die by dominant negative action of the few nuclear Mitfmi-ew proteins. 3 weeks of age (data not shown). This effect varies with re- The osteopetrosis observed in the Mitf Mi-or mutation is more spect to the Tfe3 concentration as the osteopetrosis in Mitf mi-ew͞ severe than that observed in Mitf mi-ew homozygotes but less Mitf mi-ew,Tfe3Fcr͞ϩ animals is more severe than in Mitf mi-ew severe than in Mitfmi animals (28). Thus, the three Mitf alleles homozygotes but less severe than in the double homozygotes form an allelic series with respect to osteopetrosis. (Table 1). No such concentration-dependence is observed in Osteopetrosis is not observed in all dominant negative Mitf Mitf mi-vga9͞Mitf mi-vga9,Tfe3Fcr͞ϩ animals, suggesting that this mutations. For example, the Mitf Mi-wh mutation has normal bone effect is caused by dominant negative action of the Mitfmi-ew development in the homozygous condition (Fig. 2D). Similarly, protein against Tfe3. the dominant negative Mitf Mi-brownish (Mitf Mi-b) mutation is nor- Although the coat and eye phenotype associated with the

Table 2. Mitf alleles used in this study Phenotype

Allele symbol Heterozygote Homozygote Molecular defect

Mitf mi Iris pigment less than in wild type; White coat; microphthalmia; incisors fail to Deletion of R216; basic domain occasional spots on belly, head, or tail erupt; osteopetrosis Mitf Mi-or Slight dilution of coat color; pale ears and White coat; microphthalmia; incisors fail to R216K; basic domain tail; belly streak or head spot erupt; osteopetrosis Mitf Mi-wh Dilute coat color; eyes dark ruby; white White coat; eyes small and slightly I212N; basic domain spots on feet, tail, and belly pigmented Mitf mi-ew Normal White coat; microphthalmia; hyperosteosis 25 amino acid deletion; basic domain Mitf mi-vga9 Normal White coat; microphthalmia Insertion and deletion; regulatory

4480 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.072071099 Steingrı´msson et al. Downloaded by guest on September 29, 2021 Discussion Here we describe the generation of knockout mutations in Tfe3 and Tfec, two members of the Mitf-Tfe family of bHLH-Zip transcription factors. We report the phenotype of these mice in the heterozygous and homozygous condition and when combined with each other and with mutations in Mitf and Tfeb,two other family members. Our studies show that Tfec is not essential for mammalian development; TfecFcr mutant mice are phenotypically normal and the TfecFcr mutation does not exacerbate the phenotype of mutations in genes of the other family members. In contrast, we uncover an important but functionally redundant role for Tfe3 and Mitf in osteoclasts, the cells required for bone remodeling. Although osteoclasts appear normal in Mitf and Tfe3Fcr null mice, their combined loss results in severe osteopetrosis. These results explain previous genetic data showing that some strong dominant negative alleles of Mitf induce osteopetrosis on their own. Our results indicate that dominant negative mutations of Mitf down-regulate Tfe3 activity and that the combined loss of Mitf and Tfe3 results in osteopetrosis. Two human deafness and pigmentation disorders, Waarden- Fig. 3. Osteoclast defects in Mitf and Tfe3Fcr mutant animals. (A–H) Sections burg Syndrome type 2A (WS2A) (30) and Tietz Syndrome (31), through femurs of wild-type and mutant mice showing different levels of are caused by mutations in MITF. Individuals with Tietz syn- osteoclast activity as determined by tartrate-resistant alkaline phosphatase drome have profound congenital deafness and generalized hy- staining. Arrows point to osteoclasts. popigmentation, whereas individuals with WS2A have milder manifestations of deafness and pigmentation disturbances. Both diseases are dominant. To date, osteopetrosis has not been Mitf Mi-wh mutation is unchanged in Mitf Mi-wh,Tfe3Fcr double observed in any individuals with WS2A or Tietz. However, mutants, the mice are smaller and more runted than their osteopetrosis is seen only in a few mouse Mitf mutations that single-mutant littermates [5.6 Ϯ 1.3g (n ϭ 17) vs. 10.0 Ϯ 1.2g encode strong dominant negative proteins and only when they (n ϭ 8)]. Although most of these mice die at 3–4 weeks of age, are homozygous. It thus remains possible that individuals with some survive to adulthood. In all cases, bones show clear signs WS2A or Tietz will be identified who suffer from osteopetrosis. of osteopetrosis with intermediate extensions of bony trabeculae Mitf mutations have also been observed in rats (15, 32), (Fig. 2I), similar to that seen in Mitf Mi-or mice (28). The residual hamsters (33), quail (34), and zebrafish (35). Molecular analyses activity of the MitfMi-wh protein is probably responsible for the of these mutations show that they are loss of function mutations fact that this osteopetrosis is less severe than that seen in the rather than dominant negative mutations. Despite this, osteo- other double mutants analyzed. In all cases, the effects of petrosis has been reported as a phenotype associated with the the Mitf;Tfe3 mutant combinations are specific to osteoclasts, quail silver mutation and the rat Mitf mib mutation (36, 37). In because in no case do they alter the mast cell, eye, or pigmen- both cases the osteopetrosis is mild. For example, newborn rats tation defects of Mitf mice (Table 1). show sclerosis that is reduced significantly during the first Unlike Tfe3Fcr, TfecFcr does not induce osteopetrosis when months, with bone marrow cavities shaping to near-wild-type mib combined with Mitf Mi-wh or with any of the other mutations levels. The Mitf mutation is caused by a deletion, but the full tested (Table 1), nor does it exacerbate the osteopetrosis ob- extent of this deletion is unknown. It is possible that this mild GENETICS served in Mitf Mi-wh;Tfe3Fcr;TfecFcr triple-mutant mice (Table 1). osteopetrosis is caused by the removal of an as yet unidentified Although the embryonic lethality of TfebFcr precludes analysis of flanking gene that is important for osteoclast function. It is also osteoclasts in homozygotes, Mitf Mi-wh͞Mitf Mi-wh,TfebFcr͞ϩ ani- possible this phenotype reflects species differences in Mitf mals do not develop osteopetrosis, and reducing the Tfeb gene function in rats or genetic background. In this light, it will be of Tfe3 dosage does not change the appearance of these mice in any interest to investigate expression, function, and sequence in normal and Mitf mib rats. The quail mutation is unusual because other way (Table 1). Furthermore, explant culture and chimera it carries two alterations in the gene, a histidine to arginine analysis strongly indicate that Tfeb is not necessary for melano- change in the basic domain, and a stop codon shortly after the cyte development (data not shown). zipper domain. Although the transactivation potential of this mutation is somewhat reduced as compared with wild type (34), Effects on Osteoclasts. Staining for osteoclast activity with tar- mi its dominant negative activity has not yet been determined. trate-resistant alkaline phosphatase (TRAP) shows that Mitf Mutations in cathepsin K (Ctsk) also produce an osteopetrotic homozygotes (Fig. 3B) have smaller osteoclasts than either Fcr mi-ew phenotype. Ctsk-deficient mice survive and are fertile but display wild-type (Fig. 3A) (13), Tfe3 (Fig. 3E), Mitf (Fig. 3C), or Ctsk mi-vga9 excessive trabeculation of the bone marrow space (38). - Mitf (Fig. 3D) mice. Consistent with their bone morphol- deficient osteoclasts manifest a modified ultrastructural appear- Fcr mi ogy, animals double homozygous for Tfe3 and Mitf (Fig. 3F) ance: their resorptive surface is poorly defined with a broad mi-ew mi-vga9 or Mitf (Fig. 3G)orMitf (Fig. 3H) have smaller demineralized matrix fringe containing undigested fine collage mi osteoclasts resembling those of Mitf (Fig. 3B) homozygotes. fibrils, their ruffled borders lack crystal-like inclusions, and they The different effects of the various Mitf alleles on osteoclast are devoid of collagen-fibril-containing cytoplasmic vacuoles. In function are likely the results of different effects on target gene humans, CTSK deficiency induces pycnodysostosis (39), a rare expression. Although osteoclasts are smaller in the mutants, inherited osteochondrodysplasia that is characterized by osteo- TRAP activity can still be detected (Fig. 3) indicating that, sclerosis, short stature, and acroosteolysis of the distal phalan- although the Mitf and Tfe3 proteins may regulate the expression ges. Interestingly, in vitro expression studies show that Ctsk of the Trap gene in vitro (29), their role can be substituted for by expression can be up-regulated by Mitf and Tfe3 (40). Mitf and other factors in vivo. Tfe3 may therefore be important activators of Ctsk expression in

Steingrı´msson et al. PNAS ͉ April 2, 2002 ͉ vol. 99 ͉ no. 7 ͉ 4481 Downloaded by guest on September 29, 2021 vivo, and loss of Ctsk expression in Mitf;Tfe3Fcr double-mutant Tfe3 and Mitf. Consistent with this explanation, the Mitf targets mice could, in part, explain the osteopetrosis we observe in these Tyrp1 and Tyr can both be stimulated by overexpression of Tfe3 animals. in vitro, whereas Tfeb can stimulate only expression of Tyrp1 (16). Immunohistochemical studies show that carbonic anhydrase In summary, our results indicate that the functional impor- II, another potential target of Mitf, is expressed in osteoclasts of tance of heterodimeric interactions varies dramatically among Mitf mi homozygous mice (41). Electron microscopic studies of bHLH-Zip families. Although a number of the Myc͞Max͞Mad osteoclasts from these mice show close contact with bone and heterodimers are thought to provide critical and essential func- some ability to resorb calcified matrix (14). Furthermore, the tions to this transcription factor network, this does not seem to cells have the capacity to adhere to and spread from surfaces and be true for the Mitf-Tfe family. At least three Mitf-Tfe family form lamellipodia (42). These observations suggest that the Mitf members (Mitf, Tfe3, and Tfec) do not serve as essential and Tfe3 mutations do not eliminate osteoclast function alto- heterodimerization partners, although Mitf-Tfe3 dimers clearly gether. Perhaps Mitf and Tfe3 are required for the expression of form in osteoclasts (15). It remains possible, however, that Tfeb genes that enhance the bone resorption ability of osteoclasts. is an essential dimerization partner for Mitf in mast cells or for Alternatively, an as yet unidentified protein(s) in osteoclasts, a Mitf or Tfe3 in osteoclasts, as the effects of the TfebFcr mutation cell type that seems to have developed genetic redundancy as have been fully explored only in melanocytes. In short, our an important evolutionary theme, could substitute for their results indicate that the outcome of heterodimerization differs function. dramatically among different bHLH-Zip families. A possible explanation for our results lies in the expression patterns of the Mitf-Tfe genes combined with tissue-specific We thank Susan Reid, Jan Flynn, Debby Swing, Bryn Eagleson, Lisa threshold requirements. In melanocytes and labyrinthine cells of Secrest, Joanne Dietz, Fran Dorsey, and Linda Cleveland for expert the placenta, which express high levels of Mitf and Tfeb, respec- technical assistance; Miriam Anver, Becky Oden, and the staff of the tively, only the highly expressed family member may be able to histopathology laboratory (Science Applications International Corpo- meet the threshold requirement. In osteoclasts, Tfe3 and Mitf are ration, Frederick, MD) for help with histology. We also thank Francesca Pignoni, Nick Short, Pamela Schwartzberg, and Lynn Hudson for expressed at similar levels (15), the threshold requirement could comments on the manuscript; and Richard Frederickson for illustrations. be lower, and either protein may be able to replace the other. The This work was supported by the National Cancer Institute, the Depart- sequences of the Mitf-Tfe genes differ outside of their basic and ment of Health and Human Services (N.G.C. and N.A.J.), by a North HLHZip domains, however, and these nonconserved sequences Atlantic Treaty Organization science fellowship (E.S.), and by the may be responsible for the different phenotypes of mutations in Icelandic Research Council (E.S.).

1. Schreiber-Agus, N. & DePinho, R. A. (1998) BioEssays 20, 808–818. 23. Siracusa, L. D., Russell, L. B., Jenkins, N. A. & Copeland, N. G. (1987) Genetics 2. Knoepfler, P. S. & Eisenman, R. N. (1999) Cell 99, 447–450. 117, 85–92. 3. O’Hagan, R. C., Schreiber-Agus, N., Chen, K., David, G., Engelman, J. A., 24. Jenkins, N. A., Copeland, N. G., Taylor, B. A. & Lee, B. K. (1982) J. Virol. 43, Schwab, R., Alland, L., Thomson, C., Ronning, D. R., Sacchettini, J. C., et al. 26–36. (2000) Nat. Genet. 24, 113–119. 25. Merrell, K., Wells, S., Henderson, A., Gorman, A., Alt, F., Stall, A. & Calame, 4. Moore, K. J. (1995) Trends Genet. 11, 442–448. K. (1997) Mol. Cell. Biol. 17, 3335–3344. 5. Steingrı´msson, E., Tessarollo, L., Reid, S. W., Jenkins, N. A. & Copeland, N. G. 26. Makita, Y., Nishimura, G., Ikegawa, S., Ishii, T., Ito, Y. & Okuno, A. (2000) (1998) Development (Cambridge, U.K.) 125, 4607–4616. Am. J. Med. Genet. 91, 153–156. 6. Hodgkinson, C. A., Moore, K. J., Nakayama, A., Steingrı´msson, E., Copeland, 27. Takebayashi, K., Chida, K., Tsukumoto, I., Morii, E., Munakata, H., Arnheiter, N. G., Jenkins, N. A. & Arnheiter, H. (1993) Cell 74, 395–404. H., Kuroki, T., Kitamura, Y. & Nomura, S. (1996) Mol. Cell. Biol. 16, 7. Steingrı´msson, E., Moore, K. J., Lamoreux, M. L., Ferre´-D’Amare´, A. R., 1203–1211. Burley, S. K., Zimring, D. C., Skow, L. C., Hodgkinson, C. A., Arnheiter, H., 28. Nii, A., Steingrı´msson, E., Copeland, N. G., Jenkins, N. A. & Ward, J. M. (1995) Copeland, N. G. & Jenkins, N. A. (1994) Nat. Genet. 8, 256–263. Am. J. Pathol. 147, 1871–1882. 8. Steingrı´msson, E., Nii, A., Fisher, D. E., Ferre´-D’Amare´,A. R., McCormick, 29. Luchin, A., Purdom, G., Murphy, K., Clark, M.-Y., Angel, N., Cassady, A. I., R. J., Russell, L. B., Burley, S. K., Ward, J. M., Jenkins, N. A. & Copeland, N. G. Hume, D. A. & Ostrowski, M. C. (2000) J. Bone Miner. Res. 15, 451–460. 15, (1996) EMBO J. 6280–6289. 30. Read, A. P. & Newton, V. E. (1997) J. Med. Genet. 34, 656–665. 9. Hemesath, T. J., Steingrı´msson, E., McGill, G., Hansen, M. J., Vaught, J., 31. Amiel, J., Watkin, P. M., Tassabehji, M., Read, A. P. & Winter, R. M. (1998) Hodgkinson, C. A., Arnheiter, H., Copeland, N. G., Jenkins, N. A. & Fisher, Clin. Dysmorphol. 7, 17–20. D. E. (1994) Genes Dev. 8, 2770–2780. 32. Opdecamp, K., Vanvooren, P., Riviere, M., Arnheiter, H., Motta, R., Szpirer, 10. Hughes, M. J., Lingrel, J. B., Krakowsky, J. M. & Anderson, K. P. (1993) J. Biol. J. & Szpirer, C. (1998) Mamm. Genome 9, 617–621. Chem. 268, 20687–20690. 33. Hodgkinson, C. A., Nakayama, A., Li, H., Swenson, L. B., Opdecamp, K., 11. Walker, D. G. (1975) Science 190, 784–785. Asher, J. H., Jr., Arnheiter, H. & Glaser, T. (1998) Hum. Mol. Genet. 7, 12. Walker, D. G. (1975) Science 190, 785–786. 703–708. 13. Holtrop, M. E., Cox, K. A., Eilon, G., Simmons, H. A. & Raisz, L. G. (1981) Dev. Biol. 196, Metab. Bone Dis. Rel. Res. 3, 123–129. 34. Mochii, M., Ono, T., Matsubara, Y. & Eguchi, G. (1998) 14. Thesingh, C. W. & Scherft, J. P. (1985) Bone 6, 43–52. 145–159. 15. Weilbaecher, K. N., Hershey, C. L., Takemoto, C. H., Horstmann, M. A., 35. Lister, J. A., Robertson, C. P., Lepage, T., Johnson, S. L. & Raible, D. W. (1999) Hemesath, T. J., Tashjian, A. H. & Fisher, D. E. (1998) J. Exp. Med. 187, Development (Cambridge, U.K.) 126, 3757–3767. 775–785. 36. Cielinski, M. J. & Marks, S. C. (1994) Bone 15, 707–715. 16. Verastegui, C., Bertolotti, C., Bille, K., Abbe P., Ortonne, J. P. & Ballotti, R. 37. Kawaguchi, N., Ono, T., Mochii, M. & Noda, M. (2001) Dev. Dyn. 220, 133–140. (2000) Mol. Endocrinol. 14, 449–456. 38. Saftig, P., Hunziker, E., Wehmyer, O., Jones, S., Boyde, A., Rommerskirch, W., 17. Beckmann, H., Su, L.-K. & Kadesch, T. (1990) Genes Dev. 4, 167–179. Moritz, J. D., Schu, P. & von Figura, K. (1998) Proc. Natl. Acad. Sci. USA 95, 18. Zhao, G.-Q., Zhao, Q., Zhou, X., Mattei, M.-G. & DeCrombrugghe, B. (1993) 13453–13458. Mol. Cell. Biol. 13, 4505–4512. 39. Gelb, B. D., Shi, G. P., Chapman, H. A. & Desnick, R. J. (1996) Science 273, 19. Roman, C., Matera, A. G., Cooper, C., Artandi, S., Blain, S., Ward, D. C. & 1236–1238. Calame, K. (1992) Mol. Cell. Biol. 12, 817–827. 40. Motyckova, G., Weilbaecher, K. N., Horstmann, M., Rieman, D. J., Fisher, 20. Soriano, P., Montgomery, C., Geske, R. & Bradley, A. (1991) Cell 64, 693–702. D. Z. & Fisher, D. E. (2001) Proc. Natl. Acad. Sci. USA 98, 5798–5803. 21. Swiatek, P. J. & Gridley, T. (1993) Genes Dev. 7, 2071–2084. 41. Jilka, R. L., Rogers, J. I., Khalifah, R. G. & Vaananen, H. K. (1985) Bone 6, 22. Laird, P. W., Zijderveld, A., Linders, K., Rudnicki, M. A., Jaenisch, R. & Berns, 445–449. A. (1991) Nucleic Acids Res. 19, 4293. 42. Helfrich, M. H. & Mieremet, R. H. P. (1988) Bone 9, 113–119.

4482 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.072071099 Steingrı´msson et al. Downloaded by guest on September 29, 2021