Gene fusions involving PAX and FOX family members in alveolar rhabdomyosarcoma


The chromosomal translocations t(2;13)(q35;q14) and t(1;13)(p36;q14) are characteristic of alveolar rhabdomyosarcoma, a pediatric soft tissue cancer related to the striated muscle lineage. These translocations rearrange PAX3 and PAX7, members of the paired box transcription factor family, and juxtapose these genes with FKHR, a member of the fork head transcription factor family. This juxtaposition generates PAX3–FKHR and PAX7–FKHR chimeric genes that are expressed as chimeric transcripts that encode chimeric proteins. The fusion proteins, which contain the PAX3/PAX7 DNA binding domain and the FKHR transcriptional activation domain, activate transcription from PAX-binding sites with higher potency than the corresponding wild-type PAX proteins. This increased function results from the insensitivity of the FKHR activation domain to inhibitory effects of N-terminal PAX3/PAX7 domains. In addition to altered function, the fusion products are expressed in ARMS tumors at higher levels than the corresponding wild-type PAX products due to two distinct mechanisms. The PAX7–FKHR fusion is overexpressed as a result of in vivo amplification while the PAX3–FKHR fusion is overexpressed due to a copy number-independent increase in transcriptional rate. Finally, though FKHR subcellular localization is regulated by an AKT-dependent pathway, the fusion proteins are resistant to these signals and show exclusively nuclear localization. Therefore, these translocations alter biological activity at the levels of protein function, gene expression, and subcellular localization with the cumulative outcome postulated to be aberrant regulation of PAX3/PAX7 target genes. This aberrant gene expression program is then hypothesized to contribute to tumorigenic behavior by impacting on the control of growth, apoptosis, differentiation and motility.

Clinical, pathologic and cytogenetic features of alveolar rhabdomyosarcoma

Alveolar rhabdomyosarcoma (ARMS) is one subtype of rhabdomyosarcoma (RMS), a family of pediatric soft tissue tumors that are related to the skeletal muscle lineage (Wexler and Helman, 1997). ARMS was initially distinguished from the other major RMS subtype, embryonal rhabdomyosarcoma (ERMS), by histologic criteria. As the criteria were refined, the subtypes were found to be associated with distinct clinical behaviors (Tsokos et al., 1992). ARMS presents mainly in adolescents and young adults, often in the vicinity of skeletal muscle in the extremities and trunk, and is associated with an unfavorable prognosis. This unfavorable prognosis is related to the propensity for early and wide dissemination, often involving bone marrow, and to poor response to chemotherapy. In contrast, ERMS mainly presents in children less than 10 years old in regions without abundant skeletal muscle such as the head and neck, genitourinary tract, and retroperitoneum, and has a more favorable prognosis.

Diagnosis of ARMS is often complicated by scant evidence of striated muscle differentiation and relatively subtle histologic criteria for distinguishing ERMS and ARMS (Tsokos, 1994). The problem is compounded by the fact that other pediatric solid tumors including neuroblastoma, Ewing's sarcoma and lymphoma can present as collections of poorly differentiated cells. Subtle evidence of myogenic differentiation can be detected with immunohistochemical reagents specific for muscle proteins (such as MyoD, desmin, or muscle-specific actin) and electron microscopic analysis of myofilaments. However, there is no well-established immunohistochemical or ultrastructural marker that distinguishes ARMS and ERMS.

Chromosomal studies identified nonrandom translocations that distinguish the majority of ARMS tumors from ERMS and other pediatric solid tumors. The most prevalent finding in ARMS is a translocation, t(2;13)(q35-37;q14), which was detected in 70% of ARMS cases (Douglass et al., 1987; Turc-Carel et al., 1986; Wang-Wuu et al., 1988) (Figure 1). In addition, a variant translocation, t(1;13)(p36;q14), was identified in a smaller subset of ARMS cases (Biegel et al., 1991; Douglass et al., 1991). The 2;13 and 1;13 translocations have not been detected in any other tumor type and appear to be specific and sensitive markers of ARMS.

Figure 1

Diagram of t(2;13)(q35;q14) and t(1;13)(p36;q14) chromosomal translocations

Mapping and cloning of loci involved in 2;13 and 1;13 translocations

Based on physical mapping studies, PAX3 was localized to the vicinity of the t(2;13) breakpoint on chromosome 2 (Barr et al., 1993). Analysis of somatic cell hybrids derived from ARMS cells indicated that PAX3 is split such that the 5′ PAX3 region is located on the derivative chromosome 13 [der(13)] and the 3′ PAX3 region is on the der(2) (Figure 2). In conjunction with Southern blot findings of PAX3 structural alterations in ARMS cell lines, these studies established that PAX3 is the chromosome 2 locus rearranged by the t(2;13). Northern analysis with a 5′ PAX3 probe demonstrated a novel 7.2 kb transcript unique to cell lines with the t(2;13). Clones containing 5′ PAX3 sequences fused to a novel sequence from chromosomal region 13q14 were then isolated from cDNA libraries constructed from ARMS cell lines (Galili et al., 1993; Shapiro et al., 1993) (Figure 3). The corresponding wild-type chromosome 13 gene generates a 6.5 kb widely expressed transcript that encodes a 655 amino acid protein. A BLAST database search showed homology to the fork head transcription factor family, and thus this chromosome 13 gene was named FKHR (fork head in rhabdomyosarcoma) (Galili et al., 1993); this gene was also termed ALV (Shapiro et al., 1993), and was recently renamed FOXO1 (Kaestner et al., 2000). These findings indicate that the t(2;13) results in a chimeric transcript composed of 5′ PAX3 sequences fused to 3′ FKHR sequences.

Figure 2

Generation of chimeric genes by the t(2;13)(q35;q14) translocation in ARMS. The exons of the wild-type and fusion genes are shown as boxes above each map and the translocation breakpoint distributions are shown as line segments below the map of the wild-type genes

Figure 3

Comparison of wild-type and fusion products associated with the 2;13 and 1;13 translocations. The paired box, octapeptide, homeobox and fork head domain are indicated as open boxes, and transcriptional domains (DNA binding domain, DBD; transcriptional activiation domain, TAD; transcriptional inhibitory domain, TID) are shown as solid bars. The sites phosphorylated by Akt are indicated by stars, and the alternative splice in the paired box is shown by an arrowhead. The vertical dash line indicates the translocation fusion point

Analysis of the PAX3–FKHR cDNA revealed that the 5′ PAX3 and 3′ FKHR coding sequences are fused in-frame to encode an 836 amino acid fusion protein (Galili et al., 1993; Shapiro et al., 1993) (Figure 3). The consistency of this chimeric transcript was confirmed by reverse transcriptase (RT)–PCR experiments, and sequence analysis of these RT–PCR products revealed an invariant fusion point. The reciprocal translocation product, the der(2) chimeric transcript consisting of 5′ FKHR and 3′ PAX3 exons, was not detected in ARMS lines by Northern blot analysis (Barr et al., 1993), but low level expression was detected by RT–PCR analysis in a subset of ARMS cell lines and specimens (Galili et al., 1993). The findings of higher, more consistent expression of the 5′ PAX3-3′ FKHR fusion indicate that the der(13) encodes the product involved in the pathogenesis of ARMS.

In ARMS cases with the 1;13 translocation, attention focused on PAX7, another member of the paired box-containing transcription factor family located in chromosomal region 1p36 (Davis et al., 1994). RT–PCR and Southern blot studies demonstrated that PAX7 is rearranged and fused to FKHR in tumors containing this translocation. This fusion results in a chimeric transcript consisting of 5′ PAX7 and 3′ FKHR regions (Figure 3), which is nearly identical in structure and organization to the 5′ PAX3–3′ FKHR transcript formed by the 2;13 translocation.

Paired box family of transcription factors

PAX3 and PAX7 encode members of the paired box or PAX transcription factor family that is characterized by a conserved paired box DNA binding domain first identified in Drosophila segmentation genes (Tremblay and Gruss, 1994). There are nine human members of this family. Several family members also contain a complete or truncated version of a homeobox DNA binding domain and a short conserved octapeptide motif distal to the paired box. PAX3 and PAX7 constitute a subfamily within the PAX family and encode very similar proteins containing an N-terminal DNA binding domain consisting of a paired box, octapeptide and complete homeodomain, and a proline-, serine- and threonine-rich C-terminal domain (Figure 3).

PAX family members function in the transcriptional control of pattern formation during embryogenesis (Tremblay and Gruss, 1994). Each gene has a unique temporal and spatial expression pattern during early development, and some are also expressed with a restricted distribution in the adult. In situ hybridization analysis of embryos revealed that Pax3 and Pax7 are both expressed in the developing nervous system and in somite compartments that give rise to skeletal muscle progenitors (Goulding et al., 1991; Jostes et al., 1990). Pax7 expression is activated later and persists longer than that of Pax3. In addition, Pax3 becomes localized to the lateral dermomyotome while Pax7 becomes more prominent in the medial dermomyotome, and Pax3 but not Pax7 is expressed by myogenic progenitors that migrate to the limbs (Bober et al., 1994; Goulding et al., 1994; Williams and Ordahl, 1994). Finally, while Pax3 expression was not detected in the adult mouse, Pax7 expression was detected in muscle satellite cells of the adult mouse (Seale et al., 2000). Therefore, PAX3 and PAX7 are expressed with overlapping but distinct patterns in myogenic precursors that are the presumed progenitors of ARMS.

The important developmental role of the paired box genes is further highlighted by phenotypic analyses of germline mutations (Dahl et al., 1997; Tremblay and Gruss, 1994). Mutations of these genes were identified in several spontaneous murine and human heritable developmental disorders. Point mutations and deletions affecting functional domains of the Pax3 gene were identified in the splotch mouse (Epstein et al., 1991), which is characterized by abnormalities of the neural tube, neural crest-derived structures and peripheral musculature. The human disease Waardenburg syndrome, characterized by deafness and pigmentary disturbances, is caused by mutations in the PAX3 gene (Baldwin et al., 1992; Tassabehji et al., 1992). Functional studies showed that splotch and Waardenburg mutations alter or abolish transcriptional activity (Chalepakis et al., 1994a), and result in a loss of function. In addition to these spontaneous mutations, gene targeting technology was used to generate murine strains in which other PAX genes were inactivated (Dahl et al., 1997). Animals that are homozygous for a disrupted Pax7 gene demonstrate abnormalities in neural crest-derived structures and musculature (Mansouri et al., 1996; Seale et al., 2000). These studies thus provide functional evidence of an important role for Pax3 and Pax7 in myogenic development.

Fork head family of transcription factors

The fork head domain was first identified as a 100 amino acid region of sequence similarity between the Drosophila fork head and the rat HNF-3α proteins, and was subsequently identified in a large family of genes in species ranging from yeast to human (Kaufmann and Knochel, 1996). Genetic, expression and functional analyses of the various fork head (or FOX) genes indicate diverse roles including control of embryonic development and adult tissue-specific gene expression. The FOX gene products function as transcription factors in which the fork head domain constitutes a functional unit that is necessary and sufficient for DNA binding, and cannot be further subdivided without loss of the DNA binding function. In addition, other divergent regions of the proteins confer transcriptional regulatory function.

Within the FOX family, FKHR is the prototype of a subfamily with highly similar sequence in the fork head domain and additional regions of sequence similarity outside the fork head domain (Anderson et al., 1998; Borkhardt et al., 1997; Hillion et al., 1997). Other subfamily members include AFX (also called FOXO4) and FKHRL1 (also called FOXO3a). A subfamily member named DAF-16 was also identified in C. elegans (Lin et al., 1997; Ogg et al., 1997). Analysis of adult and embryonic murine tissues demonstrated that the three mammalian subfamily members are widely expressed, with quantitative differences in expression that are somewhat complementary (Furuyama et al., 2000). Following the finding that DAF-16 functions in insulin-like signaling pathways and longevity control in C. elegans (Lin et al., 1997; Ogg et al., 1997), subsequent studies of FKHR, AFX, and FKHRL1 demonstrated roles in insulin-response and growth arrest/apoptosis pathways (Brunet et al., 1999; Guo et al., 1999; Medema et al., 2000). A further striking role of this subfamily in tumorigenesis is indicated by the finding that AFX and FKHRL1 are both fused to the MLL gene by chromosomal translocations in acute leukemias (Borkhardt et al., 1997; Hillion et al., 1997).

Structural analyses of translocation breakpoints

The structure of the chimeric PAX3–FKHR and PAX7–FKHR products is clarified by a consideration of the genomic organization of the wild-type and chimeric genes (Figure 2). In addition to the high degree of sequence similarity between the PAX3 and PAX7 products, the organization of the PAX3 and PAX7 loci is very similar, with highly comparable exon-intron boundaries and exon distributions (Burri et al., 1989; Macina et al., 1995; Vorobyov et al., 1997). Each locus consists of nine exons dispersed over 100 kb; exons 2, 3 and 4 encode the paired box whereas the homeodomain is encoded by exons 5 and 6 and the transactivation domain is encoded by exons 6, 7 and 8 (Bennicelli et al., 1995, 1999). A recently discovered exon 9 in each gene encodes a C-terminal extension with unknown function (Barr et al., 1999). The t(2;13) and t(1;13) breakpoints consistently disrupt the seventh introns, which span 17.5 and 32 kb in the PAX3 and PAX7 loci, respectively; the distribution of breakpoints within these introns appears random (Barr et al., 1998; Fitzgerald et al., 2000). Therefore, these breakpoints are situated to maintain the integrity of the N-terminal DNA binding domain and separate it from an essential part of the transactivation domain.

The FKHR locus consists of three exons spanning 140 kb (Davis et al., 1995) (Figure 2). The fork head domain is encoded by portions of exons 1 and 2 and the transcriptional activation domain is encoded by exon 2; the last exon consists entirely of 3′ untranslated region. The translocation breakpoints occur within the first intron, which spans 130 kb. This intron provides a large target for rearrangements and allows disruption of the fork head DNA binding domain and fusion of the C-terminal FKHR transactivation domain to the N-terminal PAX DNA binding domain. Furthermore, the translational reading frame is maintained from PAX3/PAX7 exon 7 to FKHR exon 2. A similar fusion cannot be created by any other combination of PAX3/PAX7 and FKHR exons, because of incompatible reading frames or loss of functional domains. These findings support the premise that rearrangements of PAX3/PAX7 intron 7 and FKHR intron 2 are selected due to functional constraints related to the genomic organization of these loci.

Expression characteristics of wild-type and chimeric PAX genes

Ribonuclease protection assays showed several-fold higher levels of PAX3–FKHR mRNA relative to wild-type PAX3 mRNA in most PAX3–FKHR-expressing ARMS specimens (Davis and Barr, 1997). Immunoprecipitation analysis confirmed that PAX3–FKHR is overexpressed at the protein level. Similar studies of PAX7–FKHR-expressing tumors revealed that this fusion is consistently overexpressed relative to wild-type PAX7. These findings indicate that overexpression of PAX3–FKHR and PAX7–FKHR relative to the corresponding wild-type PAX genes is characteristic of ARMS tumors, and suggest that overexpression is required to generate a level of fusion product above a critical threshold for oncogenic activity.

Genomic studies revealed a striking difference in the basis of PAX3–FKHR and PAX7–FKHR overexpression. Fluorescence in situ hybridization assays showed in vivo amplification of the fusion gene on extrachromosomal elements in PAX7–FKHR-positive ARMS cases (Barr et al., 1996). These findings were confirmed and extended by quantitative Southern blot assays of the relative copy number of wild-type and rearranged alleles (Barr et al., 1996; Davis and Barr, 1997). In a series of ARMS cases, fusion gene amplification was detected in one of 24 PAX3–FKHR cases and 10 of 11 PAX7–FKHR cases (Fitzgerald et al., 2000). Therefore, in PAX7–FKHR tumors, translocation and amplification often occur sequentially to alter both gene structure and copy number and thereby activate oncogenic activity by complementary strategies.

The mechanism of overexpression in PAX3–FKHR-expressing tumors was elucidated by analyses of mRNA stability and transcription rate (Davis and Barr, 1997). Ribonuclease protection analysis of RNA from actinomycin D-treated ARMS cell lines showed that PAX3–FKHR and PAX3 transcripts have comparable stabilities. However, nuclear runoff analysis using hybridization targets flanking the t(2;13) breakpoint revealed that PAX3–FKHR is more actively transcribed than PAX3. Therefore, PAX3–FKHR overexpression results from a copy number-independent increase in transcriptional rate that is postulated to result from synergism of PAX3 and FKHR regulatory elements. In contrast, PAX7–FKHR overexpression results from a second genetic event, fusion gene amplification. These findings indicate biological differences between the two fusion genes that are probably related to differences in the regulation of their expression.

Regulation of subcellular localization of wild-type and chimeric FOX proteins

Genetic studies in C. elegans revealed that DAF-16 is involved in a signaling pathway involving the homologues of the insulin receptor (DAF-2), phosphoinositide 3-kinase (AGE-1), and AKT serine/threonine kinases (Lin et al., 1997; Ogg et al., 1997). In mammalian cells, various growth/survival factors (including insulin and IGF-1) activate phosphoinositide 3-kinase and subsequently activate AKT. Based on the C. elegans findings, AKT was shown to phosphorylate FKHR, FKHRL1 and AFX at three conserved sites that conform to the AKT consensus target sequence (Brunet et al., 1999; Takaishi et al., 1999; Tang et al., 1999) (Figure 3). Following phosphorylation, these proteins localize within the cytoplasm, and thus this signaling pathway inactivates transcriptional function by preventing these proteins from reaching their site of action. The mechanism for this change in subcellular localization appears to involve interaction with 14-3-3 scaffolding proteins and cytoplasmic sequestration.

Of the three AKT phosphorylation sites, two are retained in the PAX3–FKHR and PAX7–FKHR proteins in the region distal to the fork head domain (Figure 3). In that ARMS cells produce high levels of IGF2 (Toretsky and Helman, 1996), the fusion proteins would be sequestered in the cytoplasm if these proteins are regulated by this pathway. However, transfection experiments showed that PAX3–FKHR has a nuclear localization and is transcriptionally active in the presence of AKT (del Peso et al., 1999). Therefore, the fusion protein is resistant to upstream regulatory signals and is constitutively retained in the nucleus. Possible explanations include a necessary role for the N-terminal FKHR phosphorylation site that is not retained by the fusion protein or antagonism by N-terminal PAX domains, such as the nuclear localization signal.

DNA binding properties of wild-type and chimeric PAX proteins

In PAX3–FKHR and PAX7–FKHR, the paired box and homeobox are intact whereas the fork head domain is truncated (Davis et al., 1994; Galili et al., 1993; Shapiro et al., 1993) (Figure 3). Since mutations of other fork head domains inactivate DNA binding function (Kaufmann and Knochel, 1996), the truncated fork head domain in the fusion proteins is probably inactive or modifies DNA binding by the PAX3/PAX7 domains. Therefore, the PAX3 and PAX7 DNA binding domains are postulated to provide the DNA binding specificity for the fusion proteins.

Initial studies focused on binding to the e5 sequence, which was identified in the Drosophila even-skipped gene as a binding site for the paired gene product, which also contains a paired box and homeodomain (Goulding et al., 1991; Treisman et al., 1991). Footprinting experiments with e5 and its derivatives showed that the paired and PAX3 proteins protect an 18–27 bp region, which can be subdivided into paired box and homeodomain binding sites that include GTTCC and ATTA motifs, respectively (Chalepakis et al., 1994a; Treisman et al., 1991). Optimal binding of PAX3 to e5 or its derivatives requires both sites (Chalepakis et al., 1994a; Goulding et al., 1991). Comparison of full-length PAX3 to a truncated form lacking the homeodomain demonstrates a qualitative difference in binding affinity. Subsequent studies of PAX3 proteins with paired box or homeodomain mutations (from splotch and Waardenburg syndrome cases) revealed altered binding activity of both the unmutated and mutated domains, and thus indicated a functional interaction between the paired box and homeodomain within the wild-type PAX3 protein (Fortin et al., 1997; Underhill and Gros, 1997; Underhill et al., 1995).

Additional studies of PAX family members indicated that the paired box is comprised of two subdomains (Czerny et al., 1993). PAX5 targets can be divided into two sets of binding sites, the longer class I sites that require both N-terminal and C-terminal subdomains, and the shorter class II sites (which includes e5 and related sequences) that only require the N-terminal subdomain. A comparison of PAX3 and PAX5 DNA binding initially indicated that only PAX5 can bind class I sites whereas both PAX3 and PAX5 can bind class II sites (Schafer et al., 1994). However, subsequent studies identified an alternative splice that removes a single glutamine residue in the paired box of PAX3 and increases the affinity for class I sites (Vogan et al., 1996) (Figure 3). An analysis of PAX7 showed that the analogous glutamine-containing isoform interacts with e5 and related class II sites with a binding affinity similar to that of PAX3 (Schafer et al., 1994). Furthermore, a comparable alternative splice that removes a glutamine from the paired box was also identified in PAX7, though binding activity was not assayed (Vogan et al., 1996). Since both splice forms of PAX3 and PAX7 are co-expressed, the differing binding activity of the two isoforms will contribute to a complex pattern of overall binding function.

Electrophoretic mobility shift assays of the PAX3-FKHR protein showed binding to e5 and related sites (Fredericks et al., 1995; Sublett et al., 1995). Whereas mutations within the paired box or homeodomain of PAX3–FKHR abolished interaction with these sites, deletion of the truncated fork head domain did not affect binding. Comparison of the autoradiographic intensities of the protein–DNA complexes formed with PAX3 and PAX3–FKHR indicated that the binding affinity of isolated wild-type PAX3 protein for e5 was greater than that of PAX3–FKHR. Therefore, even though the wild-type and fusion proteins contain the same PAX3 DNA binding domain, in vitro binding function is partly dependent on the larger protein context.

PAX3 binding sites were also isolated from random oligonucleotide pools by a PCR-based selection strategy. A bacterially synthesized protein containing the PAX3 paired box identified consensus binding sequences of TCGTCAC(G/A)C(T/C/A)(T/C)(C/A/T)A and CGTCACG(G/C)TT (Chalepakis and Gruss, 1995; Epstein et al., 1996). Similar experiments with a bacterial synthetic protein containing the PAX3 paired box and homeodomain identified ATTA and GTNNN motifs in the majority of isolated binding sites, and the sequence ATTA-(N)n-GTTAT in 20% of the binding sites (Phelan and Loeken, 1998). This finding is consistent with the hypothesis of paired box and homeobox binding motifs connected by a variable spacer.

Several studies also demonstrated protein–protein interactions involving PAX3 and/or PAX7 that may influence DNA binding activity. PAX3 and PAX7 are capable of forming homodimers (PAX3/PAX3 or PAX7/PAX7) and heterodimers (PAX3/PAX7) that can bind to a palindromic homeodomain-binding site (P2-TAATCAATTA) with comparable or enhanced binding compared to the monomers (Schafer et al., 1994). In addition, the homeobox-containing protein Msx1, which is coexpressed with Pax3 in migrating limb muscle precursors, can interact with the Pax3 protein and inhibit Pax3 DNA binding activity (Bendall et al., 1999).

Transcriptional properties of wild-type and chimeric PAX proteins

To analyse sequence-specific transcriptional regulatory function of the wild-type and fusion proteins, constructs expressing the full-length proteins were transfected into mammalian cells with a reporter construct containing a minimal adenoviral E1b promoter and PAX3/PAX7 binding sites (Bennicelli et al., 1996, 1999; Fredericks et al., 1995). In experiments using various cell lines and binding sites, wild-type PAX3 expression resulted in a low but detectable level of transcriptional activation whereas wild-type PAX7 expression did not result in detectable transcriptional activity. In contrast, the PAX3–FKHR and PAX7–FKHR proteins similarly induced much higher levels of transcriptional activity, greater than 10-fold more activity than wild-type PAX3. In separate experiments using reporter constructs with a herpesvirus thymidine kinase promoter and PAX3 binding sites, transcriptional activation by PAX3–FKHR was only twofold higher than that by PAX3 (Sublett et al., 1995). These experiments demonstrate that the PAX3–FKHR and PAX7–FKHR proteins can function as transcription factors, specifically activating expression of genes containing PAX3/PAX7 DNA binding sites. Furthermore, the fusion proteins are more potent transcriptional activators than the wild-type proteins, though the increased potency may be promoter-dependent.

To investigate the differing transcriptional potencies of the wild-type and fusion proteins, initial studies focused on the activities of the C-terminal PAX3, PAX7, and FKHR regions (Bennicelli et al., 1995, 1999; Chalepakis et al., 1994b; Sublett et al., 1995). These C-terminal regions were examined independent of the respective DNA binding domains by generating fusions with the GAL4 DNA binding domain. In these GAL4 fusion constructs, the C-terminal PAX3, PAX7, and FKHR regions acted as highly potent activation domains with essential transactivation domains located at the 3′ end of each coding region (Figure 3). The PAX3 and PAX7 transactivation domains are serine- and threonine-rich, whereas the FKHR transactivation domain contains both acidic- and serine-, threonine-rich regions. In addition to these domains, positive modifying elements were found in adjacent regions. These data demonstrate that the wild-type and fusion proteins contain comparably potent, yet structurally distinct transcriptional activation domains which are switched by the translocations in ARMS.

The finding of similar transcriptional potencies of the C-terminal domains and contrasting potencies of the full-length proteins suggested that the activity of the C-terminal activation domains is modulated by other portions of the full-length protein. To explore this hypothesis, the GAL4 DNA binding domain was joined to the full-length wild-type and fusion proteins (Bennicelli et al., 1996, 1999). When fused to GAL4, full-length PAX3–FKHR or PAX7–FKHR activated transcription of a GAL4-dependent reporter gene with an activity 10-fold higher than the corresponding full-length wild-type protein. In addition, the full-length PAX–FKHR constructs had several-fold lower activity than the C-terminal FKHR construct whereas the full-length wild-type PAX3 or PAX7 construct had one to two orders of magnitude less activity than the C-terminal PAX3 or PAX7 construct. These findings indicate that an inhibitory domain is present in the N-terminal PAX3 and PAX7 regions that effectively inhibits the activity of the C-terminal PAX3 and PAX7 domains but only has a modest effect on the C-terminal FKHR domain (Figure 3). Therefore the translocations create potent transcriptional activators by introducing a C-terminal transactivation domain that is relatively insensitive to the inhibitory effects of the N-terminal PAX3 and PAX7 domains. Furthermore, differing interactions between the N-terminal inhibitory and C-terminal activation domains in PAX3 and PAX7 may explain the differing transcriptional activities of these wild-type proteins.

Additional studies suggest that the N-terminal PAX3 and PAX7 domains function as transcriptional inhibitory modules in intermolecular and intramolecular regulation by binding co-repressors. Deletion analysis localized the inhibitory activity to two N-terminal domains, the homeodomain and the N-terminus including the first half of the paired box (Bennicelli et al., 1996, 1999) (Figure 3). These same two regions were also shown to have independent transcriptional repression activity when fused to the GAL4 DNA binding domain (Chalepakis et al., 1994b). Protein interaction studies subsequently determined that the putative co-repressors DAXX, RB1, and HIRA bind to PAX3, and for DAXX and HIRA, interactions with PAX7 were also noted (Hollenbach et al., 1999; Magnaghi et al., 1998; Wiggan et al., 1998). Deletion studies suggest that these proteins interact with the PAX3 homeodomain, though the DAXX interactions also appear to involve the octapeptide and other N-terminal regions. When co-expressed with PAX3, both DAXX and RB1 result in titratable inhibition of PAX3-mediated transcriptional activation, thus supporting their function as transcriptional co-repressors. In addition, DAXX can also bind PAX3–FKHR but is unable to repress transcriptional activation by the fusion transcription factor. These findings support the transcriptional gain of function model and suggest that increased transcriptional potency of the fusion proteins results from unresponsiveness to the repressive activity of bound co-repressors.

Transcriptional targets of wild-type and chimeric PAX proteins

Using electrophoretic mobility shift assays in conjunction with partial or full-length protein constructs, putative PAX3 binding sites were identified in several mammalian genes. These sites were located within the 5′ promoters of the MET, MITF, TYRP1, PDGFRA, BCL2L1 (BCL–XL), and RET genes as well as in the 3′ untranslated region of the NF1 gene and the first intron of the NRCAM gene (Epstein et al., 1995, 1996, 1998; Galibert et al., 1999; Kallunki et al., 1995; Lang et al., 2000; Margue et al., 2000; Watanabe et al., 1998). These putative binding sites contain sequences resembling a paired box binding motif (NF1, MET and RET), a homeobox binding motif (PDGFRA and BCL–XL), or both motifs (MITF, TYRP1 and NRCAM). In one other study, a fragment from the NCAM promoter was shown to bind to cellular extracts expressing the paired box but not extracts expressing the entire PAX3 protein and thus the significance of this binding site is unclear (Chalepakis et al., 1994b).

These binding sites were further investigated in assays in which a PAX3 or PAX3–FKHR expression construct was co-transfected with a reporter construct containing the binding sites. In these assays, PAX3 activated transcription from MITF, TYRP1 and RET binding sites, whereas PAX3 repressed transcription from the NCAM element (Chalepakis et al., 1994b; Galibert et al., 1999; Lang et al., 2000; Watanabe et al., 1998). In the case of the PDGFRA site, the finding of activation by PAX3–FKHR but not PAX3 supports the gain of transcriptional function model described previously (Epstein et al., 1998). In contrast, both PAX3 and PAX3–FKHR activated transcription from NF1 and BCL–XL sites without substantial differences in transcriptional activity between the wild-type and fusion proteins (Epstein et al., 1995; Margue et al., 2000).

Additional studies investigated endogenous gene expression changes following manipulations of PAX3 or PAX3–FKHR expression. Two widely used strategies are introduction of a PAX3 or PAX3–FKHR expression construct into specific cell types, and analysis of homozygous splotch mice. The transfection approach demonstrated enhanced expression of MET, BCL–XL, and RET and thus provides further support that these genes are direct transcriptional targets of PAX3 or PAX3–FKHR (Epstein et al., 1996; Lang et al., 2000; Margue et al., 2000). A more complicated situation exists for PDGFRA in which transient transfection of PAX3–FKHR but not PAX3 into P19 cells activates PDGFRA expression whereas stable integration of either PAX3 or PAX3–FKHR in NIH3T3 cells represses expression (Epstein et al., 1998; Khan et al., 1998). These transfection studies also identified additional candidate targets, as exemplified by the cDNA array analysis of NIH3T3 cells that stably express PAX3–FKHR (Khan et al., 1998). This transcriptional profiling study showed increased expression of IGF2, IGFBP5 and MERTK; and decreased expression of FISP12, PMX1 and DAF. The homozygous splotch expression studies also identified genes whose expression is increased (e.g., versican) or decreased (e.g., lbx1) in specific tissues in the absence of functional Pax3 protein (Henderson et al., 1997; Mennerich et al., 1998). Though these various studies suggest possible target genes, these approaches do not specifically determine whether these genes are directly bound and regulated by PAX3 or PAX3–FKHR. Some of these altered expression events may occur further downstream such that additional steps or signals intervene between the PAX transcription factor and the assayed gene.

Phenotypic roles of wild-type PAX proteins: gene knockout studies

The transcriptional studies of PAX3–FKHR and PAX7–FKHR are consistent with the hypothesis that the translocations activate the oncogenic potential of PAX3 and PAX7 by dysregulating or exaggerating their normal function in the myogenic lineage. Clues for this normal function may be deduced from the skeletal muscle phenotype of splotch and Pax7 knockout mice. In the homozygous splotch animals, limb musculature fails to develop whereas the axial musculature shows varying degrees of malformation (Franz et al., 1993). The defects in muscle development appears to result from increased apoptosis within the myogenic precursor pool of the lateral dermomyotome (Borycki et al., 1999) and from failure of surviving myogenic precursors to migrate from the somites into the limb buds (Bober et al., 1994; Daston et al., 1996; Goulding et al., 1994). This limb musculature defect in splotch mice is associated with reduced expression of the Met receptor in myogenic progenitors (Daston et al., 1996; Epstein et al., 1995; Yang et al., 1996), and is also found in murine strains in which the gene encoding Met was mutated by gene targeting (Bladt et al., 1995). The established role of Met in cell motility signaling and the finding that the gene encoding Met is a potential transcriptional target of the Pax3 protein suggest that the migration problem is due to defective regulation of Met expression by the mutant Pax3 transcription factor. Therefore, these studies of splotch mice suggest possible roles for PAX3 in the stimulation of motility and maintenance of a viable population by facilitating growth or inhibiting apoptosis.

Additional studies of splotch mice and Pax7 knockout mice revealed roles for Pax3 and Pax7 in facilitating myogenic differentiation. In particular, when the homozygous Pax3 mutation is combined with a homozygous Myf5 mutation, almost all trunk (as well as limb) musculature fails to form (Tajbakhsh et al., 1997). This defect in myogenic development involves a failure to activate MyoD expression and subsequent myogenic events. Furthermore, though mice with a homozygous inactivating Pax7 mutation generally have normal muscle development, these animals have a postnatal muscular deficiency evidenced by smaller muscle fibres and decreased size of muscles such as the diaphragm (Seale et al., 2000). This muscle deficiency is attributed to the absence of satellite cells, and implicates Pax7 in the specification of myogenic satellite cells from uncommitted progenitors. Therefore, in addition to motility and viability, PAX3 and PAX7 are proposed to regulate downstream myogenic events. Exaggeration of these activities by PAX3–FKHR and PAX7–FKHR suggests several hypotheses relevant to ARMS pathogenesis.

Phenotypic roles of wild-type and chimeric PAX proteins: gene transfer studies

Gene transfer studies in several model systems further investigated specific functions of these proteins and suggest that the fusion proteins may exert an oncogenic effect through multiple pathways that exaggerate the normal role of the wild-type PAX proteins. In studies investigating growth control, under conditions in which wild-type PAX3 does not transform NIH3T3 cells or chicken embryo fibroblasts, introduction of a PAX3–FKHR expression construct resulted in potent transforming activity (Lam et al., 1999; Scheidler et al., 1996). As a complementary experiment, introduction of a fusion of PAX3 to the KRAB transcriptional repression domain into ARMS cells resulted in loss of transforming activity in culture and tumor suppression in mice (Fredericks et al., 2000). Mutation analysis demonstrated that the homeodomain but not the paired box is required for transformation by PAX3–FKHR (Lam et al., 1999; Scheidler et al., 1996), and thus altered expression of target genes whose binding sites require paired box function may not be required for cellular transformation. Mutation analysis also indicated that the C-terminal FKHR activation domain is required for transformation. Furthermore, substitution of the VP16 activation domain for the C-terminal domain of PAX3 activated transforming activity, whereas substitution of the PAX3 activation domain for the C-terminal domain of PAX3–FKHR ablated transforming activity (Cao and Wang, 2000). These swapping studies support the gain of transcriptional function model, and suggest that the effect of the C-terminal FKHR domain can be mimicked by other activation domains.

In addition to stimulating transforming activity, gene transfer studies also indicated that PAX3–FKHR functions in the maintenance of cell viability by inhibiting apoptosis. Treatment of ARMS cells with an antisense oligonucleotide directed against the PAX3 translational start site resulted in a transient decrease in PAX3–FKHR protein expression (Bernasconi et al., 1996). This expression change was associated with a significant decrease in cell number as well as morphological characteristics and a DNA fragmentation pattern characteristic of apoptosis. In a parallel set of studies, introduction of a tamoxifen-inducible PAX3–KRAB construct into ARMS cells revealed evidence of apoptosis when the construct is induced in low serum conditions or in tumor xenografts (Ayyanathan et al., 2000). A role for BCL–XL is supported by the finding that BCL-XL was downregulated in association with apoptosis in these PAX3–KRAB studies.

A final set of studies addressed the roles of PAX3 and PAX3–FKHR in regulating myogenic pathways. The ability of PAX3 to activate the myogenic program in vitro was revealed by studies in which Pax3-expressing retroviruses were introduced into explanted paraxial mesoderm and other embryonic tissues (Maroto et al., 1997). In this setting, Pax3 induced expression of the myogenic transcription factors MyoD, Myf5, and myogenin as well as myogenic differentiation products. Similarly, when a PAX3–FKHR expression construct was introduced into NIH3T3 cells, cDNA array analysis revealed induction of a large set of genes involved in myogenesis, including the transcription factors MyoD, myogenin, Six1, and Slug (but not Myf5) as well as downstream myogenic products such as various troponins and myosin light chain (Khan et al., 1998). The finding that PAX3 was not able to induce this myogenic transcription program in NIH3T3 cells suggests that PAX3–FKHR has greater transcriptional potency in this less permissive cell environment. A final study focused on the ability of these proteins to inhibit terminal differentiation in two myogenic cell lines, C2C12 myoblasts and MyoD-expressing 10T1/2 cells (Epstein et al., 1995). Whereas differentiation is typically induced in most cells (86–87%) by growth factor withdrawal, PAX3 cDNA transfection reduced the percentage of myosin-expressing colonies to 40–43%, and PAX3–FKHR transfection further reduced this percentage to 17–27%. In contrast to the transformation studies, both the paired box and homeodomain were required to inhibit terminal differentiation. The apparent contrast of these findings from those of the preceding two studies may be explained by the hypothesis that these PAX proteins facilitate entry into the myogenic pathway but inhibit the final steps of the pathway. Alternatively, the myogenic consequences of aberrant PAX3 and PAX3–FKHR expression may be dependent on the specific cell type and other environmental features.

Several recent studies investigated the effect of PAX3–FKHR on mammalian development, and have found a diversity of phenotypic effects. Transgenic mice expressing PAX3–FKHR from a cloned murine Pax3 promoter demonstrate pigmentary and neurological alterations reminiscent of the splotch phenotype (Anderson et al., 2001). In a second study that used a gene targeting strategy to introduce the 3′ FKHR cDNA into the endogenous murine Pax3 locus, mice with the heterozygous knock-in allele were not viable, probably as a result of heart defects and a small disorganized diaphragm (Lagutina et al., 2000). Finally, in a third study, ES cells expressing the PAX3–FKHR cDNA from CMV expression elements were used to generate chimeric mice that demonstrated an overgrowth phenotype reminiscent of Beckwith–Wiedemann syndrome (Xie et al., 2001). Though the phenotypic features varied in these animal systems, tumor susceptibility was not found in any of these studies. Therefore, the tumorigenic effects of the gene fusion gene may require cooperating events or other features that are not present in these animal models.

Molecular diagnostic and therapeutic studies of ARMS gene fusions

As the genetics and biology of the gene fusions in ARMS were elucidated, studies were initiated to explore clinical applications of these gene fusions (Arden et al., 1996; Barr et al., 1995; de Alava et al., 1995; Downing et al., 1995; Frascella et al., 1998; Reichmuth et al., 1996). Using RT–PCR assays for the PAX3–FKHR and PAX7–FKHR transcripts, multiple tumors were assayed to determine the frequency of these fusions. Approximately 80% of ARMS cases expressed one of the two fusions, with the frequency of PAX3–FKHR being 5–10 times that of PAX7–FKHR. Molecular diagnostic studies also revealed a small subset of ARMS cases that do not express either gene fusion. Though these negative results may be explained by variable application of histopathologic diagnostic criteria or the suboptimal quality of some samples, there is clearly a small subset of well-characterized ARMS cases in which these gene fusions are not detectable by standard RT–PCR assays. In these cases, the possibilities of rare variant fusions, other genetic events that can substitute for the characteristic fusions, or a biologically distinct entity with a similar histological appearance should also be considered.

Comparison of the clinical characteristics of PAX3–FKHR and PAX7–FKHR-expressing tumors showed that PAX7–FKHR tumors tend to occur in younger patients, more often present in extremities, are more often localized lesions, and are associated with longer event-free survival (Kelly et al., 1997). In contrast, PAX3–FKHR tumors tend to occur in adolescents as higher stage lesions in a wider variety of sites. Therefore, clinical heterogeneity in ARMS is associated with genetic heterogeneity in the involved fusion genes, and thus identification of the specific gene fusion may provide valuable information for management and outcome assessment.

In addition to providing diagnostic information, studies are also being conducted to investigate whether the fusions can be targeted by therapeutic strategies. As described, strategies were developed to antagonize PAX3–FKHR expression with antisense oligonucleotides or to antagonize PAX3–FKHR function with an engineered PAX3–KRAB transcriptional repressor (Ayyanathan et al., 2000; Bernasconi et al., 1996; Fredericks et al., 2000). Another therapeutic strategy involves transfer of a construct consisting of PAX binding sites that regulate expression of a toxin-encoding gene (Massuda et al., 1997). Cell culture studies demonstrated that this construct is selectively toxic for cells expressing the PAX3–FKHR fusion. In a final strategy, the immunogenicity of peptides spanning the fusion point is being tested (Mackall et al., 2000). Animal studies provided evidence that cytotoxic T lymphocytes directed against the fusion peptide can be generated and will recognize and kill PAX3–FKHR-expressing tumor cells.


In ARMS, the 2;13 and 1;13 translocations are consistent and specific events that provide opportunities to investigate the molecular etiology of this cancer, identify novel markers for diagnosis and monitoring, and develop therapeutic approaches against tumor-specific targets. These translocations juxtapose the transcription factor-encoding genes PAX3 or PAX7 with FKHR to generate PAX3–FKHR and PAX7–FKHR chimeric genes. A consideration of the biological and clinical data on these two fusion genes indicates important similarities that point to a common fundamental mechanism in the pathogenesis of this tumor. However, the biological and clinical data also reveal several striking differences between the two fusions that highlight the heterogeneity within this tumor category and distinctions between two highly related members of the paired box family.

The genetic changes resulting from the 2;13 translocation alter biological activity at the levels of protein function, gene expression, and protein localization. The result is high level, constitutively nuclear expression of a potent fusion transcription factor. These biological changes are postulated to promote aberrant expression of genes that are transcriptional targets of wild-type PAX3 and that normally function in the control of early myogenic development. The aberrant or inappropriate expression of these gene products is postulated to contribute to oncogenic initiation or progression by stimulating cell behaviors such as growth and motility or inhibiting cell behaviors such as apoptosis and terminal differentiation.

The t(1;13) bears a strong similarity to the t(2;13) in that the PAX7–FKHR fusion product joins the PAX7 DNA binding domain and FKHR activation domain. Functional studies indicate that the PAX7–FKHR protein functions an aberrant transcription factor similar to PAX3–FKHR. In contrast to PAX3–FKHR, the juxtaposition of PAX7 and FKHR regulatory elements is apparently not sufficient for PAX7–FKHR overexpression, and instead an amplification event ensues to increase PAX7–FKHR gene copy number. This finding indicates a biological distinction between the two fusions, the potential significance of which is indicated by the finding of clinical differences between PAX3–FKHR and PAX7–FKHR-expressing tumors.

Future investigations will integrate the gene expression, target gene, and functional studies to develop a comprehensive understanding of the aberrant genetic program induced by these translocations. Furthermore, increased efforts must be directed at identifying the collaborating events in ARMS tumorigenesis and developing appropriate model systems to recapitulate these oncogenic pathways in a controlled fashion. In addition to further elucidating how the PAX3–FKHR and PAX7–FKHR fusions contribute to tumorigenesis, these studies need to address the differences between the two fusions in tumorigenic pathways. These studies will ultimately enhance the utility of molecular genetics in the diagnosis and monitoring of this cancer, and will indicate directions in which possible therapeutic strategies can be developed to interrupt these tumorigenic pathways.


  1. Anderson MJ, Shelton GD, Cavenee WK, Arden KC . 2001 Proc. Natl. Acad. Sci. USA 98: 1589–1594

  2. Anderson MJ, Viars CS, Czekay S, Cavenee WK, Arden KC . 1998 Genomics 47: 187–199

  3. Arden KC, Anderson MJ, Finckenstein FG, Czekay S, Cavenee WK . 1996 Genes Chrom. Cancer 16: 254–260

  4. Ayyanathan K, Fredericks WJ, Berking C, Herlyn M, Balakrishnan C, Gunther E, Rauscher III FJ . 2000 Cancer Res. 60: 5803–5814

  5. Baldwin CT, Hoth CF, Amos JA, da-Silva EO, Milunsky A . 1992 Nature 355: 637–638

  6. Barr FG, Chatten J, D'Cruz CM, Wilson AE, Nauta LE, Nycum LM, Biegel JA, Womer RB . 1995 JAMA 273: 553–557

  7. Barr FG, Fitzgerald JC, Ginsberg JP, Vanella ML, Davis RJ, Bennicelli JL . 1999 Cancer Res. 59: 5443–5448

  8. Barr FG, Galili N, Holick J, Biegel JA, Rovera G, Emanuel BS . 1993 Nat. Genet. 3: 113–117

  9. Barr FG, Nauta LE, Davis RJ, Schafer BW, Nycum LM, Biegel JA . 1996 Hum Mol. Genet. 5: 15–21

  10. Barr FG, Nauta LE, Hollows JC . 1998 Cancer Genet. Cytogenet. 102: 32–39

  11. Bendall AJ, Ding J, Hu G, Shen MM, Abate-Shen C . 1999 Development 126: 4965–4976

  12. Bennicelli JL, Advani S, Schafer BW, Barr FG . 1999 Oncogene 18: 4348–4356

  13. Bennicelli JL, Edwards RH, Barr FG . 1996 Proc. Natl. Acad. Sci. USA 93: 5455–5459

  14. Bennicelli JL, Fredericks WJ, Wilson RB, Rauscher FJ, Barr FG . 1995 Oncogene 11: 119–130

  15. Bernasconi M, Remppis A, Fredericks WJ, Rauscher FJ, Schafer BW . 1996 Proc. Natl. Acad. Sci. USA 93: 13164–13169

  16. Biegel JA, Meek RS, Parmiter AH, Conard K, Emanuel BS . 1991 Genes Chrom. Cancer 3: 483–484

  17. Bladt F, Riethmacher D, Isenmann S, Aguzzi A, Birchmeier C . 1995 Nature 376: 768–771

  18. Bober E, Franz T, Arnold HH, Gruss P, Tremblay P . 1994 Development 120: 603–612

  19. Borkhardt A, Repp R, Haas OA, Leis T, Harbott J, Kreuder J, Hammermann J, Henn T, Lampert F . 1997 Oncogene 14: 195–202

  20. Borycki AG, Li J, Jin F, Emerson CP, Epstein JA . 1999 Development 126: 1665–1674

  21. Brunet A, Bonni A, Zigmond MJ, Lin MZ, Juo P, Hu LS, Anderson MJ, Arden KC, Blenis J, Greenberg ME . 1999 Cell 96: 857–868

  22. Burri M, Tromvoukis Y, Bopp D, Frigerio G, Noll M . 1989 EMBO J. 8: 1183–1190

  23. Cao Y, Wang C . 2000 J. Biol. Chem. 275: 9854–9862

  24. Chalepakis G, Goulding M, Read A, Strachan T, Gruss P . 1994a Proc. Natl. Acad. Sci. USA 91: 3685–3689

  25. Chalepakis G, Gruss P . 1995 Gene 162: 267–270

  26. Chalepakis G, Jones FS, Edelman GM, Gruss P . 1994b Proc. Natl. Acad. Sci. USA 91: 12745–12749

  27. Czerny T, Schaffner G, Busslinger M . 1993 Genes Dev. 7: 2048–2061

  28. Dahl E, Koseki H, Balling R . 1997 Bioessays 19: 755–765

  29. Daston G, Lamar E, Olivier M, Goulding M . 1996 Development 122: 1017–1027

  30. Davis RJ, Barr FG . 1997 Proc. Natl. Acad. Sci. USA 94: 8047–8051

  31. Davis RJ, Bennicelli JL, Macina RA, Nycum LM, Biegel JA, Barr FG . 1995 Hum Mol. Genet. 4: 2355–2362

  32. Davis RJ, D'Cruz CM, Lovell MA, Biegel JA, Barr FG . 1994 Cancer Res. 54: 2869–2872

  33. de Alava E, Ladanyi M, Rosai J, Gerald WL . 1995 Am. J. Pathol. 147: 1584–1591

  34. del Peso L, Gonzalez VM, Hernandez R, Barr FG, Nunez G . 1999 Oncogene 18: 7328–7333

  35. Douglass EC, Rowe ST, Valentine M, Parham DM, Berkow R, Bowman WP, Maurer HM . 1991 Genes Chrom. Cancer 3: 480–482

  36. Douglass EC, Valentine M, Etcubanas E, Parham D, Webber BL, Houghton PJ, Houghton JA, Green AA . 1987 Cytogenet. Cell Genet. 45: 148–155

  37. Downing JR, Khandekar A, Shurtleff SA, Head DR, Parham DM, Webber BL, Pappo AS, Hulshof MG, Conn WP, Shapiro DN . 1995 Am. J. Pathol. 146: 626–634

  38. Epstein DJ, Vekemans M, Gros P . 1991 Cell 67: 767–774

  39. Epstein JA, Lam P, Jepeal L, Maas RL, Shapiro DN . 1995 J. Biol. Chem. 270: 11719–11722

  40. Epstein JA, Shapiro DN, Cheng J, Lam PY, Maas RL . 1996 Proc. Natl. Acad. Sci. USA 93: 4213–4218

  41. Epstein JA, Song B, Lakkis M, Wang C . 1998 Mol. Cell. Biol. 18: 4118–4130

  42. Fitzgerald JC, Scherr AM, Barr FG . 2000 Cancer Genet. Cytogenet. 117: 37–40

  43. Fortin AS, Underhill DA, Gros P . 1997 Hum. Mol. Genet. 6: 1781–1790

  44. Franz T, Kothary R, Surani MA, Halata Z, Grim M . 1993 Anat. Embryol. 187: 153–160

  45. Frascella E, Toffolatti L, Rosolen A . 1998 Cancer Genet. Cytogenet. 102: 104–109

  46. Fredericks WJ, Ayyanathan K, Herlyn M, Friedman JR, Rauscher FJ III . 2000 Mol. Cell. Biol. 20: 5019–5031

  47. Fredericks WJ, Galili N, Mukhopadhyay S, Rovera G, Bennicelli J, Barr FG, Rauscher FJ . 1995 Mol. Cell. Biol. 15: 1522–1535

  48. Furuyama T, Nakazawa T, Nakano I, Mori N . 2000 Biochem. J. 349: 629–634

  49. Galibert MD, Yavuzer U, Dexter TJ, Goding CR . 1999 J. Biol. Chem. 274: 26894–26900

  50. Galili N, Davis RJ, Fredericks WJ, Mukhopadhyay S, Rauscher FJ, Emanuel BS, Rovera G, Barr FG . 1993 Nat. Genet. 5: 230–235

  51. Goulding M, Lumsden A, Paquette AJ . 1994 Development 120: 957–971

  52. Goulding MD, Chalepakis G, Deutsch U, Erselius JR, Gruss P . 1991 EMBO J. 10: 1135–1147

  53. Guo S, Rena G, Cichy S, He X, Cohen P, Unterman T . 1999 J. Biol. Chem. 274: 17184–17192

  54. Henderson DJ, Ybot-Gonzalez P, Copp AJ . 1997 Mech. Dev. 69: 39–51

  55. Hillion J, Le Coniat M, Jonveaux P, Berger R, Bernard OA . 1997 Blood 90: 3714–3719

  56. Hollenbach AD, Sublett JE, McPherson CJ, Grosveld G . 1999 EMBO J. 18: 3702–3711

  57. Jostes B, Walther C, Gruss P . 1990 Mech. Dev. 33: 27–37

  58. Kaestner KH, Knochel W, Martinez DE . 2000 Genes Dev. 14: 142–146

  59. Kallunki P, Jenkinson S, Edelman GM, Jones FS . 1995 J. Biol. Chem. 270: 21291–21298

  60. Kaufmann E, Knochel W . 1996 Mech. Dev. 57: 3–20

  61. Kelly KM, Womer RB, Sorensen PH, Xiong QB, Barr FG . 1997 J. Clin. Oncol. 15: 1831–1836

  62. Khan J, Simon R, Bittner M, Chen Y, Leighton SB, Pohida T, Smith PD, Jiang Y, Gooden GC, Trent JM, Meltzer PS . 1998 Cancer Res. 58: 5009–5013

  63. Lagutina I, Sublett J, McPherson C, Conway S, Grosveld G . 2000 Dev. Biol. 222: 236

  64. Lam PY, Sublett JE, Hollenbach AD, Roussel MF . 1999 Mol. Cell. Biol. 19: 594–601

  65. Lang D, Chen F, Milewski R, Li J, Lu MM, Epstein JA . 2000 J. Clin. Invest. 106: 963–971

  66. Lin K, Dorman JB, Rodan A, Kenyon C . 1997 Science 278: 1319–1322

  67. Macina RA, Barr FG, Galili N, Riethman HC . 1995 Genomics 26: 1–8

  68. Mackall C, Berzofsky J, Helman LJ . 2000 Clin. Orthop. 373: 25–31

  69. Magnaghi P, Roberts C, Lorain S, Lipinski M, Scambler PJ . 1998 Nat. Genet. 20: 74–77

  70. Mansouri A, Stoykova A, Torres M, Gruss P . 1996 Development 122: 831–838

  71. Margue CM, Bernasconi M, Barr FG, Schafer BW . 2000 Oncogene 19: 2921–2929

  72. Maroto M, Reshef R, Munsterberg AE, Koester S, Goulding M, Lassar AB . 1997 Cell 89: 139–148

  73. Massuda ES, Dunphy EJ, Redman RA, Schreiber JJ, Nauta LE, Barr FG, Maxwell IH, Cripe TP . 1997 Proc. Natl. Acad. Sci. USA 94: 14701–14706

  74. Medema RH, Kops GJ, Bos JL, Burgering BM . 2000 Nature 404: 782–787

  75. Mennerich D, Schafer K, Braun T . 1998 Mech. Dev. 73: 147–158

  76. Ogg S, Paradis S, Gottlieb S, Patterson GI, Lee L, Tissenbaum HA, Ruvkun G . 1997 Nature 389: 994–999

  77. Phelan SA, Loeken MR . 1998 J. Biol. Chem. 273: 19153–19159

  78. Reichmuth C, Markus MA, Hillemanns M, Atkinson MJ, Unni KK, Saretzki G, Hofler H . 1996 J. Pathol. 180: 50–57

  79. Schafer BW, Czerny T, Bernasconi M, Genini M, Busslinger M . 1994 Nucleic Acids Res. 22: 4574–4582

  80. Scheidler S, Fredericks WJ, Rauscher FJ, Barr FG, Vogt PK . 1996 Proc. Natl. Acad. Sci. USA 93: 9805–9809

  81. Seale P, Sabourin LA, Girgis-Gabardo A, Mansouri A, Gruss P, Rudnicki MA . 2000 Cell 102: 777–786

  82. Shapiro DN, Sublett JE, Li B, Downing JR, Naeve CW . 1993 Cancer Res. 53: 5108–5112

  83. Sublett JE, Jeon IS, Shapiro DN . 1995 Oncogene 11: 545–552

  84. Tajbakhsh S, Rocancourt D, Cossu G, Buckingham M . 1997 Cell 89: 127–138

  85. Takaishi H, Konishi H, Matsuzaki H, Ono Y, Shirai Y, Saito N, Kitamura T, Ogawa W, Kasuga M, Kikkawa U, Nishizuka Y . 1999 Proc. Natl. Acad. Sci. USA 96: 11836–11841

  86. Tang ED, Nunez G, Barr FG, Guan KL . 1999 J. Biol. Chem. 274: 16741–16746

  87. Tassabehji M, Read AP, Newton VE, Harris R, Balling R, Gruss P, Strachan T . 1992 Nature 355: 635–636

  88. Toretsky JA, Helman LJ . 1996 J. Endocrinol. 149: 367–372

  89. Treisman J, Harris E, Desplan C . 1991 Genes Dev. 5: 594–604

  90. Tremblay P, Gruss P . 1994 Pharmacol. Ther. 61: 205–226

  91. Tsokos M . 1994 Semin. Diagn. Pathol. 11: 26–38

  92. Tsokos M, Webber BL, Parham DM, Wesley RA, Miser A, Miser JS, Etcubanas E, Kinsella T, Grayson J, Glatstein E, Pizzo P, Triche TJ . 1992 Arch. Pathol. Lab. Med. 116: 847–855

  93. Turc-Carel C, Lizard-Nacol S, Justrabo E, Favrot M, Philip T, Tabone E . 1986 Cancer Genet. Cytogenet. 19: 361–362

  94. Underhill DA, Gros P . 1997 J. Biol. Chem. 272: 14175–14182

  95. Underhill DA, Vogan KJ, Gros P . 1995 Proc. Natl. Acad. Sci. USA 92: 3692–3696

  96. Vogan KJ, Underhill DA, Gros P . 1996 Mol. Cell. Biol. 16: 6677–6686

  97. Vorobyov E, Mertsalov I, Dockhorn-Dworniczak B, Dworniczak B, Horst J . 1997 Genomics 45: 168–174

  98. Wang-Wuu S, Soukup S, Ballard E, Gotwals B, Lampkin B . 1988 Cancer Res. 48: 983–987

  99. Watanabe A, Takeda K, Ploplis B, Tachibana M . 1998 Nat. Genet. 18: 283–286

  100. Wexler LH, Helman LJ . (1997). Principles and Practices of Pediatric Oncology Pizzo PA and Poplack DG. (eds) Lippincott-Raven: Philadelphia pp. 799–829

  101. Wiggan O, Taniguchi-Sidle A, Hamel PA . 1998 Oncogene 16: 227–236

  102. Williams BA, Ordahl CP . 1994 Development 120: 785–796

  103. Xie E, Zhu Q, Barr F, Robinson M, Sferra T, Muthusamy R, Qualman S, Durbin J . 2001 Mod. Pathol. 14: 4P

  104. Yang XM, Vogan K, Gros P, Park M . 1996 Development 122: 2163–2171

Download references


This work was supported in part by NIH grants CA64202, CA71838 and CA89461.

Author information

Correspondence to Frederic G Barr.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Barr, F. Gene fusions involving PAX and FOX family members in alveolar rhabdomyosarcoma. Oncogene 20, 5736–5746 (2001).

Download citation


  • rhabdomyosarcoma
  • transcription factor
  • translocation
  • paired box
  • fork head

Further reading