BIOLOGICAL OVERVIEW

Wnt transduction is mediated by the association of ß-catenin (see Shaggy) with nuclear TCF DNA binding factors (see Pangolin). The products of two newly identified Drosophila segment polarity genes, legless (lgs), and pygopus (pygo: Pygopus refers to a legless lizard with scaly skin) are required for Wnt signal transduction at the level of nuclear ß-catenin. Lgs encodes the homolog of human BCL9; genetic and molecular evidence is provided that these proteins exert their function by physically linking Pygo to ß-catenin. These results suggest that the recruitment of Pygo permits ß-catenin to transcriptionally activate Wnt target genes and raise the possibility that a deregulation of these events may play a causal role in the development of B cell malignancies (Kramps, 2002).

In an attempt to identify new components of the Wnt signal transduction pathway, a screen was performed for dominant suppressors of the rough eye phenotype caused by a sevenless-wg transgene (sev-wg) that drives ectopic expression of wingless (wg) during eye development. The majority of suppressors found in this screen belong to one of three complementation groups, two of which represent alleles of the previously identified loci armadillo (arm) and pangolin (pan), which encode the Drosophila homologs of ß-catenin and TCF, respectively. However, six mutations were recessive alleles of a locus on chromosome 4 that was designated legless (lgs). In transheterozygous situations, three lgs alleles interact genetically with either arm or pan mutations, causing adult phenotypes characteristic of reduced wg activity (see below). These results suggested that lgs, like arm and pan, encodes a component of the Wg transduction pathway and raised the possibility that these components may participate in a common, critical task (Kramps, 2002).

Chromosome 4 does not permit meiotic mapping because, unlike all other Drosophila chromosomes, it fails to undergo spontaneous recombination during female meiosis. To genetically map lgs, an alternative approach was used. A series of 60 terminal deficiencies of Dp(4;Y)E were used, where Y is a chromosome that carries at its tip an extra copy of chromosome 4. These deficiencies resulted from random breaks of dicentric Dp(4;Y)E chromosomes, which were produced by Flp-mediated unequal sister-chromatid exchange. Each terminal deficiency of Dp(4;Y)E was analyzed cytologically and tested for its ability to rescue a homozygous lgs mutant genotype. Two terminal deficiencies were identified that barely differ cytologically, yet one, PE6.5, is lgs⁺ whereas the other, PE9.1, is lgs^-. Based on these results, the lgs gene must be located at position 102B7 (Kramps, 2002).

A chromosomal walk was initiated with probes flanking P element MS209 to isolate a contiguous stretch of 150 kb DNA covering the breakpoints of both PE6.5 and PE9.1, as determined by chromosome in situ hybridization. Because the region between these two breakpoints must contain at least an essential part of the lgs gene, focus was placed on transcribed regions within this interval and their sequences were identified and compared with those derived from lgs mutant chromosomes. One transcript with an open reading frame encoding a protein of 1464 amino acids contained point mutations in five out of six lgs alleles, two of which, lgs^20F and lgs^7I, are nonsense mutations predicted to cause a premature termination of translation. This candidate gene must represent lgs, since a full-length cDNA driven by the ubiquitous promoter of the tubulinalpha1 gene was found to completely rescue the lethality and other phenotypes associated with homozygous lgs mutations (Kramps, 2002).

Key to the discovery of a role of lgs in the WNT pathway was a sensitized morphological reporter system for Wg signaling. lgs is required in Wg-receiving cells for the expression of Wg target genes and it encodes a nuclear protein with slight, but significant, structural similarities to human BCL9. Genetic epistasis experiments and protein-protein interaction assays indicate that Lgs and BCL9 bind to Arm and ß-catenin, respectively. The identification of Lgs permitted the isolation of Pygo, which like Lgs, is required for Wg signaling. Pygo and its human homologs bind to Lgs and BCL9, respectively. Lgs/BCL9 and the DNA binding proteins Pan/TCF can bind simultaneously to Arm/ß-catenin. Together, these findings provide the basis for postulating the existence of a nuclear multiprotein complex, in which ß-catenin ties Lgs/BCL9 and thus Pygo to Wnt-inducible promoters (Kramps, 2002).

Three sources of in vivo evidence for this model have been provided by the nature of the mutations recovered in the sensitized screen. (1) The two alleles found in the pan gene both encode proteins with N-terminal mutations that have a deleterious effect on the affinity to ß-catenin. (2) The lgs^17E allele encodes a protein with a mutation in the HD2 domain that abolishes the interaction between Lgs and Arm. (3) The pygo¹³⁰ allele encodes a protein lacking its PHD finger and hence the ability to bind to Lgs. In all cases, the disruption of only one of the three protein-protein interactions within the tetrapartite ß-catenin complex leads to a common outcome, namely a decrease in Wg signal response (Kramps, 2002).

All previously identified proteins that have been implicated in mediating ß-catenin-dependent activation of Wnt target genes, such as TATA binding protein (TBP; Hecht, 1999), TIP49 (Bauer, 1998), p300/CBP (Drosophila homolog: Nejire), or Brg-1 (Drosophila homolog: Brahma), have pleiotropic roles, and their specific in vivo contribution to Wnt transduction cannot be assessed by genetic means. In sharp contrast, the newly identified proteins Lgs/BCL9 and Pygo appear to represent components that are dedicated to the Wnt/Wg signaling pathway. This argument is based largely on an inability to detect lgs or pygo phenotypes that differ from those caused by reduction of Wg activity. The argument is reinforced, however, by the observation that the PHD finger of Pygo and all but HD2 of Lgs are dispensable if Pygo is allowed to directly interact with ß-catenin in the form of a hybrid Pygo[DeltaPHD]-HD2 protein (Kramps, 2002).

The activities of Pygo and Lgs are specific to the Wg signaling pathway, where they appear to be universally required to mediate all apparent Wg responses during normal Drosophila development. There is no evidence for any tissue in which Arm transduces a Wg signal in the absence of Pygo and Lgs. However, it is noted that the results do not rule out the possibility that there may be situations in mammalian systems where Pygo or BCL9 proteins are specifically absent or limiting (Kramps, 2002).

The results suggest that the primary function of Lgs/BCL9 is to recruit Pygo into the Arm/ß-catenin complex. An understanding regarding the mechanism of action of Pygo is less complete. In reporter gene assays, Pygo effectively enhances the transcription of TCF reporter genes, indicating that Pygo possesses properties of a transcriptional activator. The human and Drosophila Pygo proteins share two domains of structural similarity: the N-terminal homology domain (NHD) and the C-terminal PHD finger. The PHD finger appears to mediate the binding of Pygo to the HD1 domain of Lgs/BCL9. Hence, it is likely the NHD motif by which Pygo exerts its transcriptional activator function. While it is not yet known which proteins are targeted by the NHD, plausible candidates include TBP-associated factors (TAFs), histone acetyl transferases, and components of chromatin-remodeling complexes. A recent study suggested a role for the Brahma chromatin-remodeling complex in repressing Wg target genes, indicating that altering chromatin conformation may be an obligatory step when activating these genes. Firm answers regarding the mode of action of Pygo will have to await the identification of NHD target proteins (Kramps, 2002).

BCL9 has been identified by the translocation t(1;14)(q21;q32) from a patient with precursor-B-cell acute lymphoblastic leukemia (ALL; Willis, 1998). B-cell malignancies and, in particular, the B-cell non-Hodgkin's lymphomas are often associated with chromosomal translocations in which certain genes become overexpressed due to a juxtaposition to immunoglobulin loci. BCL9 transcript levels are normally very low in B cells, but 50-fold higher in the CEMO-1 cell line from which the translocation break point was cloned (Willis, 1998). The finding that the Drosophila homolog of BCL9, Lgs, is a component of the Wg signaling pathway raises the possibility that activation of the Wnt pathway may be causally linked to certain forms of B-cell leukemia or lymphoma (Kramps, 2002).

The formation of mature B- and T-lymphocytes and other hematopoetic lineages is guided by complex genetic and environmental cues. Some recent reports have implicated Wnt signaling in these processes. Proliferation and differentiation of CD4^-CD8^- double-negative thymocytes require an intact Wnt pathway. Hence, in both pro-T and pro-B cells, Wnt signals might provide important mitogenic stimuli at central developmental stages, although some of the effects of these signals may depend on inputs from additional receptor systems. Further support for the notion that deregulated Wnt signaling could play a role in blood cell cancers comes from recent findings that Wnt3A has a mitogenic effect on pro-B cells and that in pre-B cell ALL patients, Wnt 16 is overexpressed (Kramps, 2002 and references therein).

However, in Drosophila the mere overexpression of Lgs does not lead to an activation of the Wg pathway, and compared to hPYGO1 and 2, expression of BCL9 has only a mild stimulatory effect in TCF reporter assays. It is possible, though, that the expression of high BCL9 levels may render certain cell types more responsive to Wnt inputs. Hence, an additional event, such as the upregulation of Wnt expression, could create conditions in which the high levels of BCL9 cause a deregulated cellular behavior resulting in pre-B ALL. Such a scenario could explain the apparent low incidence of deregulated BCL9 expression in tumor samples (Kramps, 2002 and references therein).

An interesting observation in this respect is the finding that Jurkat T cells and normal T lymphocytes exhibit a vast difference in their responsiveness to ß-catenin. Overexpression of ß-catenin leads to a 150-fold induction of reporter gene expression in Jurkat cells but has no discernible effect in normal T cells, although both cells show a comparable response to a VP16 control. This result has been interpreted to indicate that normal T lymphocytes lack a component necessary for gene activation by nuclear ß-catenin. If such a situation occurs in vivo where the amounts of either BCL9 or Pygo are limiting, a transcriptional upregulation of the respective gene might have a significant impact (Kramps, 2002 and references therein).

The most prevalent activation of the Wnt pathway in cancer is caused by the loss of APC. Mutations in APC occur in >85% of inherited and sporadic colorectal cancers. Such mutations result in the accumulation of nuclear ß-catenin and the concomitant overexpression of Wnt target genes. The ß-catenin-TCF complex has therefore emerged as an attractive target for anticancer drugs. The discovery that Lgs/BCL9 and Pygo are required for the transcriptional activity of Arm/ß-catenin raises the possibility that the protein-protein contacts between the three components may represent additional targets for anticancer intervention. As a proof of principle the effect of disrupting one such interaction was tested on the consequences of mutations in Drosophila APC2. Embryos devoid of wild-type maternal and zygotic dAPC2 function die during embryogenesis with a naked cuticle phenotype caused by the constitutive activation of the Wg pathway. This situation is reverted to a lawn-of-denticle phenotype (loss of Wg signaling) if dAPC2^DeltaS embryos are also mutant for pygo¹³⁰. The Pygo¹³⁰ protein lacks the C-terminal PHD finger. Hence, in dAPC2^DeltaS animals, the loss of wild-type APC2 function is largely, if not completely, ineffective if Pygo can no longer interact with Lgs (Kramps, 2002).

Any drug designed to disrupt a protein-protein interaction involving ß-catenin must be highly specific and should not interfere, for example, with the binding of E-cadherin to ß-catenin. E-cadherin has properties of a tumor suppressor, and its loss has been implicated in the transition from adenoma to carcinoma. Recent structural studies indicate that E-cadherin and TCF use many of the same protein-protein contacts to bind to ß-catenin, complicating the strategy to identify substances that specifically disrupt the ß-catenin-TCF interaction. The observation that TCF and BCL9 do not compete for their binding to ß-catenin suggests that the ß-catenin-BCL9 interaction may provide an attractive alternative for therapeutic intervention (Kramps, 2002).

GENE STRUCTURE

cDNA clone length - 5307

Bases in 5' UTR - 447

Exons - 6

Bases in 3' UTR - 450

PROTEIN STRUCTURE

Amino Acids - 1464

Structural Domains

The Lgs protein sequence contains neither a recognizable protein motif nor does it show sequence homologies to any other Drosophila protein. Also, no proteins encoded in the genomes of Caenorhabditis elegans, mouse, and humans were predicted to share extensive sequence similarities with Lgs. However, a few stretches of ~30 amino acids were identified in Lgs that show a statistically significant match to sequences in human and mouse BCL9 proteins (Willis, 1998). Although short, these patches of protein similarities are arranged in a colinear fashion in these proteins. These regions are referred to as homology domains 1-3 (HD1-3). Intriguingly, all three missense mutations isolated in lgs map within, or in the immediate vicinity of, HD2 (Kramps, 2002).

BCL9 was identified as the gene overexpressed in a cell line derived from a patient with precursor B cell acute lymphoblastic leukemia. It was found that a translocation caused the juxtaposition of the BCL9 gene with regulatory elements of an immunoglobulin locus (Willis, 1998). In both mouse and humans there is an additional BCL9-related gene, referred to as mLgs2 and hLGS2, respectively. No data has been reported describing the normal function of these genes (Kramps, 2002).

To explore the possibility that, despite the low degree of sequence similarity, BCL9 could represent the functional homolog of Lgs, a full-length BCL9 cDNA was assembled from human EST clones and used to generate a tubulinalpha1 promoter-driven transgene. This transgene is able to rescue viability and limb pattern of lgs^17E/lgs^21L animals, which normally die as pharate legless adults. Even animals homozygous for the putative null allele lgs^20F, which causes larval lethality and encodes a severely truncated protein (amino acids 1-383), are rescued to adulthood, although the majority of these individuals (80%) fail to hatch from their pupal cases. Finally, the segment polarity phenotype of embryos lacking both maternal and zygotic lgs function is fully rescued by expression of BCL9. Together, these results provide convincing evidence that lgs encodes the Drosophila homolog of human BCL9 and suggest that the role of these two proteins in Wnt signaling is evolutionarily conserved (Kramps, 2002).

date revised: 6 July 2002

Home page: The Interactive Fly © 1995, 1996 Thomas B. Brody, Ph.D.

Please e-mail comments/corrections to brodyt@codon.nih.gov