BIOLOGICAL OVERVIEW

Lola is a transcription factor that regulates axon guidance. One spliced variant of Lola, that contains a DNA binding domain, is made in glial cells, the accessory cells of neurons. This molecular form probably regulates the production of glial factors, which in turn regulate guidance. Another alternatively spliced variant lacking a DNA binding domain, is produced in neurons. In this case the Lola variant regulates neural molecules responsible for guidance (Giniger, 1994).

Lola is required for pathfinding and targeting of the SNb motor nerve in Drosophila. With complete penetrance and very high expressivity, the motor axons that project through the SNb peripheral nerve in lola mutants fail to form connections to their cognate muscles. Most commonly, SNb appear to 'stall' somewhere between the point at which they would normally separate from the intersegmental nerve and the muscle 6/3 junction. In other hemisegments, SNB axons project through the muscle field but fail to branch into the muscle clefts where synapses should be formed. Careful analysis of muscle phenotypes show that muscle identities are not specified by lola, suggesting that the observed phenotypes are directly due to defective axon guidance. lola is shown to be a dose-dependent regulator of SNb development: by varying the expression of one lola isoform, the extent of interaction of SNb motor axons with their target muscles can be progressively titrated, from no interaction at all, through wild-type patterning, to apparent hyperinnervation. In embryos that overexpress lola in muscle, an apparent 'overgrowth' of Fasciclin 2-immunoreactive is observed where SNb axons contact the VLMs. Some of this material is in the clefts between adjacent muscles, but more is spread out of the clefts and over the surfaces of the muscles. Moreover, in many cases, broad branches extend over the muscles quite separate from the normal projections in the clefts. The Fas 2 immunoreactivity is likely to reflect expanded domains of synaptogenesis. The phenotypes observed from altered expression of Lola suggest that this protein may help orchestrate the coordinated expression of the genes required for faithful SNb development. In addition to its control of SNb morphogenesis, lola also regulates growth of CNS longitudinal axons between successive neuromeres, and growth of ISN axons along lateral peritracheal cells. It may be that some cell surface proteins act in multiple axon growth and guidance decisions, and act in multiple axon growth and guidance decisions, and that lola regulates some of the same genes in all three developmental contexts (Madden, 1999).

Alternative splicing of the lola gene creates a family of 19 transcription factors. All lola isoforms share a common dimerization domain, but 17 have their own unique DNA-binding domains. Seven of these 17 isoforms are present in the distantly-related Dipteran Anopheles gambiae, suggesting that the properties of specific isoforms are likely to be critical to lola function. Analysis of the expression patterns of individual splice variants and of the phenotypes of mutants lacking single isoforms supports this idea and establishes that the alternative forms of lola are responsible for different functions of this gene. Thus, in this system, the alternative splicing of a key transcription factor helps to explain how a small genome encodes all the information that is necessary to specify the enormous diversity of axonal trajectories (Goeke, 2003).

Two isoforms have been reported, and a third isoform is identified in the Berkeley Drosophila Genome Database. All three forms share four exons that encode a common N-terminal region, including the BTB dimerization domain. Two of these isoforms, T and F, splice to unique fifth exons, in each case encoding a pair of zinc fingers. The other isoform, A, is produced if translation continues past the common region, into the adjacent intron, bypassing the final splice. Another lola splice variant was discovered by isolating and sequencing additional lola cDNAs. This splice variant (N) follows the same pattern seen in T and F: the four common exons are spliced to a unique 3' sequence that contains a pair of zinc fingers (Goeke, 2003).

The diversity of Lola protein forms was surveyed by immunoblotting total embryo extract with antibody directed against the Lola common region, and at least 11 apparent Lola protein products were discovered. The pattern of Lola proteins evolves over the course of embryonic development, hinting that some of the different forms of Lola might have different functions (Goeke, 2003).

Since three of the known forms of Lola include zinc finger DNA-binding domains, computational searches of the lola region of the D. melanogaster genome sequence were performed, looking for additional potential zinc fingers. A variety of search motifs were used, including motifs based on a consensus of all zinc fingers or on the consensus of zinc fingers found in previously published BTB-containing proteins. Applying these searches to the sequence of the BAC that encompasses lola revealed 32 potential zinc finger sequences in the genomic DNA corresponding to the lola gene. Each of the putative fingers had strong statistical support (P < 10^-10) in at least one search and predicted protein sequences consistent with the zinc finger structure. Each of these putative fingers is embedded within a larger open reading frame. The distribution of the apparent fingers was consistent with the idea that they define 17 distinct molecular species, 15 bearing pairs of zinc fingers and two bearing single fingers (Goeke, 2003).

The expression of all 17 predicted lola 3' sequences was confirmed by performing RT-PCR on mRNA isolated from embryonic, larval and adult animals, using a forward primer from the common region and reverse primers from the putative finger containing exons. Expression of different combinations of lola isoforms was detected at different developmental stages. All isoforms except for isoform O are expressed in embryos; larvae express all isoforms, and adult males express all except isoforms C and R. Adult virgin females express all isoforms except O (Goeke, 2003).

By sequencing the RT-PCR products, it was found, in all cases, that a single finger or pair of fingers is spliced back to the common region. Eleven lola splice variants have the lola common region spliced directly to a unique final exon containing a pair of zinc fingers, whereas two isoforms have a single zinc finger. Four isoforms have the 3' portion split by an additional intron. No evidence was found for the splicing of multiple finger pair exons to make an isoform with more than two fingers. An additional isoform (M), identified from the D. melanogaster expressed sequence tag (EST) database, splices to 3' exons that do not contain zinc fingers. Remarkably, M and the finger-containing isoform N share a small interstitial exon before diverging by splices to different final exons. All isoforms include the BTB dimerization domain at their N terminus (Goeke, 2003).

The functional significance of lola gene structure was investigated by analyzing its conservation in the mosquito, Anopheles gambiae, which diverged from Drosophila ~250 million years ago. The A. gambiae genome was searched by BLAST searches with the sequence of the BTB domain and all 17 zinc-finger moieties from lola in D. melanogaster. Eight of the seventeen D. melanogaster zinc finger domains clearly had single best matches in the A. gambiae genome, as identified by high percent identity and similarity. These eight matches identified seven unique zinc finger units in A. gambiae that all mapped to a small region on a single genomic scaffold (GenBank accession AAAB008900.1, gi 19611997). Two of the D. melanogaster zinc finger sequences, K and T (which are 92% identical), identified the same zinc finger unit in mosquito. The same genomic region was also selected by BLAST searches with the lola BTB domain. These seven putative finger moieties, together with the BTB domain, identify a likely lola ortholog in A. gambiae. Several details of lola genomic organization were conserved between Anopheles and Drosophila, including the presence of an intron within the sequence encoding the first finger of isoform I (A. gambiae finger sequence 4), the existence of a single-finger isoform (isoform E and A. gambiae finger sequence 2), as well as conservation of the overall order of orthologous zinc fingers along their respective chromosomes. Analysis of the lola locus in A. gambiae also identified two lola-related finger sequences that did not have obvious closest relatives among D. melanogaster lola finger exons (Goeke, 2003).

Antisense RNA probes were generated from the lola RT-PCR products and they were used to assay the expression patterns of the embryonic lola isoforms. The probes excluded the finger sequences themselves to reduce the danger of cross-hybridization; no other portions of the 3' exons showed significant sequence identity in pairwise comparisons. The 19 probes revealed a wide variety of expression patterns for different lola isoforms. Many probes are expressed in broad, overlapping patterns, including expression in whole germ layers or even throughout the embryo. In contrast, some probes show strong enrichment in small sets of cells, such as gonad, invaginating tracheal pits, imaginal discs, a dorsal cell layer of the CNS or isolated cells in the neuroectoderm (Goeke, 2003).

Whereas most lola alleles, such as lola^ORC46 and lola^ORE76, knock out all known lola functions in parallel, several alleles disrupt specific subsets of lola-dependent guidance decisions. These have been termed 'decision-selective alleles'. For example, lola^ORC50 disrupts several CNS functions of lola but not the development of the ISN_b peripheral nerve, whereas lola^ORC4, lola^ORE50, and lola^ORE119 largely block development of ISN_b but have only mild effects on CNS patterning. The defects in decision-selective alleles do not follow any simple allelic series of overall phenotypic strength, but are more readily consistent with the hypothesis that these alleles specifically disrupt separate activities of lola (Goeke, 2003).

Sequencing the terminal lola splice variant exons in flies bearing decision-selective alleles identified molecular lesions in single lola splice variants in three cases: lola^ORE119, lola^ORC4 and lola^ORE50. These have been termed 'single-isoform mutants.' In lola^ORC4 and lola^ORE50, mutations have been identified in the same isoform (K). In lola^ORC4, a (C-->T) mutation introduced a stop codon at amino acid 771 (out of 970 residues in the predicted protein), and consistent with this, Western analysis of extract from lola^ORC4 homozygous embryos using anti-Lola antibodies verified that one of the largest Lola isoforms was absent, but a new, smaller protein species appeared. No change was observed in the level of any other Lola protein species in this experiment. In lola^ORE50, a 4-bp deletion throws the same isoform out of frame at residue 776. In both cases, the mutation is predicted to delete the zinc fingers from the protein, and is therefore expected to be null for this isoform. Consistent with sequencing data, the ISN_b phenotypes of lola^ORC4 and lola^ORE50 homozygous embryos were found to be quantitatively indistinguishable, and were also identical to the phenotype of heteroallelic embryos that are lola^ORC4/ORE50. Since the ISN_b motor nerve fails to innervate its target muscles in lola^ORC4 and lola^ORE50, and isoform K is expressed preferentially in the CNS, the simplest hypothesis is that this isoform acts cell-autonomously in motor neurons, though a contribution cannot be ruled out from low-level epithelial staining of probe K that was observe in some experiments (Goeke, 2003).

In lola^ORE119, a single base pair change in isoform L leads to a single amino acid substitution: Pro712 was changed to a leucine in the sequence between the two fingers. Nine of the 15 zinc finger pairs contain prolines at the analogous position, and moreover, a proline at this position is common in proteins containing multiple zinc fingers. Mutation of the corresponding prolines in TFIIIA and ADR1 reduces DNA binding affinity by 95%. Together, these facts suggest that the mutation in isoform L would likely knock out its function. lola^ORE119 disrupts ISN_b innervation of target muscles, as do lola^ORC4 and lola^ORE50, but in this case probably reflecting the expression of isoform L in mesoderm. The basis of the modest CNS defects in lola^ORE119 are not yet clear; perhaps they arise from a function of isoform L in the mesoderm-derived CNS midline cells. Alternatively, low-level CNS expression is observed of isoform L and it is possible that this contributes to one or both of the phenotypes of lola^ORE119. As expected, Western analysis did not detect a change in the level of any Lola protein species in embryos homozygous for the lola^ORE119 missense mutation. Surprisingly, lola^ORC4 (a mutation inactivating isoform K) did not fully complement the isoform L mutation, lola^ORE119 (Goeke, 2003).

The observation that three single-isoform mutants produce decision-selective mutant phenotypes provides direct evidence that different lola isoforms make unique contributions to specific lola functions. Mutations that inactivate isoform K (lola^ORC4 and lola^ORE50) largely block muscle innervation by ISN_b motoneurons, whereas a mutation inactivating isoform L (lola^ORE119) disrupts both muscle innervation and, to a lesser extent, CNS development. Isoforms K and L differ in sequence, with 27 differences among 55 amino acids in the zinc fingers. The expression patterns of isoforms K and L also differ, with K expressed mainly in the CNS (but with a low level of epithelial expression) and L expressed mainly in mesoderm (but with a low level of CNS expression). It will be interesting to determine how much of the functional difference between isoforms K and L arises from the difference in their expression patterns, and how much arises from the difference in protein sequence (Goeke, 2003).

The data show that some lola isoforms have unique functions. Other experiments demonstrate that some lola functions require combinations of different isoforms. The phenotypes of mutations in isoforms K and L establish that each makes unique contributions to motoneuron pathfinding. However, it is striking that both are required for the same process: muscle innervation by the ISN_b motor nerve. Evidently, innervation of these muscles requires cooperation between at least these two lola isoforms. The need for functional cooperation between these isoforms is driven home by the observation that embryos that are doubly heterozygous for mutations in isoforms K and L display highly penetrant defects in ISN_b development (lola^ORC4/ORE119), even though each mutation by itself is recessive. Trans-heterozygous mutant phenotypes have often been observed for genes whose products function together closely. Since it is not possible to rule out some coexpression of K and L, it may be that their synthetic phenotype arises from the requirement for a K/L heterodimer. Alternately, it may be that these two isoforms regulate the expression of, respectively, a receptor and its ligand that function together in ISN_b development, much as lola co-regulates robo and slit in the CNS. The trans-heterozygous interaction of K and L mutations in this case could reflect their concerted effect on the expression of these target genes (Goeke, 2003).

The data above demonstrate that the diversity of lola isoforms contributes to the complex pattern of lola functions. Yet more diversity could potentially arise from heterodimerization. Taking into account overlap in expression patterns, if the Lola isoforms dimerize with each other, then association of different isoforms could produce over 100 distinct combinations of zinc finger DNA-binding domains. Moreover, the BTB domain is highly conserved among BTB proteins, and there are indications of heterodimerization between lola and other BTB-containing proteins. In addition, multiple splice variants have been identified for some of these potential lola partners. If lola dimerization partners show as much alternate splicing as does lola, the potential diversity of Lola-containing species would increase geometrically (Goeke, 2003).

With the sequencing of various complete genomes, a common problem has emerged: understanding how functional diversity is generated from a compact genome. The role of alternative splicing in increasing the diversity of cell surface proteins is well established. These data extend this principle to gene regulatory proteins, and it is suggested that alternative splicing may be used by transcription factors as extensively as it is by cell surface proteins as a means to increase the overall functional diversity of the genome (Goeke, 2003).

GENE STRUCTURE

lola codes for three RNA transcripts of 3.8, 4.7 and 4.9 kb sharing a common core sequence with different 5' and 3' ends. Both 3.8 and 4.7 kb transcripts have a 399 base pair 5' UTR whose sequences differ in the first 178 bases. The 3.8 kb transcript has a 180 base pair 5' UTR. The 4.7 kb transcript has a 750 base pair 5' UTR. The proteins are identical up to the 454th amino acid. The long form continues from there, while the short form has 13 amino acids that differ from the long form.

PROTEIN STRUCTURE

Amino Acids - 467 for the 3.8 kb RNA and 894 for the 4.7 kb RNA

Structural Domains and Evolutionary Homologs

The long transcript has a Cys2-His2 type of double zinc finger. The short transcript has no zinc finger. Both transcripts have a BTB domain (Giniger, 1994).

A novel zinc finger protein, ZID (standing for zinc finger protein with interaction domain) was isolated from humans. ZID has four zinc finger domains and a BTB domain, also know ans a POZ (standing for poxvirus and zinc finger) domain. At its amino terminus, ZID contains the conserved POZ or BTB motif present in a large family of proteins that include otherwise unrelated zinc fingers, such as Drosophila Abrupt, Bric-a-brac, Broad complex, Fruitless, Longitudinals lacking, Pipsqueak, Tramtrack, and Trithorax-like (GAGA). The POZ domains of ZID, TTK and TRL act to inhibit the interaction of their associated finger regions with DNA. This inhibitory effect is not dependent on interactions with other proteins and does not appear dependent on specific interactions between the POZ domain and the zinc finger region. The POZ domain acts as a specific protein-protein interaction domain: The POZ domains of ZID and TTK can interact with themselves but not with each other, or POZ domains from ZF5, or the viral protein SalF17R. However, the POZ domain of TRL can interact efficiently with the POZ domain of TTK. In transfection experiments, the ZID POZ domain inhibits DNA binding in NIH-3T3 cells and appears to localize the protein to discrete regions of the nucleus (Bardwell, 1994).

Transcriptional Regulation

The wide-ranging defects in dendrites and axons indicate that sequoia functions to regulate axonal and dendritic morphogenesis in most neurons. Alternatively, it is conceivable that sequoia regulates the expression of genes generally required for neuronal differentiation. To gain mechanistic insight into sequoia function, the transcript profiles in wild-type and sequoia mutant embryos were compared based on microarray analyses of over 3,000 genes or ESTs, corresponding to about 25% of the Drosophila genome. The vast majority of these genes show comparable expression levels, including genes for cytoskeletal elements, genes that specify neuronal cell fates, and genes generally required for neurite outgrowth such as cdc42. Interestingly, a small fraction of the genes/ESTs analyzed showed clearly distinct expression ratios in sequoia mutants. Of these, 93 (3.1%) different transcripts were reduced by at least one-third of the wild-type level, and 34 (1.1%) different transcripts were increased by at least 75% of the wild-type level. A number of genes that appear to be regulated by sequoia, directly or indirectly, correspond to genes implicated in the control of axon morphogenesis rather than neuronal fate. These include known genes such as connectin, frazzled, roundabout 2, and longitudinals lacking, in addition to novel molecules with homology to axon guidance molecules including slit/kekkon-1 and neuropilin-2. It is noteworthy that two of the genes showing increased transcript ratios, roundabout 2 and CG1435, a novel calcium binding protein, were both also identified in a gain-of-function screen affecting motor axon guidance and synaptogenesis. In addition to genes that have clearly been implicated in axon development based on previous studies or sequence similarity, microarray data reveal that other genes potentially regulated by sequoia include peptidases, lipases, and transporters, as well as novel zinc finger proteins. It should be noted that transcripts that are broadly expressed and increased or decreased in sequoia mutants may actually be altered to a greater extent within neurons, because sequoia likely functions cell autonomously and is only expressed in the nervous system (Brenman, 2001).

Targets of Activity

By transfection experiments, a 72-bp enhancer sequence has been identified within the Drosophila copia retrotransposon, which is involved in the control of the transcription level of this mobile element in cells in culture. Gel shift assays with nuclear extracts from Drosophila hydei-derived DH-33 cells demonstrate specific interactions of at least two nuclear factors with this enhancer sequence. Using this sequence as a probe for the screening of an expression cDNA library constructed from DH-33 cells RNA, a cDNA clone encoding a 110-kDa protein was isolated with features common to those of known transcription factors: these include a two-zinc-finger motif at the C terminus; three glutamine-rich domains in the presumptive activation domain of the protein, and an N-terminal domain that shares homology with the Bric-a-brac, Tramtrack, and Broad-Complex BTB boxes. The precise DNA recognition sequence for this transcription factor has been determined by both gel shift assays and footprinting experiments with a recombinant protein made in bacteria. The functionality of the cloned element has been demonstrated upon transcriptional activation of copia reporter genes, as well as of a minimal promoter coupled with the identified target DNA sequence, in cotransfection assays in cells in culture with an expression vector for the cloned factor. Southern blot and nucleotide sequence analyses reveal a related gene in Drosophila melanogaster (the lola gene) previously identified by a genetic approach as involved in axon growth and guidance. Transfection assays in cells in culture with lola gene expression vectors and in situ hybridization experiments with lola gene mutants finally provided evidence that the copia retrotransposon is regulated by this neurogenic gene in D. melanogaster, with a repressor effect in the central nervous systems of the embryos (Cavarec, 1997).

The pattern and level of expression of axon guidance proteins must be choreographed with exquisite precision for the nervous system to develop its proper connectivity. Previous work has shown that the transcription factor Lola is required for central nervous system (CNS) axons of Drosophila to extend longitudinally. Lola is simultaneously required to repel these same longitudinal axons away from the midline, and it acts, in part, by augmenting the expression of both the midline repellant, Slit, and its axonal receptor, Robo. Lola is thus the examplar of a class of axon guidance molecules that control axon patterning by coordinating the regulation of multiple, independent guidance genes, ensuring that they are co-expressed at the correct time, place and relative level (Crowner, 2002).

The reduction of Robo expression seen in lola mutants is relatively modest (~40%). It is known, however, that a 50% diminution in Robo is sufficient by itself to cause some inappropriate midline crossing, and this effect is strongly enhanced by a simultaneous 50% reduction in Slit. Loss of lola causes a greater reduction than this in Slit levels. Thus, it is plausible that the change in Slit and Robo levels could account for much of the midline phenotype observed in embryos that bear strong lola mutations. But why are weaker lola alleles like lola^1A4 able to cause extra midline crossing when their effect on target gene expression is presumably proportionately less? It is likely that regulation of Slit and Robo expression is only one part of the control of midline crossing by lola, and that a significant contribution to the phenotype is made by changes in the expression of other, interacting guidance genes that are also controlled by lola. For example, aspects of the lola midline phenotype resemble details of the axon pattern observed upon mutation of genes encoding receptor tyrosine phosphatases, suggesting that these are good candidates for potential lola effectors. Moreover, it is known that the Notch-dependent mechanism that promotes the alternative (longitudinal) trajectory of CNS axons also requires lola. The multiplicity of genes contributing to the midline/longitudinal axon growth decision underscores the need for a gene, like lola, to coordinate the expression of all these cooperating guidance factors. It is suggested that it is the combination of many quantitative effects, each individually modest, which together produce the profound effects of lola on axon patterning (Crowner, 2002).

Many questions remain from these studies. (1) Though Lola itself is a transcriptional regulator, it is not known whether robo and slit are direct Lola targets or whether Lola initiates a longer chain of events leading only indirectly to robo and slit. For example, Lola could regulate other genes that themselves control the stability of robo or slit RNA or protein, or the splicing or translation efficiency of these genes. Analysis of this issue will require unambiguous identification of the exact lola isoforms required for expression of robo and slit, and characterization of their DNA-binding specificities in combination with their appropriate dimerization partner(s). (2) Only the accumulation of Robo and Slit protein has been characterized in lola mutants, and not transcript levels. The inherent variability of whole-mount RNA in situ hybridization has prevented sufficiently precise quantification of robo and slit RNA levels for this purpose. Nonetheless, the observation that ectopic expression of lola 4.7 leads to ectopic expression of slit RNA strongly argues that lola is upstream of slit transcription, though it remains possible that Robo and Slit expression are also subject to lola-dependent regulation at some post-transcriptional level (Crowner, 2002).

lola does not just regulate midline crossing, but also controls extension of some peripheral motor axons and orientation of lateral chordotonal neurons in the embryo, as well as pathfinding of some axons of the adult wing. In each case, it apparently establishes a precise balance of guidance factors. How can one transcription factor exert such subtle control over such a diverse array of developmental events? This remains to be determined, but it has recently been found that lola encodes a large number of protein isoforms. At least in some cases, lola isoforms with different predicted DNA binding specificities are expressed in different tissue specific patterns, potentially allowing the regulation of distinct cohorts of downstream target genes. Moreover, it is known that a single, direct Lola target gene can be activated by lola in one tissue and repressed in another . Both of these properties are likely to contribute to the ability of Lola to modulate gene expression programs in distinct ways in different cells (Crowner, 2002).

The problem of ensuring appropriate relative levels of multiple guidance genes is not unique to the midline crossing versus longitudinal growth decision in the fly CNS, but rather is an inherent feature of all axon guidance decisions. It is therefore imagined that lola is not unique in its property of co-regulating multiple, interacting guidance genes, but rather is the exemplar of a class of transcriptional regulators that will be found to be widespread in distribution and critical in importance in the regulation of axon patterning (Crowner, 2002).

Alternative trans-splicing of constant and variable exons of lola

longitudinals lacking (lola) is a complex Drosophila gene encoding at least 20 protein isoforms, each bearing the same N-terminal constant region linked to a different C-terminal variable region. Different isoforms specify different aspects of axon growth and guidance. lola mRNAs are generated by alternative trans-splicing of exons sequentially encoded by the same DNA strand. Chromosomal pairing facilitates interallelic trans-splicing, allowing complementation between mutations in the constant and those in the variable exons. At least one variable exon is transcribed from its own promoter and is then trans-spliced to the constant exons transcribed separately (Horiuchi, 2003).

To dissect the genetic properties of the lola locus, interallelic complementation tests were performed among five ethylmethane sulfonate (EMS)-induced mutant alleles, lola^ORE120, lola^ORC46, lola^ORC4, lola^ORE50, and lola^ORE119, whose molecular lesions are already identified. Exons C5-8 constitute the constant region for all isoforms, and V10-11, V21-22, V23, and V32 correspond to the variable region of isoforms B, K, L, and T, respectively. All are homozygous lethal with severe defects in embryonic axon guidance. Flies die that are heterozygous for alleles mutated in the same exon unit (lola^ORE120/lola^ORC46 for C5-8 and lola^ORE50/lola^ORC4 for V21-22). In contrast, animals heterozygous for mutations in exons V21-22 and V23 (lola^ORE50/lola^ORE119 and lola^ORC4/lola^ORE119) show nearly normal viability. The interallelic complementation between alleles mutated in different variable exons is reasonable, since a functional version of each isoform is supplied by one of the homologous chromosomes. Surprisingly, however, combinations of mutations in the constant exons with those in the variable exons (lola^ORC46/lola^ORE50, lola^ORE120/lola^ORE119, and so on) also result in viable flies (Horiuchi, 2003).

To explain these unexpected results, three models were postulated: (1) Functional mRNAs are produced through interallelic trans-splicing or any other mechanisms recombining two pre-mRNAs; (2) a wild-type locus is generated through recombination or gene conversion, or (3) functional complementation occurs between the two mutant proteins. Model 3 is rather unlikely, because lola^ORC46 contains a nonsense mutation within a constant exon, so that its mRNA would be degraded quickly via nonsense-mediated mRNA decay; even if it could survive, the translated protein lacks the entire variable region. In models 1 and 2, wild-type mRNA variants should be present in the hybrid animals. To examine this possibility, the sequences were determined of C5-8/V21-22 and C5-8/V23 cDNAs amplified by reverse transcriptase PCR (RT-PCR) from total RNA fractions of flies of the genotype lola^ORE120/lola^ORC4 and lola^ORC46/lola^ORE119, respectively. Indeed, C5-8/V21-22 and C5-8/V23 cDNA clones were found with no molecular lesions. In model 2, DNA recombination or gene conversion at the lola locus should occur at an extraordinarily high frequency at a very early stage of development. Whether it happened in the germ-line cells, which start dividing at a very early stage, was examined. lola^ORE120/lola^ORC4 and lola^ORC46/lola^ORE119 flies were crossed to Df(2R)stan2, a deficiency uncovering the entire lola region. None of the hemizygous offspring were viable, suggesting that DNA recombination or gene conversion at the lola locus is unlikely. Therefore, model 1 seemed the most likely mechanism to explain the interallelic complementation between mutations in the constant and those in the variable regions and the presence of chimeric mRNAs. In all sequenced clones including those described below, joining of exons occurs invariably at consensus splicing sites, suggesting that recombinant mRNAs are generated through trans-splicing, rather than any other mechanism (Horiuchi, 2003).

It is important to note that the variable exons (V23 and V32) assort independently of the constant exons; they are trans-spliced to the constant exons derived from either homologous chromosome at nearly equal frequency. This suggests that the splicing must occur when the pre-mRNAs from the two homologous chromosomes are in close proximity. In Drosophila and other dipteran insects, genes on homologous chromosomes are thought to be physically proximal to each other through pairing. Transvection and trans-sensing effects are severely affected when homologous chromosome pairing is disrupted by chromosomal rearrangements. To examine the influence of chromosomal rearrangements on the frequency of interallelic trans-splicing in the lola locus, SNP analyses of F1 hybrids was performed with a CKG30 balancer chromosome with multiple inversions, and those with T(2;3)X3, a translocation involving the lola locus. The frequency of cDNA clones with interallelic hybrid haplotypes was significantly reduced by CKG30 and by T(2;3)X3. These results clearly indicate that interallelic trans-splicing occurs frequently when the two chromosomal loci are positioned such that they can pair effectively (Horiuchi, 2003).

In summary, therefore, alternative trans-splicing at the constant/variable junction generates a significant portion of the mRNA from the lola locus. For some isoforms, nearly half of the mature RNAs were chimeras between pre-mRNAs derived from the homologous chromosomes, indicating that the mRNAs were generated through trans-splicing exclusively. Finally, in one case, a variable exon was found being expressed from a dedicated promoter in the preceding intron. lola provides the first demonstration of abundant mRNAs generated by alternative trans-splicing of exons sequentially encoded by the same DNA strand. The frequent interallelic trans-splicing events support the pairing of homologous chromosomes in Drosophila. In contrast, mammalian hybrid alpha/gamma protocadherin mRNAs, which are most likely generated by trans-splicing, can be spliced only intrachromosomally. This difference may be due to a lack of chromosomal pairing during transcription in mammalian cells, and/or a high competence of Drosophila pre-mRNAs to be trans-spliced (Horiuchi, 2003).

There are important implications regarding the mechanism of trans-splicing. Because the efficiency of interchromosomal trans-splicing was dramatically reduced by chromosomal rearrangements, it is inferred that trans-splicing occurs locally either cotranscriptionally or posttranscriptionally. Cotranscriptional trans-splicing is consistent with a recent view of gene expression in which all steps of mRNA processing take place cotranscriptionally. It is also supported by reported examples of trans-splicing events detected in mammalian cells in which trans-splicing occurred between pre-mRNAs derived from the same genes or closely linked ones. Trans-splicing of mod(mdg4) pre-mRNAs occurs between those derived from the original locus and a transgene inserted in a different chromosome. However, this may not represent the natural situation, because a high level of transgene expression was induced by the GAL4-UAS system, and RT-PCR not controlled for template switching was used to detect trans-spliced mRNAs. Therefore, it may be reasonable to assume that natural trans-splicing occurs locally, most likely cotranscriptionally, as is the case for cis-splicing. An additional requirement for trans-splicing would be a mechanism allowing integration of pre-mRNAs encoding variable exons into the spliceosome operating in the constant region. The outron sequences (transcribed region preceding the variable exons) may contain cis-elements required for this process (Horiuchi, 2003).

In addition, variable regions may consist of multiple independent transcription units whose pre-mRNAs are each trans-spliced to the constant exons. With internal promoter(s) for variable exons and cotranscriptional alternative trans-splicing, the regulation of lola isoform expression could be much simplified. Unlike alternative cis-splicing, simultaneous expression of multiple isoforms can be achieved without complex regulation of splicing and polyadenylation sites of a single mRNA. Multiple pre-mRNAs containing different variable exons could be expressed in the same cells depending on their own promoter activities, which would then be trans-spliced to the constant exons, possibly in a concentration-dependent manner. Furthermore, the internal promoter(s) for variable exons may be tissue-specific, and thus it could also simplify the mechanism of tissue-specific expression of particular variants as well (Horiuchi, 2003).

lola shares many molecular features with mod(mdg4), and it is therefore possible that mod(mdg4) isoforms sequentially encoded in the same DNA strand could also be generated through trans-splicing. Furthermore, there are other fly genes encoding BTB-Zn finger proteins, such as fruitless and Br-C, some of which show complex patterns of intragenic complementation reminiscent of what is seen with lola. These genes produce multiple splice variants, often with a complex pattern of expression. Trans-splicing may be a general means to generate splice variants in this family of genes (Horiuchi, 2003).

DEVELOPMENTAL BIOLOGY

Embryonic

All three RNA transcripts of lola are present maternally, although the 4.9 kb RNA disappears rapidly after fertilization. The 3.8 kb RNA continues to be present throughout embryogenesis, whereas the 4.7 kb transcript decays at about 9 hours, at which time the 4.9 kb RNA reappears. The 4.7 kb isoform preferentially appears in mesodermal and mesectodermal cells after germ band extension. The mesectodermal cells are probably glial precursors. Label associated with the short form concentrates in neural tissue between stage 11 and 13, concommitant with the reappearance of the 4.9 kb form.

RNA detectable with the long probe appears in clusters of peripheral cells that are probably precursors of imaginal discs and abdominal histoblasts and tracheal histoblasts and in a small sector of the brain, possibly the optic lobe anlagen (Giniger, 1994).

Effects of mutation or deletion

lola was isolated from a screening for disrupted axons of 550 homozygotic lethal insertions of a transposon. Disrupting genes with exogenous DNA is a convenient method to search for and manipulate new genes. Mutation of lola leads to disrupted axon tracts in the central nervous system, and displacement of chordotonal organs (Giniger, 1994).

Epigenetic silencers Lola and Pipsqueak collaborate with Notch to promote malignant tumours by Rb silencing

Cancer is both a genetic and an epigenetic disease. Inactivation of tumour-suppressor genes by epigenetic changes is frequently observed in human cancers, particularly as a result of the modifications of histones and DNA methylation. It is therefore important to understand how these damaging changes might come about. By studying tumorigenesis in the Drosophila eye, two Polycomb group epigenetic silencers, Pipsqueak and Lola, have been identified that participate in this process. When coupled with overexpression of Delta, deregulation of the expression of Pipsqueak and Lola induces the formation of metastatic tumours. This phenotype depends on the histone-modifying enzymes Rpd3 (a histone deacetylase), Su(var)3-9 and E(z), as well as on the chromodomain protein Polycomb. Expression of the gene Retinoblastoma-family protein (Rbf ) is downregulated in these tumours and, indeed, this downregulation is associated with DNA hypermethylation. Together, these results establish a mechanism that links the Notch-Delta pathway, epigenetic silencing pathways and cell-cycle control in the process of tumorigenesis (Ferres-Marco, 2006).

Correct organ formation depends on the balanced activation of conserved developmental signalling pathways (such as the Wnt, Hedgehog and Notch pathways). If insufficient signals are received, organ growth may be deficient. By contrast, excess signalling leads to an overproduction of progenitor cells and a propensity to develop tumours. When such hyperproliferation is associated with the capacity of cells to invade surrounding tissue and metastasis to distant organs, cancer develops. Indeed, activation of the Wnt, Hedgehog and Notch pathways is a common clinical occurrence in cancers. Curiously, activation of any of these pathways in animal models seems to be insufficient for cancer to develop, indicating that synergism with other genes is required for these pathways to produce cancer (Ferres-Marco, 2006).

Cellular memory or the epigenetic inheritance of transcription patterns has also been implicated in the control of cell proliferation during development, as well as in stem-cell renewal and cancer. Proteins of the Polycomb group (PcG) are part of the memory machinery and maintain transcriptional repression patterns. The upregulation of several PcG proteins has been associated with invasive cancers. Thus, increased amounts of EZH2, the human homologue of the Drosophila histone methyltransferase E(z), is associated with poorer prognoses of breast and prostate cancers (Ferres-Marco, 2006).

Another histone methyltransferase implicated both in gene silencing and in cancer is SUV39H1, a homologue of Drosophila Su(var)3-9. SUV39H1 and Su(var)3-9 methylate histone H3 on lysine 9 (H3K9me), and this epigenetic tag is characteristic of heterochromatin and DNA sequences that are constitutively methylated in normal cells. DNA methylation is another mechanism involved in cellular memory that actively contributes to cancer. Indeed, numerous tumour-suppressor genes, including the retinoblastoma gene RB, are silenced in cancer cells by DNA hypermethylation. Inactivation of the RB tumour-suppressor pathway is considered an important step towards malignancy; thus, it is important to understand how these damaging epigenetic changes are initiated in cells that become precursors of cancer. Moreover, it is equally important to determine the connection between these processes and the developmental pathways controlling proliferation (Ferres-Marco, 2006).

Forward genetic screening in suitable animals is a powerful tool with which to identify tumour-inducing genes and to reveal changes that precede neoplastic events in vivo. The developing eye of Drosophila melanogaster is a good model for such studies because it is a simple and genetically well-defined organ. The growth of the eye depends on Notch activation in the dorsal-ventral organizer by its ligands Delta (human counterparts, DLL-1, -3, -4) and Serrate (human counterparts, JAGGED-1, -2). This study used the 'large eye' phenotype, produced by overexpression of Delta, as a tool to screen for mutations that interact with the Notch pathway and convert tissue overgrowths into tumours. One mutation, eyeful, was isolated that combined with Delta induces metastatic tumours. eyeful forces the transcription of two hitherto unsuspected growth and epigenetic genes, lola and pipsqueak (psq). The identification of eyeful has been a starting point from which to unravel crosstalk between the Notch and epigenetic pathways in growth control and tumorigenesis. The fact that many epigenetic factors are involved in cancer suggests that these processes may be more generally involved in tumorigenesis than at first it might seem (Ferres-Marco, 2006).

To identify genes that interact with the Notch pathway and that influence growth and tumorigenesis, the Gene Search (GS) system was used to screen for genes that provoked tumours when coexpressed with Delta in the proliferating Drosophila eye. The ey-Gal4 line was used for both eye-specific and ubiquitous induction, resulting in the transactivation of UAS-linked genes throughout the proliferating eye discs. It was through such a screen that the GS88A8 line was isolated. Generalized overexpression of Delta by ey-Gal4 (hereafter termed ey-Gal4 > Dl) produces mild eye overgrowth. In most of the flies in which the GS88A8 line was coexpressed with Delta, tumours developed in the eyes. Moreover, in ~30% of the mutant flies, secondary eye growths were observed throughout the body, and in some flies the whole body filled up with eye tissue. These secondary eye growths had ragged borders, indicating invasion of the mutant tissue into the surrounding normal tissue. As a result, the GS88A8 line was named 'eyeful' (Ferres-Marco, 2006).

A developmental analysis of the tumours was undertaken. To facilitate analyses, a triple mutant strain was generated carrying the eyeful, UAS-Dl and ey-Gal4 transgenes all on the same chromosome (ey-Gal4 > eyeful > Dl. In this strain, mutant eye discs showed massive uncontrolled overgrowth (some discs were more than five times their normal size). In most discs, the epithelial cells had lost their apical-basal polarity, and some had a disrupted basement membrane and grew without differentiating (Ferres-Marco, 2006).

These results were extended to the wing disc. (1) dpp-Gal4 was used to direct coexpression of eyeful and Delta along the anterior-posterior boundary of the wing (perpendicular to the endogenous Delta domain along the dorsal-ventral boundary. In a normal wing disc, the dpp-Gal4 driver typically establishes a stripe of green fluorescent protein (GFP) expression with a sharp border at the boundary. Whereas wild-type (or single eyeful) cells expressing GFP conformed with this pattern, some of the eyeful and Delta cells were found outside this stripe, indicating that the mutant cells can disseminate and invade adjacent regions of the disc. (2) The MS1096-Gal4 line was used to direct expression in the dorsal wing disc compartment. Under these conditions, the wing tissue grew massively and aggressively, and the mutant tissue failed to differentiate. These observations suggest that, when coupled with Delta overexpression, an excess of the gene products flanking the eyeful insertion site induces the formation of tumours capable of metastasising (Ferres-Marco, 2006).

The genomic DNA flanking the eyeful P-element was isolated and sequenced. eyeful is inserted in an intron of the gene longitudinals lacking (lola), which is known to be a chief regulator of axon guidance. lola encodes 25 messenger RNAs that are produced by alternative splicing and that generate 19 different transcription factors. All of the different isoforms share four exons that encode a common amino terminus, which contains a BTB or POZ domain. In addition, all but one of these transcription factors are spliced to unique exons encoding one or a pair of zinc-finger motifs (Ferres-Marco, 2006).

The GS P-elements allow Gal4-dependent inducible expression of sequences flanking the insertion site in both directions. The nearest gene in the opposite direction to transcription of lola is the psq gene. This gene encodes nine variants produced by alternative splicing and alternative promoter use, generating four different proteins. Three of the psq isoforms contain a BTB or POZ domain in the N terminus, and a histidine- and glutamine-rich region downstream of this domain. Two of the BTB-containing isoforms and the isoform that lacks this domain contain four tandem copies of an evolutionarily conserved DNA-binding motif, the Psq helix-turn-helix (HTH) motif (Ferres-Marco, 2006).

psq was initially identified for its 'grandchildless' and posterior group defects and was subsequently shown to have a role in retinal cell fate determination. Psq is essential for sequence-specific targeting of a PcG complex that contains histone deacetylase (HDAC) activity. Psq binds to the GAGA sequence, which is present in many Hox genes and in hundreds of other chromosomal sites (Ferres-Marco, 2006).

Both polymerase chain reaction with reverse transcription (RT-PCR) and in situ hybridization experiments confirmed that transcription of lola and psq was influenced by eyeful in response to Gal4 activation (Ferres-Marco, 2006).

To determine whether lola and/or psq was responsible for the tumour phenotype, 11 enhancer promoter (EP) P-elements inserted into the lola and psq region were tested. In contrast to the GS lines, the EP lines allows Gal4-dependent inducible expression of sequences flanking only one end of the P-element. It was found that none of the EP lines induced tumours; thus, it was reasoned that the deregulation of both genes might be required to produce the tumours (Ferres-Marco, 2006).

The complexity of lola and psq loci, which together produce 23 proteins, hampers identification of the transcripts responsible for the eyeful phenotype by gain-of-expression mutants (that is, by expressing individual or combinations of isoforms). Therefore, this issue was resolved by isolating point mutations that reverted the phenotype caused by deregulated expression of lola and psq. In this analysis, the chemical mutagen ethyl-methane sulphonate (EMS) was used to induce preferentially single nucleotide changes (Ferres-Marco, 2006).

The parental eyeful GS line was viable in trans with deficiencies that removed both lola and psq. In contrast, a set of 14 EMS-induced mutations on the eyeful chromosome failed to complement these deficiencies and were found to be alleles of psq or lola. The EMS-induced mutations that best recovered a normal eye size were sequenced. Each individual mutation had a single base change or a small deletion that considerably altered the predicted Psq or Lola proteins (Ferres-Marco, 2006).

All psq^- mutations induced on the eyeful chromosome prevented eyeful from producing eye tumours and metastases. Three alleles affected the BTB domain (psq^rev2, psq^rev7 and psq^rev9), and three other alleles contained either a premature stop codon that would produce truncated proteins lacking the Psq HTH repeats (psq^rev4 and psq^rev14) or a missense mutation that would change a conserved amino acid in the third Psq HTH repeat (psq^rev12). All lola^- mutations induced on the eyeful chromosome, including the presumptive null allele (lola^rev6), reduced eye tumour size but still permitted sporadic secondary growth (Ferres-Marco, 2006).

These data show an unequal contribution of Psq and Lola in this process, whereby Psq is the most important factor in the tumorigenic phenotype. The BTB subfamily of transcriptional repressors includes the human oncogenes BCL6 and PLZF. In these oncogenes, the BTB domain is crucial for oncogenesis through the recruitment of PcG and HDAC complexes. It is therefore speculated that deregulated Psq and Lola could lead to tumorigenesis by epigenetic processes and that Drosophila counterparts of HDACs and PcG proteins might be involved in the progression of these tumours. Indeed, genetic evidence was found that both Lola and Psq function as epigenetic silencers in vivo (Ferres-Marco, 2006).

Attempts were made to determine the specific epigenetic mechanisms through which deregulation of Psq and Lola might induce tumorigenesis in conjunction with Delta overexpression. Methylation of histone on lysine is a central modification in both epigenetic gene control and in large-scale chromatin structural organization. For example, trimethylation of histone H3 on K4 (H3K4me3) is associated with the active transcription of genes and open chromatin structure. By contrast, histone hypoacetylation and H3K9 and H3K27 methylation are characteristic of heterochromatin state and gene silencing. To determine whether any changes in these epigenetic markers might coincide with the induction of tumorigenesis, eye discs were immunolabelled with antibodies against specific histone H3 modifications. Because dorsal eye disc cells are refractory to Delta, the dorsal region of the discs provided an internal control for these studies (Ferres-Marco, 2006).

With the exception of some scattered cells, a prominent loss or strong reduction of H3K4me3 was observed in the ventral region of the mutant discs. Notably, although the loss of Notch in clones does not affect this epigenetic tag, overexpression of Delta caused a significant reduction in staining for H3K4me3. The H3K4me3 depletion was already apparent in discs showing moderate hyperplasia and thus preceded neoplasm formation. Changes in other epigenetic tags (such H3K9me3 and H3K27me2) could not be reproducibly resolved; perhaps more sensitive methods or antibodies might facilitate detection of such changes (Ferres-Marco, 2006).

H3K4 methylation is thought to be permissive for maintaining and propagating activated chromatin and is thought to neutralize repressor tags by precluding binding of the HDAC complex and impairing SUV39H1-mediated H3K9 methylation. Thus, H3K4me3 depletion may contribute to tumour formation by permitting aberrant chromatin silencing. It was found that a 50% reduction in dosage of the HDAC gene Rpd3 or of Su(var)3-9 decreased the tumour phenotype dominantly. In contrast, reducing the activity of the H3K4 histone methyltransferase genes Trx (known as ALL1 or MLL in humans) or Ash1, which would be expected to deplete the H3K4me3 tag further, did not visibly enhance the tumours (Ferres-Marco, 2006).

E(z) when complexed with the Extra sex combs (Esc) protein becomes a histone methyltransferase. The E(z)-Esc complex and its mammalian counterpart Ezh2-Eed show specificity for H3K27 but may also target H3K9. The complex also contains the HDAC Rpd3, and this association with Rpd3 is conserved in mammals. H3K27 methylation facilitates binding of the chromodomain protein Pc (HPC in humans), which then creates a repressive chromatin state that is a stable silencer of genes (Ferres-Marco, 2006).

Although loss of E(z) does not cause proliferation defects within discs, halving the E(z) gene dosage dominantly suppressed tumorigenesis, indicating that histone methylation by the E(z)-Esc complex is also a prerequisite for the excessive proliferation of these tumours. Accordingly, Esc^- or Pc^- mutations also notably reduced the tumours (Ferres-Marco, 2006).

Together, these findings suggest that the development of these tumours involves, at least in part, changes in the structure of chromatin brought about by covalent modifications of histones. These changes probably switch the target genes from the active H3K4me3 state to a deacetylated H3K9 and H3K27 methylation silent state (Ferres-Marco, 2006).

From this above data, it is considered that the tumours might form as a result of aberrant gene silencing. If so, then the expression of genes involved in cell-cycle control is likely to be altered in the mutant cells. The transcription of 12 tumour-related genes in the mutant and wild-type discs was compared. Transcription of the gene Rbf, a fly homologue of the RB/Rb family of genes, was strongly downregulated in this assay (and even in ey-Gal4 > Dl flies). A second Rb gene, Rbf2, remained unchanged in the different genetic backgrounds, highlighting the specificity of Rbf silencing (Ferres-Marco, 2006).

It was found that Rbf depletion seems to be intricately associated with tumorigenesis: (1) reducing Rbf gene dosage by 50% enhanced tumour growth; (2) re-establishing Rbf expression in the eye (using an UAS-Rbf transgene) consistently prevented eye tumours and occurrence of secondary growths (Ferres-Marco, 2006).

Inactivation of RB1 in retinoblastoma, a form of eye cancer in children, can occur through DNA hypermethylation of the promoter. Unlike in mammals, however, there is little cytosine methylation of the genome in Drosophila during developmental stages, and its potential role during tumorigenesis is unknown. DNA methylation seems to depend on one DNA methyltransferase, Dnmt2, that preferentially methylates cytosine at CpT or CpA sites. The fly genome also encodes one methyl-CpG DNA-binding MBD2/3 protein. Because there are no known Dnmt2 loss-of-function mutations, the role of this gene in tumorigenesis could not be tested (Ferres-Marco, 2006).

Nevertheless, whether the CpG islands that were observed in the Rbf gene were potential targets for repression by DNA methylation was tested by two methods. (1) Methylation-sensitive restriction enzymes analysis was used; this showed that the regions around the promoter and transcription start site of the Rbf gene are susceptible to methylation. This approach showed aberrant DNA hypermethylation of Rbf in eyeful and Delta eye discs and mild hypermethylation in Delta discs; however, at best only very mild methylation was detected in discs from wild-type flies or from flies with the control psq gene (Ferres-Marco, 2006).

(2) Direct bisulphite sequencing of genomic DNA was carried out from mutant discs. This approach confirmed the notable increase in methylated DNA in eyeful and Delta discs when compared with wild-type discs (and a moderate increase in methylated DNA in the Delta discs). Hypermethylation of the Rbf promoter was not simply the result of de novo transcription of Dnmt2 (ey-Gal4 > Dnmt2), indicating that activation of the Notch pathway is a crucial step in this de novo hypermethylation of Rbf (Ferres-Marco, 2006).

This study has used Drosophila genetics to search for genes that collaborate with the Notch pathway during tumorigenesis in vivo. Psq and Lola were identified as decisive factors to foment tumour growth and invasion when coactivated with the Notch pathway. These proteins are presumptive transcription repressors that contain a BTB domain and sequence-specific DNA-binding motifs and behave as epigenetic silencers in vivo (Ferres-Marco, 2006).

In addition, crosstalk between the Notch pathway and different epigenetic regulators was identified. It is likely that alterations in this crosstalk provoke the aberrant epigenetic repression (and perhaps also derepression) of genes that contribute to cellular transformation. The Rbf gene was identified as one target for this epigenetic regulation and Rbf depletion was shown to directly contribute to the tumours (Ferres-Marco, 2006).

It is proposed that the sequence of events that leads to these tumours commences with hyperactivation of the Notch pathway, which initiates gene repression. Subsequently, or at the same time as Notch, Psq-Lola could bind to the silenced genes and enforce silencing by recruitment of HDAC or PcG repressors. Given the conservation of the Psq-like HTH domains in Psq and of BTB domains, it seems likely that other transcriptional repressors containing such domains strongly influence the tumour-inducing capacities of HDACs and PcG repressors in human cancers (Ferres-Marco, 2006).

Finally, the collaboration between PcG-mediated cellular memory and the Notch pathway may have implications in other processes controlled by Notch, including the second mitotic wave in the Drosophila eye, and the organization of eye and wing growth. In these processes, the memory mechanism could ensure that cells keep a record of the Notch signals received at an earlier stage or when the progenitor cells were closer to the Delta source. In this way, they might remain proliferative without having to receive continuous instructions from Notch. Likewise, such a situation could be conceived for tumorigenesis. The oncogenic signals could opportunistically take advantage of the memory mechanism to fix and to maintain their instructions of continuous proliferation in progenitor or stem cells, thereby fostering tumour growth and metastasis (Ferres-Marco, 2006).

Lola regulates cell fate by antagonizing Notch induction in the Drosophila eye

Lola is a transcription repressor that regulates axon guidance in the developing embryonic nervous system of Drosophila. Lola regulates two binary cell fate decisions guided by Notch inductive signaling in the developing eye: the R3-R4 and the R7-cone cell fate choices. Lola is required cell-autonomously in R3 for its specification, and Lola transforms R4 into R3 if overexpressed. Lola also promotes R7 fate at the expense of cone cell fate. Lola antagonizes Notch-dependent gene expression and Notch-dependent fate transformation. Expression analysis shows that Lola is constitutively present in all photoreceptors and cone cells. It is proposed that when a precursor cell receives a weak Notch inductive signal, it is not sufficiently strong to overcome the constitutive repression of target gene transcription by Lola. A precursor that receives a strong Notch inductive signal expresses target genes despite a constitutive repression by Lola. The predicted consequence of this mechanism is to sharpen a cell’s responsiveness to Notch signaling by creating a threshold (Zheng, 2008).

Lola regulates two cell-fate decisions in the eye: the R3 vs. R4 decision and the R7 vs. cone cell decision. Lola promotes R3 fate at the expense of the R4 fate, and it promotes R7 fate at the expense of the cone cell fate. Lola appears to influence these binary fate decisions by acting on responsiveness to Dl-N signaling. Lola regulates N-dependent gene expression, and Lola is able to suppress the effects of constitutively-active N on cell fate decisions. This latter result argues that Lola either acts downstream of N in the N signaling pathway or acts in a pathway that is convergent downstream with the N signaling pathway (Zheng, 2008).

Lola antagonizes 'strong N' signaling, which is evident by overexpressed Lola antagonizing strong N-dependent gene expression in R4 and transforming it into an R3-type. This is interpreted to mean that Lola normally makes the R3 precursor more resistant against induction into a R4-type, an interpretation that is consistent with mosaic analysis indicating that Lola is required in the R3 cell. Within the R3 precursor cell, Lola might repress transcription of genes that are normally activated by strong N signaling, such as the E(spl)m∂ gene. Doing so, Lola then would ensure the precursor would not inappropriately adopt an R4 cell fate. However, this notion raises a paradox: since Lola is expressed somewhat equivalently in all photoreceptors including R4, why does it not repress genes such as E(spl)m∂ in the R4 cell? One explanation is that it represses these genes in the R4 precursor but this repression is not strong enough to block their expression when induced by strong N signaling. Two lines of evidence support this explanation. Overexpression of Lola in the R4 precursor is sufficient to block expression of the E(spl)m∂ gene. This argues that Lola activity is a limiting component in N-dependent gene expression. Second, analysis of mutant lola clones reveals that R4 cells in mosaic ommatidia are more likely to be mutant for the lola gene. This result implies that Lola antagonizes N-activated programming of R4 fate in the R4 cell itself (Zheng, 2008).

Based on these observations, a model is proposed in which Lola acts as a transcription repressor of certain N-dependent genes that program the R4 cell fate. It acts on these genes in all cells as they are induced by EGFR to differentiate. In cells that receive no Dl-N signal or a weak Dl-N signal, repression is sufficiently strong to block N-dependent gene expression. However, in cells such as the R4 precursor that receive a strong Dl-N signal, repression is not sufficient to block gene expression, and the cells follow an R4 fate. Thus, Lola functions to make cells respond with a sharper threshold to Dl-N signaling. Since Lola functions in this manner only in cells that have already been induced by EGFR signaling, Lola seems to aid in crosstalk between the two signal transduction pathways. Moreover, the lola gene is repressed by Dl-N signaling; loss of N activity results in ectopic lola expression in many precursor cells. This maybe ensures that only cells induced to differentiate will respond with a sharp threshold to later Dl-N signals they might receive (Zheng, 2008).

Lola protein has sequence-specific transcriptional repressor activity and it genetically interacts with histone remodeling proteins. Thus, Lola could antagonize N signaling by virtue of it specifically binding and repressing transcription of genes that are activated by N signaling. Alternatively, Lola might render genes less responsive to N by specifically interacting with effectors of N signaling. In this regard, strong genetic interactions have been observed between Lola and Hairless, a co-repressor of Su(H) in Drosophila (Muller, 2005). It is not yet possible to distinguish between these two models of interaction (Zheng, 2008).

The Dl-N pathway is also required to specify both R7 and cone cell fates; strong Dl-N signaling induces cone fate and weaker Dl-N signaling induces R7 fate. Precursors with high levels of constitutively-active Notch become cone cells. This transformation is suppressed by Lola overexpression and leads to formation of more R7 cells. Conversely, loss of Lola leads to fewer precursors adopting an R7 fate and more adopting a cone fate, Again, these data fit with a model in which Lola sharpens the threshold response to levels of Dl-N signaling. It helps block strong N-dependent expression in the R7 precursor receiving a weak Dl-N signal, while it is insufficient to block the expression of the same genes in the cone precursors that receive a stronger Dl-N signal. However, in contrast to Lola's role in the R3/R4 fate switch, Lola has less effect on the R7/cone fate switch. Loss of lola causes infrequent R7 cell loss, and there is no significant correlation between the lola genotype of R7 cells in mosaic ommatidia and their normal development. Moreover, flies overexpressing Lola in cone cells only occasionally exhibit ectopic R7 formation. The reason why Lola has less impact on the R7/cone cell fate choice is unclear. Perhaps other factors play a redundant role with Lola in this fate switch (Zheng, 2008).

REFERENCES

Bardwell, V. J. and Treisman, R. (1994). The POZ domain: a conserved protein-protein interaction motif. Genes Dev. 8: 1664-1677. PubMed citation: 7958847

Brenman, J. E., Gao, F.-B., Jan, L. Y. and Jan, Y. N. (2001). Sequoia, a Tramtrack-related zinc finger protein, functions as a pan-neural regulator for dendrite and axon morphogenesis in Drosophila. Dev. Cell 1: 667-677. 11709187

Cavarec, L., et al. (1997). Molecular cloning and characterization of a transcription factor for the copia retrotransposon with homology to the BTB-containing lola neurogenic factor. Mol. Cell. Biol. 17(1): 482-94. PubMed citation: 8972229

Crowner, D., Madden, K., Goeke, S. and Giniger, E. (2002). Lola regulates midline crossing of CNS axons in Drosophila. Development 129: 1317-1325. 11880341

Ferres-Marco, D., et al. (2006). Epigenetic silencers and Notch collaborate to promote malignant tumours by Rb silencing. Nature 439(7075): 430-6. 16437107

Giniger, E., Tietje, K., Jan, L.Y. and Jan, Y.N. (1994). lola encodes a putative transcription factor required for axon growth and guidance in Drosophila. Development 120: 1385-1398. PubMed citation: 8050351

Goeke, S., et al. (2003). Alternative splicing of lola generates 19 transcription factors controlling axon guidance in Drosophila. Nature Neurosci. 6: 917-924. 12897787

Horiuchi, T., Giniger, E. and Aigaki1, T. (2003). Alternative trans-splicing of constant and variable exons of a Drosophila axon guidance gene, lola. Genes Dev. 17: 2496-2501. 14522953

Madden. K., Crowner, D. and Giniger, E. (1999). lola has the properties of a master regulator of axon-target interaction for SNb motor axons of Drosophila. Dev. Biol. 213(2): 301-13. PubMed citation: 10479449

Muller, D., Kugler, S. J., Preiss, A., Maier, D. and Nagel, A. C. (2005). Genetic modifier screens on Hairless gain-of-function phenotypes reveal genes involved in cell differentiation, cell growth and apoptosis in Drosophila melanogaster. Genetics 171(3): 1137-52. PubMed citation: 16118195

Zheng, L. and Carthew, R. W. (2008). Lola regulates cell fate by antagonizing Notch induction in the Drosophila eye. Mech. Dev. 125(1-2): 18-29. PubMed citation: 18053694

date revised: 30 July 2008

The Interactive Fly resides on the
Society for Developmental Biology's Web server.