Molecular Structure offrizzled , a Drosophila Tissue Polarity Gene Paul

The function of thefrizzled V;) locus is required to coordinate the cytoskeletons of pupal epidermal cells so that a parallel array of cuticular hairs and bristles is produced. We report here the molecular cloning and characterization of the fz locus. The locus is very large. Mutations that inactivate the gene are spread over 100 kb of genomic DNA. The major mRNA product of the gene is a 4-kb RNA that is encoded by 5 exons spread over more than 90 kb of genomic DNA. Conceptual translation of this mRNA indicates that it encodes an integral membrane protein that is likely to contain both extracellular and cytoplasmic domains. T HE adult cuticle of Drosophila melanogaster has a rich morphology, containing a large number of polarized structures, principally bristles (4-cell sense organs) and hairs (cellular extensions of individual cells that later become sclerotinized). These polarized structures are evenly spaced, aligned in parallel, and typically point distally on appendages and posteriorly o n the thorax and abdomen, thus defining a tissue polarity. The development of tissue polarity is expected to require intercellular signaling; however, very little is known about this process. Embryological experiments have led to suggestions that tissue polarity is a manifestation of a gradient of cell adhesiveness (NARDI and KAFATOS 1976), a gradient of a diffusible morphogen (LAWRENCE 1966; STUMPF 1966), or direct cytoskeleta1 and plasma membrane connections between cells (TUCKER 198 1). Examination of pupal wings indicates that the developing hairs contain large bundles of actin filaments, and that the developing hairs are aligned at the earliest stages of hair morphogenesis (IA. Wong and P. N. ADLER, unpublished results). Cell geometry appears to play a role in tissue polarity on the wing since the bundles of actin filaments that take part in forming the hair extend istally from the distal vertex of each hexagonally shaped wing cell. In Drosophila a small number of genes have been identified on the basis of mutant phenotypes as being essential for the generation of tissue polarity (GUBB and GARC~A-BELLIDO 1982). These mutations, to a first approximation, do not alter the morphology of structures produced by individual cells (or developmental units). Rather they alter the orientation of these structures (e .g . , hairs) with respect to their neighbors and the organism as a whole. The best studied ' Cul-rent address: (hrnegie Institution of Washington, 1 1 5 West Univer' C:ur!-ent address: Crop Genetlcs International, Dorsey, Maryland 21076. b i t v Parkway, Baltin~ore, Maryland 21 210. (;twetir\ 1 2 6 401-416 (Octoher. 1990) member of this group of genes is the frizzled (fz) locus (ADLER, CHARLTON and VINSON 1987; VINSON and ADLER 1987). Most mutations in this gene disrupt normal tissue polarity in all body regions. We have primarily studied the effects of fz mutations in the wing (ADLER, CHARLTON and VINSON 1987; VINSON and ADLER 1987) because of the ease of examining this flat structure microscopically. In the most severely affected region of the wing, null mutations result in a loss of both proper polarity and the parallel alignment of neighboring hairs, giving rise to a relatively random pattern of hair polarity. In less severely affected regions of the wing, polarity is abnormal although neighboring hairs usually remain aligned, thus resulting in a distinctive swirling pattern. This swirling effect is the principal phenotype seen in weak mutations that only partially inactivate the fz locus. frizzled mutations also typically cause a rough eye. Here again the mutation does not grossly alter the differentiation of individual cells or ommatidia, rather it alters their spatial relationship to the rest of the eye (R. CARTHEW, personal communication; L. MARSH, personal communication). There are several exceptional fz alleles that do not fall into the hypomorphic to amorphic series defined by the majority of alleles. The exceptional alleles were originally identified as tissue specific alleles, because they do not result in rough eyes, as do all other moderate or strong fz alleles (ADLER, CHARLTON and VINSON 1987). Mitotic clones on the wing of several fz alleles (including a null allele) disrupt the polarity of wild type cells located distal (but not proximal) to the clone of mutant cells (VINSON and ADLER 1987). This non-cell autonomous behavior argues that the polarity information itself is altered in fz mutant tissue, and that the function of the fz locus is essential for the transmission of a polarity signal along the proximal-distal axis of the wing. When similar mitotic clones were 402 P. N. Adler et al. generated for the two exceptional (tissue specific) alleles the polarity of the surrounding wild type cells was normal. This cell autonomous behavior suggests that these alleles do not alter the polarity signal, rather, they alter the ability of cells to respond to the signal. Thus thefz gene has two functions. It is required for the transmission of a polarity signal (noncell autonomous function) as well as for the cellular response to (transduction of) the polarity signal (cell autonomous function). The fz locus could encode either one bifunctional protein or two single function proteins. To try to distinguish between these possibilities and to help in our efforts to understand the molecular mechanism(s) underlying the development of tissue polarity we have undertaken the molecular cloning of thefz locus. We report here the cloning of thefz locus via the transposon tagging strategy (BINGHAM, LEVIS and RUBIN 1981; RUBIN, KIDWELL and BINGHAM 1982; SEARLES et al. 1982), and the mapping of the locations of 11 independent fz mutations. Somewhat unexpectedly, the fz gene is quite large-mutations were found to be located over approximately 100 kb of genomic DNA. A cDNA clone and Northern analyses indicate that the principal RNA product of the fz locus is a 4-kb RNA that is encoded by 5 exons that span over 90 kb of genomic DNA. Conceptual translation of this RNA indicates that thefz locus encodes an integral membrane protein that, if located in the plasma membrane, would be able to interact with both extracellular and cytoplasmic molecules, and thus potentially function in both the transmission of tissue polarity information between epidermal cells and the transduction of the polarity signal to the cytoskeleton. Evidence for a second substantially rarer@ transcript has also been obtained. Conceptual translation of the sequence of a cDNA clone of this RNA yields a truncated version of the protein predicted from the major transcript. Thefz gene is the first tissue polarity gene molecularly cloned. METHODS AND MATERIALS Drosophila strains and mutant isolation: The cytology, mode of induction, and phenotypic strength of fz alleles mentioned in the text are shown in Table 1. Marker mutations and balancer chromosomes are described in LINDSLEY and GRELL (1 968). Hybrid dysgenesis was induced by crossing Harwich (P) males and Oregon-R (M) females. The dysgenic F, progeny of these flies were crossed tofz th st inlTM3 flies and the F2 progeny screened for newfz mutations. Flies carrying new mutations were then crossed to TM3 (st)#% th st flies and stocks carrying the new mutations established over TM3. Newfz mutations were isolated after y-ray (3500 R, from a I3'Cs source (Isomedix) or EMS (LEWIS and BACHER (1968)) mutagenesis by crossing marked mutagenized males to fz females and screening the F1 progeny for phenotypicallyfz flies. Stocks of each new mutation were established over a TM3 balancer chromosome. Revertants and stronger variants were derived fromfzc'FKc by crossingfzCT8'/TM3 (P) males and TM3IDCXF (M) females. The dysgenic fz""" carrring female progeny from this cross were then mated tofzp th stlTM3 ( s t ) males. The f i c T 8 c ' I f p 2 1 progeny of this cross were then screened for flies with either a wild type or stronger mutant phenotype. Mutant or revertant stocks were then established using the TM3 balancer chromosome. Cytogenetic analysis: Salivary gland squashes were done by standard technique except no stain was used. Chromosomes were examined under phase optics. In situ hybridizations were done as described by BINCHAM, LEVIS and RUBIN (1981) using nick translated ('H-TTP) DNA as a probe. Isolation of genomic clones: A genomic library was constructed in the Bam site of the EMBL4 vector (FRISCHAUF et al. 1983) from a partial MboI di est of DNA isolated from fzCTBc containing flies. The fi"'" containing chromosome had been outcrossed (with the fz""" stock as the female parent) to remove extraneous P elements. During this outcrossing the number of P elements apparently fell low enough to induce some mobilization of the P elements. The DNA was packaged in vitro, plated on the P2 lysogen (Q359), and this primary plating screened with a probe made by nick translating the P element containing plasmid pn25.1 (RUBIN and SPRADLING 1982; SPRADLING and RUBIN 1982). DNA from 31 P element containing recombinant phage was used to make probes for in situ hybridization to Oregon R salivary gland chromosomes. Chromosome walking was done by probing filter plaque lifts of the Maniatis library (MANIATIS et al. 1978) with probes made by nick translation or random priming. On two occasions (to get across middle repetitive sequences) we isolated recombinant clones from a genomic library of the P2 strain constructed and kindly provided by L. SEARLES (University of North Carolina). We subsequently isolated bacteriophage from the Maniatis library that contained inserts that spanned the middle repetitive element. Large scale preparation of bacteriophage DNA was carried out by the glycerol gradient technique (GARBER, KuROIWA and GEHRINC 1983). Restriction mapping was carried out using single and double digests with six enzymes (WEINER, SCOTT and KAUFMAN 1984). The map was confirmed by genomic Southern analyses. Isolation of cDNA clones: In our initial experiments three cDNA clones were isolated from a cDNA library constructed by POOLE, KAUVER and KORNBERG (1985) in the XgtlO vector using as a probe an 8-kb EcoRI fragment of genomic DNA from CV150. In subsequent experiments several additional clones were isolated using the insert from the cDNA clone ACVC22 (this clone was named CVC22 in a previous publication (VINSON, CONOVER and ADLER 1989)) as a probe. The procedures for experiments with this library and clones derived from it were the same as for the genomic libraries. Many cDNA clones were recovered by screening the cDNA libraries constructed by BROWN and KAFATOS (1 988) using random primed DNA from previously isolated cDNA clones as probes. The cDNAs in these libraries were directionally cloned into the pNB4O plasmid vector, thus facilitating the determination of 3' and 5' ends. All of the clones have an oligo(A) stretch at their 3' end, although in some cases this is due to the oligo(dT) primer hybridizing to an internal oligo(A) sequence during reverse transcription. The library was screened by the colony hybridization procedure described by BROWN and KAFATOS (1988). Isolation and analysis of genomic DNA: Drosophila DNA was isolated from adult flies as follows. Two hundred adult flies were anesthetized, and homogenized in a ground Tissue Polarity Gene in Drosophila 403