BIOLOGICAL OVERVIEW

The Drosophila 'tissue polarity' genes control the orientation of bristles and hairs in the adult cuticle. Mutant flies have the polarity of cells altered in the plane of the epithelium without gross changes in the overall shape of imaginal structures or the distribution of differentiated cell types within them. Other tissue polarity defects include mirror-image duplications of the tarsal joints, rotations of bristle sockets in the leg, and changes in ommatidial polarity. The tissue polarity genes have been divided into three groups (Wong. 1993). The type 1 genes [dishevelled (dsh), frizzled (fz), and prickle (pk)], affect the whole body surface and are therefore believed to directly establish tissue polarity (Shulman, 1998). In contrast, the type 2 [inturned (in) and fuzzy (fy) and type 3 [multiple wing hairs (mwh)] genes affect distinct subsets of body areas and are thought to interpret the polarity established by the type 1 genes (Gubb, 1999).

prickle and spiny legs (sple) were thought to be two separate, although closely linked genes. Mutations in sple roughen the eye and disrupt ommatidial polarity. pk affects tissue polarity but has no mutant eye phenotype. However, a pk-sple double mutant gives rise to a strongly roughened eye, suggesting that sple and pk act redundantly in eye development. Sections of sple and pk-sple mutant eyes show a disturbed ommatidial polarity with each ommatidium having the normal arrangement of photoreceptor cells. Together with fz, these polarity mutants can be categorized into two classes based on the adult eye phenotypes. One class, which consists of fz, pk-sple and dsh, exhibits all three aspects of polarity phenotype seen in fz adult eyes. Among them, pk-sple had the strongest eye phenotype. The other class, which consists of sple, exhibits only one aspect of the polarity phenotype seen in fz eyes. Although sple ommatidia are still arranged in antiparallel arrays and the equator is still detected, some of the ommatidia are oriented such that their R7 cells are pointing away, rather than towards the equator. Anti-Elav staining of sple eye discs reveals that the disorientation is due to incorrect direction of rotation as in fz mutants. The single mutant alleles pk^pk and pk^sple give the most extreme phenotypes, but in reciprocal regions of the body; pk^pk in the wing and notum and pk^sple in the legs, abdomen, and eyes. Complementation between these classes of allele indicates two subtly different functions at the pk locus (Zheng, 1995 and Gubb, 1999).

Thus loss-of-function lesions in alternatively spliced transcripts of prickle result in two genetically distinct phenotypes, Pk and Spiny-legs. The pk and sple transcripts encode proteins that contain three LIM motifs and a novel conserved domain that has been called PET (Prickle Espinas Testin). Surprisingly, deletion of the entire gene gives a phenotype that is much weaker than that of either the pk or sple single mutants. This unusual result can be explained by the pk and sple gene products acting in concert. The single-mutant phenotypes result from misactivation, rather than simply blocking, of a pathway of polarity formation. It is proposed that the correct balance of the Pk and Sple variants is required for normal planar polarity signaling in the Drosophila imaginal discs (Gubb, 1999).

To investigate the functional relationship between the Pk and Sple protein variants, mutant combinations that alter the ratio of the pk and sple transcripts were generated. The strongest phenotypes result from lack of either one of these two transcripts in homozygous mutants. When the single mutant alleles are combined with pk^pk-sple alleles or deficiencies, the resulting phenotypes are weaker than the corresponding homozygous single mutant phenotypes. Thus, flies that carry only a single functional copy of the sple transcript or the pk transcript, show an intermediate phenotype more similar to pk^pk-sple13 than the single mutant. These results imply that the presence of one transcript without the other (as in pk^pk or pk^sple single mutants) creates an extreme phenotype that is corrected when the dose of the remaining transcript is reduced. The implication is that both the relative and absolute levels of the Pk and Sple proteins are important for the function of putative Pk-Sple homomeric or heteromeric protein complexes (Gubb, 1999).

The importance of the levels of Pk and Sple expression and the balance between them has been investigated further with overexpression constructs. When driven by the expression of the uniform drivers gal4-da (daughterless) or gal4-C765, P[UAS:pk⁺] an extreme tarsal duplication phenotype results, including a duplicated socket structure in the proximal T1 segment. Overexpression of UAS:sple⁺ with these drivers gives a Pk phenotype in the wing triple row bristles; the tarsi, however, remain completely wild type. These results confirm that overexpression of pk gives a phenotype analogous to lack of the sple transcript, whereas overexpression of sple gives a phenotype similar to lack of pk. The da-UAS:pk⁺ and da-UAS:sple⁺ wing phenotypes show an unexpected feature that is not seen in mutant alleles. Instead of uniform polarity patterns, swirls are seen in the wing hair orientation with different patterns from wing to wing. In addition to phenotypes resembling loss of function, misexpression of pk in engrailed (en)-UAS:pk⁺ flies blocks the migration of wing hairs from the distal vertex of wing cells to the central position. This antagonism between the pk and sple transcripts implies either that both proteins compete for a target that is present in limiting amounts, or that they form protein complexes with distinct activities. In pk^pk and pk^sple mutants excess, homodimers would misactivate polarity signaling (Gubb, 1999).

Deletion of the entire prickle gene gives a phenotype over the whole body surface that is similar to dsh and fz. To test whether these mutants affect the same signaling pathway, double mutant combinations were made. Double mutants of pk^pk and dsh in the triple row give a Dsh phenotype, suggesting that dsh is epistatic to pk^pk. In other words, dsh function is blocked; pk function is irrelevant. These data suggest that many of the functions of pk require dsh. The situation within the wing blade is less clear, as the double mutants give an intermediate phenotype. The dsh; pk^sple double mutant retains a dsh phenotype in the triple row, but the wing hair pattern is altered, despite the fact that pk^sple alleles have no wing phenotype. In the leg, the dsh tarsal phenotype is not modified by pk^pk. There is a synergistic interaction between dsh and pk^sple, with the double mutant giving a more extreme mirror-image transformation of the T1 segment than either of the single mutants. This extreme transformation is also seen with UAS:pk⁺ overexpression and in fz (Gubb, 1999).

Clones of fz show a directional nonautonomy that has been interpreted to mean that fz mediates the intercellular communication of a polarity signal. On the other hand, dsh is strictly cell autonomous in clones, implying that it is involved in signal reception. To investigate whether pk participates in both reception and transmission of a polarity signal, pk^pk clones in the wing were studied. Large clones in the wing express the mutant polarity typical of that region of the wing in homozygous flies. There is an occasional nonautonomous disruption of polarity in wild-type cells adjacent to the proximal or lateral margins of a pk^pk clone. In these cases, a short range perturbation aligns wild-type cells with the mutant polarity pattern. There is no clear pattern to the position of such clones, but small peninsulas of pawn (pwn⁺) tissue surrounded by pk^pk pwn tissue tend to adopt the mutant polarity pattern. Smaller clones, induced later than 72-96 hr, do not alter the polarity of adjacent pwn⁺ tissue. In contrast to the autonomous behavior of pk, clones of fz tricornered (trc) cause long-range domineering nonautonomy both distal (Vinson, 1987) and lateral to the clone. Wild-type hairs are oriented toward the clone as though it is acting as a polarity 'sink'. Proximal cells are also directed toward the fz trc clone, but as this is the normal orientation for wing cells no polarity changes would be expected (Gubb, 1999).

There are a number of conserved features within the LIM domains of the pk family proteins. In the first LIM domain, there is a proline residue between the third (H) and fourth (C) zinc binding residues. This internal proline residue might introduce a kink within the zinc binding site. Similar LIMP domains with an internal P at this site are found in the human SLIM 2, SLIM 3, PINCH, Zyxin, and Paxillin proteins and LIM domain kinase 2 (LIK2) from chicken, mouse, rat, and human. In addition, the length of the consecutive LIM domains is constant in the pk family, with LIM1 containing 57 amino acids; LIM2 containing 52 aa, and LIM3 containing 56 aa. The implication is that the conservation of the triple LIM domain itself is important, rather than the individual LIM domains. In the case of the triple LIM domain protein Zyxin, the individual LIM domains may bind different target proteins and act as a template for the assembly of a number of structural components (Beckerle, 1997 and Gubb, 1999). A similar scaffold function for the Pk protein would be consistent with the cell autonomy of pk in clones and the expression of pk transcripts in cells that are changing shape. The blocking of the normal migration of the actin-rich prehair structure to the center of wing cells by overexpression of pk also implies a role in cytoskeletal remodeling. The lack of embryonic phenotype, despite the dynamic expression pattern, implies that pk function is redundant during embryonic development. A putative embryonic pk function could, in principle, be maternally supplied, but pk mutant strains of all three classes are fully fertile and homozygous pk embryos from homozygous mothers remain wild type. Similarly, gal4-da; P[UAS:pk⁺], and gal4-da; P[UAS:sple⁺] flies are completely viable and show no embryonic phenotype, despite the embryonic expression of the gal4-da driver (Gubb, 1999).

Perhaps the most surprising feature of the tissue polarity mutants is the precise polarity patterns seen in the wing hairs (Gubb, 1982). Rather than reflecting fine-grained positional information, however, the precision of the final pattern might be dependent on a tessellation mechanism. The orientation of the first cell would determine the alignment of subsequent cells, like sticking tiles on a bathroom wall. In pk^pk mutants, the alignment of wing hairs deviates progressively with occasional abrupt changes. In the adult mutant wing, hair polarity alters gradually with sudden topological discontinuities that resemble the stacking flaws in liquid crystals. In Df(2R)pk-30 wings, regions where the wing hairs are uniformly oriented retain a predominantly hexagonal array. Regions surrounding topological discontinuities, such as the anterior whorl, show irregular cell shapes frequently associated with duplicated wing hairs (Gubb, 1999).

Strong support for a tessellation mechanism is given by the overexpression phenotype of Pk and Sple when driven by a ubiquitous promotor. Although polarity patterns are variable, hair orientation alters smoothly from cell to cell across the wing surface, indicating that the polarity of cytoskeletal structures is aligned within large fields of cells. It is as if cell packing had nucleated randomly and then spread to neighboring cells until meeting an adjacent domain. The short-range perturbation in polarity that is occasionally seen proximal and lateral to a pk^pk clone is consistent with mechanical adjustment of cells to fit against their immediate neighbors, unlike the long-range domineering nonautonomy lateral and distal to fz clones. With both classes of clones, a tessellation mechanism might impose a threshold. Below this threshold, disruptions in polarity signaling would fail to affect the orientation of neighboring cells (Gubb, 1999).

The tarsal phenotype of dsh is very similar to pk^pksple13 causing duplications of the T3 and T4, segments. A more extreme phenotype including complete duplications of T2 to T4 together with a well-developed ectopic T1 joint are seen in dsh; pk^sple, overexpression UAS:pk⁺ and fz mutant strains, implying that a similar polarity signaling pathway is affected. It has been suggested that pk and fz are upstream in a signaling pathway leading to dsh (Wong, 1993 and Shulman, 1998). There are several problems with a simple linear pathway, however, and the relationship between pk and fz is unclear. (1) Neither fz nor pk is clearly epistatic to the other; rather, the double mutant (pk^pk; fz) phenotype is intermediate between pk^pk and fz (Gubb, 1982 and Wong, 1993). (2) Whereas fz clones cause a long-range nonautonomous disruption in surrounding tissue, pk^pk clones resemble dsh clones in being almost completely cell autonomous. (Both pk^sple1 and pk^pk-sple13 clones are cell autonomous in the eye. In contrast, the pk^pk triple row bristle phenotype (that results from Sple expression in the absence of Pk) is suppressed in dsh; pk^pk wings and the dsh polarity pattern is modified in dsh; pk^sple wings (despite pk^sple on its own not having a wing phenotype). (3) Complete lack of pk transcripts produces a phenotype very similar to dsh. Taken together, these results indicate that Pk is not downstream of Fz but may represent an alternative input into Dsh-mediated planar polarity signaling. If Dsh is acting as a scaffolding or adapter protein then it would be required in stoichiometric ratios to its target proteins, and overexpression would prevent assembly of functional complexes. It may be that the Pk protein isoforms are components of this protein complex that are expressed in cells remodeling their cytoskeletal architecture (Gubb, 1999).

GENE STRUCTURE

The 5' start of the pk transcript (located in the most 3' portion of the gene) lies within a cluster of three serpins (serine Proteinase inhibitors) (Spn43Aa, Spn43Ab, and Spn43Ac), one of which corresponds to the necrotic gene. A transcript with homology to Adenosine kinases maps just distal to pk gene termination. The first pk exon contains an untranslated leader sequence of 0.8 kb, with the putative translation start site being 39 bp 5' to the large intron. The breakpoints of pk-associated mutations map across 70 kb of genomic DNA. Single-mutant pk^pk alleles carry lesions within the proximal 30 kb (the 3' portion of the gene), whereas pk^pk-sple double-mutant alleles map between 42 and 69 kb (nearer the 6 common exons) and none of the pk^sple single mutations have been localized. Two putative pk transcripts are detected as weak bands of ~4.2 and 5.1 kb in embryos and 2-day pupae on developmental Northern blots. The shorter transcript corresponds to pk (initiating from most 3' exon) and the larger to sple. The 5' end of pk hybridizes to the region of the pk^pk breakpoints and is separated by a large intron from the remaining 6 exons. The 5' end of sple is within the large pk intron. An additional medial transcript, pkM, is only detected in embryonic stages. The identity of the pk and sple transcripts has been confirmed by rescue experiments (Gubb, 1999).

Exons - 6 common exons and three initiating exons each serving a different transcript

PROTEIN STRUCTURE

Amino Acids - The Pk protein isoforms consist of 870 (Pk), 936 (PkM), and 1206 (Sple) amino acid chains with the conserved PET and LIM

domains mapping entirely within the common exons

Structural Domains

The tissue polarity mutants in Drosophila include a set of conserved gene products that appear to be involved in the control of cytoskeletal architecture. The tissue polarity gene prickle (pk) encodes a protein with a triple LIM domain and a novel domain, designated PET, is present in human, murine, and Caenorhabditis elegans homologs. Three transcripts have been identified (pk, pkM, and sple) encoding 93-, 100-, and 129-kD conceptual proteins, respectively. The three transcripts span 70 kb and share 6 exons that contain the conserved domains. The second to seventh exons of pk are common to the three transcripts. These common exons encode a peptide sequence containing three LIM domains, but no recognized DNA binding, membrane spanning, or signal peptide motifs. LIM domains, serving to mediate protein interactions, are cysteine-rich sequences with a double zinc finger motif (for review, see Dawid, 1998). Four closely related LIM domain sequences have been found by database searches; one murine (Testin; Divecha, 1995), one human (LMO6; Fisher, 1997), and two Caenorhabditis elegans. An additional pk cognate gene in Drosophila, espinas (esn), maps just proximal to the 43A1.2 Serpin cluster (C. Green and D. Gubb, unpubl., cited in Gubb, 1999). The Serpin cluster is in the far upstream portion of the prickle gene. Comparison of these sequences identifies a novel conserved PET domain, 5' to the LIM domains. Two further incomplete sequences with PET domain homology have been identified by database searches: a human EST and a C. elegans ORF. The PET domain is unrelated to other sequence motifs in the protein database and encodes a neutral stretch of amino acids with no obvious distinguishing features. Database searches failed to identify any sequences with significant homology to peptides encoded by the unique 5' pk, pkM, or sple exons. Both the human and mouse homologs, LMO6 and Testin (Divecha, 1995 and Fisher, 1997), encode alternative 5' transcripts that may be a general feature of the pk family (Gubb, 1999).

date revised: 17 October 99

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

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