Adhesive pad differentiation in Drosophila melanogaster depends on the Polycomb group gene Su(z)2

In insects, essentially two types of anatomical structures account for the ability to climb on vertical surfaces, the claws of the pretarsus, and smooth or hairy adhesive pads located on different parts of the legs (Gorb et al., 2007). Whereas claws are mainly involved in climbing on rough ground, adhesive pad structures enable insects to walk on vertical smooth surfaces. Both smooth and hairy pads allow adaptation to different surface profiles and enhance contact areas with the ground. The hairy adhesive structures are found in evolutionarily younger insects groups, such as diptera with a huge diversity in terms of shape, which is considered as an advanced adaptation to different ecological niches (Gorb, 2001). Generally, two types of physical forces are known to contribute to the dipteran wet adhesive system: capillary forces, which play the major role, and−to a minor extent−van der Waals forces (Langer et al., 2004). Van der Waals forces occur between pad and substrate molecules, which are in very close contact. Additionally, flies secrete fluid into the contact area, which is responsible for generation of capillary forces (Bauchhenß, 1979; Walker et al., 1985; Langer et al., 2004).

In higher dipterans, such as flies, adhesion to smooth surfaces is mediated by hairy attachment devices located on the tarsus of each leg, the so-called pulvilli (West, 1862; Bauchhenß, 1979; Walker et al., 1985; Gorb, 1998). Pulvilli are covered by setae, hair-like cuticle outgrowths differentiating from epidermal cells and possessing spatula-like terminal contact zones. Each pulvillus contains about 30 individual setae. Although Drosophila offers unique experimental and genetic tools, most studies on the functional anatomy of adhesive structures in Diptera have focused on larger species such as the blowfly (Calliphora). Thus it is not surprising that our knowledge about adhesive device differentiation in Drosophila remains rather limited. One example for this lack of data is the fact that by now, it is not even known whether Drosophila setae secrete fluid similar to the setae of larger fly species. In addition, the molecular mechanisms underlying adhesive device differentiation remain to be investigated.

Here, we describe the first Drosophila mutant with malformed adhesive pad structures, which affect the mutant's climbing ability in multiple ways. In this mutant, the contact zones between the adhesive setae and the substrate are reduced in size and display a different shape, which leads to a climbing disability on smooth vertical surfaces. The mutation is caused by a transposon insertion in the untranslated region of the Su(z)2 gene as proven by analysing revertants as well as transheterozygous animals. Northern blot analysis revealed the appearance of an additional Su(z)2 transcript in mutant animals, which may cause deregulation of yet unknown target genes involved in adhesive pad differentiation. Su(z)2, a RING-finger DNA binding protein and the homolog of the human transcriptional repressor and tumor suppressor Mel-18, is known to be a key regulator of homeotic (Hox) gene activity. Polycomb group (PcG) genes, to which Su(z)2 belongs, are responsible for maintaining the silenced state of repressed target genes (Pirrotta, 1997). Thereby, PcGs regulate numerous genes in Drosophila, e.g. those involved in epidermal cell differentiation. In mammals and humans, PcG genes are well known to act as epigenetic regulators of gene silencing. Although we have not analysed the precise role of Su(z)2 in setae differentiation, we believe that our results clearly reveal the potential of Drosophila genetics for future investigations on adhesive structure morphology, development and function.

We have previously generated several transgenic Drosophila strains carrying a GFP reporter gene driven by the heart-specific enhancer element of the hand gene (Sellin et al., 2006). One of these transgenic lines (handC-GFP1.1) attracted our attention because 100% of the homozygous (but 0% of the heterozygous) viable adults failed to climb on the vertical walls of the breeding vials, indicating that the transposon integration might have caused a mutation in a climbing relevant gene (see supplementary material Movie 1).

While the mutants failed to climb on smooth vertical surfaces, they were fully able to walk on smooth horizontal or rough vertical surfaces, as tested with a simple climbing test apparatus (see Materials and methods: Climbing assay). Our initial observations prompted us to measure the climbing ability of the mutant more precisely using a slide friction tester (Fig. 1). The mutants failed to climb on all tested surfaces at significantly lower tilting angles than the wild type (***P<0.001). Surfaces with very low roughness (glass, 0.3−1 µm asperity size) proved especially difficult for the mutants. On surfaces with higher roughness (3−12 µm asperity size), mutants performed significantly better, but were not able to reach the same maximum tilting angle as the wild type.

Microscopic inspections of intact and dissected homozygous mutant animals did not reveal any severe leg defects, such as leg malformations or degenerated muscles that might interfere with locomotory ability. However, upon analysing the tarsal regions of wild-type and mutant Drosophila by scanning electron microscopy (SEM) (Fig. 2A,B), we found that the adhesive pads as well as the setae itself were malformed in the mutant. In contrast to the wild-type pulvilli, which are long and quite constant in width from basal to distal (most obvious in lateral view, Fig. 2A), the mutant pulvilli are shorter and thicker at the basal part (Fig. 2B). Furthermore, the flexible part of the pulvillus appears to be notably shorter than in the wild type. Measurements of the lengths of 10 pulvilli of each genotype on the ventral side from the first seta to the tip confirmed that the mutant pulvilli with an average length of 23.95 µm (s.d. 2.84) are significantly shorter than those of the wild type with an average length of 30.36 µm (s.d. 2.16) (***P<0.001, Fig. 3A). Additionally, we counted whether the number of setae per pulvillus differs between mutant and wild-type flies (Fig. 3B). On average, 31 (s.d. 1.81) setae were found on a wild-type pulvillus, while only 26 (s.d. 2.17) setae were observed on a mutant pulvillus (***P<0.001). This amounts to a total difference of 60 setae per animal (10 setae per leg), which reduces the capacity of the adhesive structures by 16%. We also tested for differences in the claws (length and curvature) between mutant and wild-type animals and found no differences (data not shown).

Another major difference was observed regarding the morphology of the setal tips. Whereas wild-type animals displayed thin, spatula-like tips (Fig. 2C), many of the mutant's tips were folded (Fig. 2D), resulting in a reduced contact area with the substrate and an increase in thickness. The folded structure of the setal tips could only be observed in the mutants but was never seen in wild-type animals. To quantify the effect on the contact area, we measured the width of terminal spatulae using SEM images, in which the setae were oriented in parallel with the plane of view. Wild-type tips are on average 1.97 μm (s.d. 0.16) wide, whereas mutant setae measure only 1.6 μm (s.d. 0.25) across the tip (***P<0.001, Fig. 3C). This accounts for an 18.8% reduction of the terminal width in the mutant.

Previous studies have shown that the secretion of tiny droplets of liquid from the setal tip is used by many species of the suborder Brachycera (flies with shortened antennae) to improve adhesion (Gorb, 2001). By employing an inverted microscope equipped with reflection interference contrast microscopy (RICM) illumination (see Materials and methods), we examined whether Drosophila is also able to produce liquid at the adhesive structures and subsequently compared footprints of wild-type with mutant flies. Wild-type individuals left a clear footprint consisting of liquid droplets matching the patterns of setae, indicating that both pulvilli per leg contact the surface (inset in Fig. 3D). This result shows that Drosophila secretes liquid to produce capillary forces for adhesion at the setae contact zones. We did not find any opening near the spatula on SEM pictures, therefore the secreted liquid is presumably released through the porous channels of setal cuticle as previously described in other representatives of Brachycera (Bauchhenß, 1979; Gorb, 1998). In wild-type animals, contact between pulvilli and the glass surface was detected. In contrast, mutant animals failed to establish contact with their pulvilli and walked on their claws instead (Fig. 3D,E). As a result of this lack of contact, it remains unclear whether mutant flies secrete liquid at their setae or not.

The climbing mutant (handC-GFP1.1) was discovered upon generation of transgenic fly lines carrying a green fluorescent protein (GFP) reporter (Sellin et al., 2006). We obtained several independent transgenic lines carrying the same construct at different chromosomal locations but none of them displayed climbing disability on smooth vertical surfaces except for the line handC-GFP1.1. We therefore assumed that the integration of the P-element in this particular line affects the expression of an endogenous gene at the site of insertion. An initial mapping based on balancer chromosomes showed that the P-element is located on the second chromosome. Using inverse polymerase chain reaction (PCR), we subsequently identified the precise insertion site of the P-element, which is in the first intron of isoforms B and C of the Su(z)2 gene, and thus upstream of the first untranslated exon of isoform A (see Fig. 4). Thus we consider this particular mutant a new Su(z)2 allele, which we named Su(z)2^handC-GFP1.1.

To ensure that the P-element in the Su(z)2 gene is responsible for the observed phenotype and not a second hit elsewhere in the genome, we remobilized the P-element by crossing in a transposase source to generate revertants. Since the P-element contains a white⁺ gene and drives GFP expression in the heart throughout development, we could easily identify revertants by screening for white-eyed flies that at the same time lacked GFP expression. This led to the recovery of two independent revertants with fully restored climbing ability in the climbing assay. Sequence analysis confirmed that the P-element was successfully removed from the Su(z)2 gene locus in these lines and that the wild-type sequence had been restored. Additionally, we recovered a revertant line that lacked GFP expression but still failed to climb on smooth vertical surfaces. Sequencing revealed that in this case a truncated P-element remained at the original position.

As the P-element in Su(z)2^handC-GFP1.1 is located in the first intron of Su(z)2 isoforms B and C, and thus upstream of the first untranslated exon of Su(z)2 isoform A, we asked whether the transposon insertion has an effect on the expression level of any isoform. To address this question, we performed northern blots with total RNA isolated from homozygous mutants and probed it with anti-sense riboprobes being specific to all individual Su(z)2 isoforms (see Materials and methods). In the wild type, we detected a single band that presumably represents all three isoforms with annotated lengths of 6301, 6631 and 6522 nucleotides, respectively (St Pierre et al., 2014). Interestingly, in the mutant line an additional transcript of increased size (∼7 kb) becomes apparent (Fig. 5). Because the P-element is located in an untranslated region, the additional transcript may represent a new isoform of Su(z)2, which would render the Su(z)2^handC-GFP1.1 allele into a regulatory gain-of-function mutant. We furthermore noted that the wild-type transcripts are apparently still expressed in the Su(z)2^handC-GFP1.1 allele, in amounts comparable to the wild type. Considering these data, we speculate that the expression of an additional transcript variant of Su(z)2 in the climbing mutant interferes with the morphogenesis of adhesive devices, presumably by deregulation of some or several target genes involved in adhesive device development. To elucidate the molecular and cellular processes underlying the observed malformations, additional work is necessary. Taken together, our findings demonstrate that Drosophila is a highly suitable model system to identify and analyse genes involved in adhesive device formation.

Phenotypes associated with homozygous viable Su(z)2 alleles include bristle abnormalities, suppressor of variegation phenotypes and eye color variations, to mention a few. Surprisingly, a climbing disability phenotype has not been noticed so far. Therefore we analysed available and molecularly characterized Su(z)2 alleles obtained from the Bloomington Drosophila Stock Center (BDSC), the Drosophila Genomics Resource Center (DGRC) and the Exelixis Collection at the Harvard Medical School for climbing ability (climbing assay). Because our Su(z)2^handC-GFP1.1 allele expresses an artificial transcript as revealed by northern blot analysis, and therefore most likely represents a gain-of-function allele, we wished to evaluate this hypothesis by crossing this allele to a deficiency [Df(2R)Exel6062], which uncovers the Su(z)2 locus and two neighboring genes (CG33798 and CR44339). Indeed, removal of one copy of Su(z)2 in combination with Su(z)2^handC-GFP1.1 had no effect on climbing ability, supporting the assumption that it may represent a gain-of-function mutation. To further characterize this mutation, we focused on homozygous viable Su(z)2 mutants induced by transposons with known nearby insertion sites and available data regarding their respective flanking sequences (St Pierre et al., 2014). None of the tested mutant alleles (P{SUPor-P}^KG07068, P{XP}Su(z)2^d01221, P{RS5}Su(z)2^5-SZ-3055, PBac{RB}Su(z)2^e00448) displayed any climbing disability on smooth vertical surfaces. Therefore we asked if a particular genetic combination of the Su(z)2^handC-GFP1.1 allele and one of the molecularly characterized Su(z)2 alleles in trans shows a climbing phenotype. Here we also included alleles that were annotated as being lethal. It turned out that the genotypes Su(z)2^handC-GFP1.1/P{SUPor-P}^KG07068, Su(z)2^handC-GFP1.1/P{XP}Su(z)2^d01221, Su(z)2^handC-GFP1.1/P{RS5}Su(z)2^5-SZ-3055 and Su(z)2^handC-GFP1.1/P{lacW}Su(z)2^k06344 lead to viable adults that fail to climb on smooth vertical surfaces (Fig. 4). Neither the genomic position nor the orientation of the transposon relative to the Su(z)2 open reading frame (Fig. 4) allowed us to conclusively deduce the molecular situation that leads to the climbing disability phenotype in the Su(z)2^handC-GFP1.1 allele. Nevertheless, the occurrence of climbing disability of selected Su(z)2 alleles only in trans with the Su(z)2^handC-GFP1.1 mutation let us assume that the adhesive organ malformations caused by Su(z)2^handC-GFP1.1 might be the result of a complex gain-of-function effect. Furthermore, our genetic analysis provides additional evidence that the mutated Su(z)2 gene in Su(z)2^handC-GFP1.1 is indeed responsible for the observed climbing disability phenotype.

We discovered a transgenic Drosophila line in which all individuals, independent of their sex, fail to climb on smooth vertical surfaces or ceilings, whereas their ability to walk on horizontal substrates is not affected. Their ability to climb on rough ceilings is also slightly reduced. For walking on smooth vertical surfaces, flies use attachment devices located on the tarsi of their legs (Beutel and Gorb, 2001). These structures consist of dozens of cuticular setae, which are used to generate the contact zone between the substrate and the foot. The terminal elements (spatula) of the flexible setae easily adjust to the local texture of the substrate and thereby establish a high number of micro-contact spots. For Drosophila, the distal part of the setae shows a spatula-like shape with a width of approximately 2 µm. Since the discovered mutant was unable to climb on smooth walls, but otherwise behaved normally, we assumed that this mutation may affect proper setae differentiation. SEM analysis showed that the shape of the whole pulvillus is different in mutants in comparison with the wild type. Mutants display shorter, and therefore less flexible pulvilli at the basal part and harbor less setae, which are often folded at the distal spatula-formed tip and are therefore thicker and narrower. All these morphological features must lead to a reduction of the total contact area. Furthermore, we found that wild-type animals secrete fluid into the contact area to generate capillary forces. Further investigations are necessary to elucidate where this fluid is produced and secreted. We failed to detect openings at the setae, although such openings were shown to be present in several other insects (Gorb, 1998). Whether the mutants described herein produce such an adhesive fluid cannot be answered so far. Probably they do, but the setae do not contact the surface (as far as we can tell from the RICM experiment) and thus a clear footprint with remaining liquid droplets is not detectable.

In summary, we postulate that climbing performance of the mutants is severely reduced due to several reasons. On extremely smooth surfaces such as glass, the pulvilli are too short to bring the spatula of the setae in direct contact with the surface and the flies have to rely on their claws. On increasingly rougher surfaces, the setae display increasing ability to reach the surface but their reduced number and their malformed tips do not provide the same adhesion as in the wild type. In addition, mutant animals most likely can apply their claws for interlocking with and climbing on rougher surfaces, therefore the climbing performance of mutants increases on such surfaces. This is reflected in the climbing performance measured with the slide friction tester (Fig. 1). However, whether the mutant animals sense their malformed pulvilli and adapt by using the claws instead cannot be answered by our studies.

Furthermore, we have been able to show that the climbing inability in the allele Su(z)2^handC-GFP1.1 is due to a transposon insertion in the non-coding region of the Su(z)2 gene. The transposon insertion leads to the expression of an artificial Su(z)2 transcript with an unaltered expression of the endogenous isoforms A and B. Mobilization of the P-element in Su(z)2^handC-GFP1.1 led to the recovery of the wild-type sequence and to adult flies with full climbing capability. The molecular mechanisms underlying the observed adhesive device malformation remain elusive so far, but we speculate that the appearance of an artificial Su(z)2 transcript may lead to an altered expression of genes involved in adhesive structure development. It has been shown previously that overexpression of Su(z)2 in gain-of-function mutants, or in transgenic animals carrying a heat-shock inducible hs-Su(z)2 construct, leads to bristle abnormalities (Brunk and Adler, 1990; Brunk et al., 1991; Sharp et al., 1994) that may result from altered expression of genes involved in bristle organ development, which are endogenously not regulatory targets of Su(z)2 (Sharp et al., 1994). A similar situation may explain the phenotypes observed in Su(z)2^handC-GFP1.1.

Although many questions regarding the role of Su(z)2 in adhesive device differentiation remain to be answered, we argue that the present work serves as a pioneering work that demonstrates the usefulness of Drosophila to investigate the differentiation of adhesive organs used in locomotion. The availability of mutants and genetic screening techniques will facilitate the identification of genes involved in setal morphogenesis and thereby allow new insights into the functional principles of adhesive organ development.

The handC-GFP1.1 reporter line was generated previously; it carries the complete third intron of the gene hand and drives GFP in, for example, cardiac tissues (Sellin et al., 2006; Popichenko et al., 2007). As this P-element insertion causes a mutation of the Su(z)2 gene (this work), we consider the handC-GFP1.1 line to be a Su(z)2 allele and renamed it Su(z)2^handC-GFP1.1. Additional Su(z)2 alleles, used for phenotypic characterization and transheterozygote analysis, were obtained from the Bloomington Stock Center [P{lacW}Su(z)2^k06344 (BL10619), Df(2R)Exel6062 (BL7544)], from the Drosophila Genomics Resource Center [P{SUPor-P}^KG07068 (DGRC 14511), P{RS5}Su(z)2^5-SZ-3055 (DGRC 125858), P{GSV7}^GS20608 (DGRC 201674), P{GSV6}^GS10178 (DGRC 205205), P{GSV6}^GS10954 (DGRC 202953)] and from the Harvard Exelixis Collection [P{XP}Su(z)2^d01221 (d01221), PBac{RB}Su(z)2^e00448 (e00448)]. w¹¹¹⁸ was used as wild-type control.

To assess the climbing ability of mutant and wild-type flies, two new 50 ml reaction vials (Sarstedt, Nümbrecht, Germany) connected by an outer adapter ring were used. Batches of 20 flies were tapped down to the bottom of one vial and allowed to climb up for 30 s. A light source was placed above the apparatus. After disconnecting the vials, the number of flies in the lower and upper vials were counted. Each experiment was repeated five times. Our climbing test device was built after the countercurrent apparatus used by Benzer (Benzer, 1967). Wild-type flies are highly responsive to light (positive phototaxis) when startled and we found that 100% of the tested wild-type flies climbed successfully in their breeding vials (see supplementary material Movie 1) or in the test apparatus.

The slide friction tester consists of a movable stage with a motorized rotation control (Berthé et al., 2009). The stage is fixed on one side, so that different climbing angles can be chosen. The stage was covered with different polishing paper replicas made of Spurr epoxy resin (Spurr, 1969). The replicas were obtained by filling negative dental wax casts of polishing paper of different asperity sizes (0.3, 1, 3, 9 and 12 µm) with Spurr's resin and subsequent polymerization overnight at 70°C. In order to control the walking direction of the flies, a transparent tunnel was fixed onto the replicas. On each replica, 30 flies of the mutant and 30 flies of the wild-type stock were tested. Prior to the experiment, the wings of each fly were cut off and the flies were put into a tube with wet filter paper for 10 min to remove any potential dirt from their feet. Individual flies were then put into the tunnel and the angle of the stage was continuously increased until the tested individual stopped moving and started to slide down. The maximum angle of the slide friction tester was set to 160 deg. For statistical analysis, two-way ANOVA was used.

For Cryo-SEM (−120°C), samples were mounted on a holder using fluid Tissue-Tek O.C.T. compound (Sakura Finetek, Zoeterwoude, The Netherlands) and frozen in liquid nitrogen. Specimens were sputter-coated with gold-palladium (6–10 nm thickness) using an internal sputter coater of the microscope, and examined in a Hitachi S-4800 scanning electron microscope (Hitachi, Tokyo, Japan), equipped with a Gatan ALTO 2500 cryo-preparation system (Gatan, Abingdon, UK) at an accelerating voltage of 3 kV. The images were analysed using the software Adobe Photoshop (version CS5; Adobe Systems, Germany). All SEM images shown in this article were obtained using the Cryo-SEM method. All length and counting measurements were performed with flies that were prepared using Cryo-SEM. For the statistical analysis of pulvilli length, the Mann–Whitney rank sum test was used.

Contact behavior between adhesive pads of wild type and mutants and glass substrate was visualized with an inverted light microscope (Axio Observer.A1, Carl Zeiss Microscopy, Göttingen, Germany). In the internal reflection contrast mode, the light source is positioned in a way that light is reflected at the interface of direct (real) contact between the glass slide and the object. Zones of direct contact appear as dark spots on the bright background. A cleaned glass coverslip was mounted on the stage and viewed at ×40 magnification (oil immersion). In the first set-up, a living fly was glued with its back to a thin wire and positioned in a way that its legs touched a coverslip. In the second set-up, unrestrained flies were allowed to move freely in a small chamber. The stage was then manually moved vertically and laterally and the contact of spatulae with glass was recorded with a high-speed video camera (Photron Fastcam SA1.1, VKT Video Kommunikation, Pfullingen, Germany) at 6250 frames per second (f.p.s.). The pictures shown in this article are of the unrestrained flies.

For localizing the P-element in the Su(z)2^handC-GFP1.1 line (pH-Stinger backbone; Barolo et al., 2000) by inverse PCR, genomic DNA was extracted, digested with MspI or MboI (New England BioLabs, Frankfurt am Main, Germany), religated and used for PCR. The primers used were Pry1, Pry2, Pry4, Plac1, Plac2 and Pwht1. Amplicons were extracted from agarose gels and sequenced. Primer sequences and a detailed protocol for inverse PCR are available from the Berkeley Drosophila Genome Project (Rehm, 2008).

In order to remobilize the P-element in the Su(z)2^handC-GFP1.1 line on the second chromosome, a standard excision scheme was applied as described previously (Wang et al., 2012). Briefly, the line Su(z)2^handC-GFP1.1 was crossed to a homozygous viable and fertile transposase source in a piggyBac vector (y¹ w¹¹¹⁸; PBac{Δ2-3.Exel}106, Bloomington Stock number 8200). F1 males displaying mosaic eye color were then crossed individually to virgin females of a balancer stock Kr^If-1/CyO. Single F2 males with white eyes and without GFP in cardiac cells were selected and backcrossed to Kr^If-1/CyO virgin females to establish individual stocks.

Northern blots were conducted as previously described (Meyer et al., 2009) with 15 μg of total RNA being loaded per lane and a hybridization temperature of 65°C (20 h). For probe synthesis, total RNA isolated from adult flies was treated with DNase I (Invitrogen Life Technologies, Darmstadt, Germany) according to the manufacturer's instructions and used as a template for cDNA synthesis (SuperScript III Reverse Transcriptase, Invitrogen Life Technologies). Subsequently, a 1 kb fragment was amplified by RT-PCR using the following primers: Su(z)2-fwd (5′-GAA TCG CCC ATG GCC TTC TGC-3′) and Su(z)2-rev (5′-CAC TGG CCA GTG AAT GCG ACC-3′). The resulting fragment was cloned into pGEM-T Easy (Promega, Mannheim, Germany), verified by sequencing and used as a template for in vitro transcription utilizing the ‘DIG RNA labelling kit’ (Roche, Mannheim, Germany).

We thank Martina Biedermann, Kerstin Etzold, Mechthild Krabusch and Werner Mangerich (all from the Department of Biology, University of Osnabrück, Zoology/Developmental Biology) for technical assistance.