ChtVis-Tomato, a genetic reporter for in vivo visualization of chitin deposition in Drosophila

Chitin is generally considered to be the second most abundant biomolecule. It is a polymer of N-acetylglucosamine linked by β-1,4 glycosidic bonds and is synthesized by the enzyme chitin synthase, which exists as a transmembrane protein (Merzendorfer and Zimoch, 2003). It is found in a wide variety of organisms including microorganisms such as fungi, protists and algae, arthropods such as insects, crustaceans and arachnids, and in other invertebrates such as sponges, coelenterates, molluscs and nematodes. Recently it has become clear that chitin is also produced in a number of vertebrates (Tang et al., 2015) and that some vertebrate genomes have recognizable chitin synthases (Zakrzewski et al., 2014). Chitin usually functions as a structural component of cell walls or extracellular matrix. In fungi it is often deposited at sites of rapid growth (Molano et al., 1980; Sloat and Pringle, 1978; Teparić and Mrša, 2013) and in insects it is deposited at the apical surface of epithelial cells as part of the cuticular exoskeleton (Merzendorfer, 2013; Merzendorfer and Zimoch, 2003; Moussian, 2013).

From a morphological perspective, the most complex biological structures that contain chitin are the exoskeletons of insects and other arthropods, which are decorated with a wide range of specialized structures including sensory bristles, hairs (trichomes), ridges and various types of sensilla. In addition to chitin, the insect exoskeleton cuticle is composed of cuticle proteins, lipids, minerals and water (Vincent, 2005). It has a complex layered structure and from one body region to another it varies enormously in its physical properties. These differences are presumably a result of differences in the content and arrangement of molecular constituents but for the most part, this has not been well studied. One difficulty is the large number of cuticle proteins encoded by insect genomes. For example, there are ∼150 cuticle protein genes annotated in the Drosophila genome and more than 200 in some other insects (Cornman, 2010; Willis, 2010). In contrast, there are only two chitin synthases encoded by insect genomes and only one of these functions in the synthesis of cuticle (Arakane et al., 2005; dos Santos et al., 2015; Gagou et al., 2002; Moussian et al., 2005). Hence, there are advantages to studying chitin and chitin synthase as a starting point for trying to understand the genetic basis for the differences between cuticles.

Several different approaches have been used to localize chitin in cells and tissues. Many investigators have stained fixed or living cells using Calcofluor (Sloat and Pringle, 1978), Congo Red (Michels and Buntzow, 2010) or wheat germ agglutinin (Molano et al., 1980; Tronchin et al., 1981). However, these approaches suffer from a lack of specificity as other carbohydrates can also be recognized by the reagents. They have primarily been used on yeast and fungi and in our experience are not well suited for staining insect cuticle. A more specific staining protocol has used a fluorescently labeled protein that contains a CBD. This has been used successfully in fungi, nematodes and insects (Gangishetti et al., 2009; Nagaraj and Adler, 2012; Taylor et al., 2002; Zhang et al., 2005). There are, however, limitations with this staining approach. It is most often used on fixed material, cannot be used for time lapse experiments and, at least for insect cuticle, the ability to stain chitin is progressively lost as cuticle development proceeds (Adler et al., 2013; Nagaraj and Adler, 2012). Successful staining is also sensitive to the degree of fixation and this sensitivity increases progressively as differentiation proceeds. The basis for these limitations is likely decreased accessibility for the tagged protein because of the tightly packed and cross-linked nature of cuticle. Consistent with this hypothesis is the dramatic decrease in the ability to immunostain cellular constituents through developing cuticle. To overcome these limitations we have developed a genetic in vivo chitin reporter for Drosophila. The reporter protein consists of amino and carboxy terminal segments of the Dyl protein, which is a ZP domain protein that is secreted apically and is important for cuticle deposition (Adler et al., 2013; Fernandes et al., 2010; Nagaraj and Adler, 2012). Internally, we placed the chitin-binding domain of chitinase A1 from Bacillus circulans (Watanabe et al., 1994) and a copy of td-Tomato, an extremely bright red fluorescent protein (Shaner et al., 2004). DNA encoding this fusion protein was subcloned into a UAS vector and transgenic flies generated. This reporter proved to be both effective and sensitive at detecting chitin. It enabled us to carry out time lapse experiments and to detect chitin long after this would be possible by standard staining protocols. We constructed a control gene where the CBD was not included, which we used to compare with ChtVis-Tomato. The expressed control protein failed to stain chitin-containing structures, confirming the specificity of the reporter containing the CBD. We also showed that in extracts of imaginal wing discs, the reporter would bind to chitin beads while the control did not. Using the reporter, we identified an unexpected deposit of chitin fibers located between the apical surface of imaginal epithelial cells and the pupal cuticle. These fibers were observed ∼13 h prior to the time we could detect chitin deposition in the adult cuticle and were gradually lost as development proceeded.

The reporter protein was designed to mimic several properties of the Dyl protein (Adler et al., 2013). Dyl is thought to be a single-pass transmembrane protein with the amino terminus outside of the cell. A putative furin cleavage site is found before the putative transmembrane domain, and we previously found that the Dyl protein was secreted and not tethered to wing cells (Adler et al., 2013). We hypothesize that the protein was cleaved at the furin site, which allowed it to diffuse away from the synthesizing cell. In at least one ZP domain-containing protein this cleavage is important for protein activation (Jovine et al., 2004). The reporter contains both the Dyl signal sequence to drive the secretion of the amino terminal segment and the furin site to allow release from the membrane (Fig. 1). In between these dyl-derived sequences we inserted a CBD from the Bacillus circulans chitinase A1 gene (Watanabe et al., 1994), and a copy of the very bright td-Tomato protein (Shaner et al., 2004). To test if the reporter was secreted we expressed both the reporter and GFP using en-Gal4, which drives expression in the posterior compartment of imaginal discs and in stripes in embryonic and larval segments. As expected, GFP accumulated only in posterior compartment cells where it is expressed (Fig. 2A). By contrast, we observed td-Tomato fluorescence throughout the space located between the apical surface of wing disc cells and the peripodial membrane (Fig. 2A) as expected for a secreted protein. We also examined the larval body wall and observed fluorescent stripes expressing both GFP and ChtVis-Tomato (Fig. 2B). This seemed likely to be a result of the reporter binding to larval body wall chitin and hence not diffusing away from the synthesizing cell. As a further test of the specificity of the reporter we compared the fluorescence pattern in the hypoderm of en-Gal4 UAS-GFP/UAS-ChtVis-Tomato and en-Gal4 UAS-GFP/UAS-ChtVis-Tomato-control prepupae (Fig. 2C,D). As expected, both green and red stripes were seen with the ChtVis-Tomato reporter. However, no red stripes were seen with ChtVis-Tomato-Control, which lacks the CBD (Fig. 1). This result confirmed that chitin binding was essential for the stripe fluorescence.

As an alternative method to test whether the secreted reporter would undergo limited diffusion during cuticle formation within a cell, we generated flip-out clones expressing Gal4 that contained both UAS-GFP and UAS-ChtVis-Tomato. Clones were induced during the second day of pupal life and then not examined for another 24 h. This allowed us to label cells and/or sense organs (Fig. 3A-C) and also established that the reporter could be used to visualize chitin at later stages in pupal development (day 3) than is possible using an applied stain. It was apparent that the chitin reporter was located externally to the F-actin in the growing bristle (Fig. 3A-C) as would be expected if the reporter were binding to extracellular chitin. The reporter was also located externally to F-actin in developing wing hairs (Fig. 3D).

In previous experiments where we stained fixed pupal wings (and other pupal epidermal tissues) with a tagged chitin binding protein, hair staining was first detected around 42 h after white prepupae (awp) and by about 44 h the staining was strong (Adler et al., 2013). To test the reporter we induced expression by putting the animals at 29.5°C for 4 h and then observed the pupae by in vivo imaging or by examining stained fixed tissues. Similar results were obtained in both sets of experiments. We observed strong hair fluorescence on the wing and thorax, and strong bristle fluorescence in slightly older animals (Fig. 2E; Fig. 3E,F, arrows). In previous experiments on fixed and stained material we described how chitin is prominently found in bands that run along the proximal distal axis of bristles (Nagaraj and Adler, 2012). This was also seen by in vivo imaging (Fig. 3E, arrow). By contrast, we did not observe any fluorescence in hairs or bristles in animals that were too young to have begun cuticle deposition, confirming the specificity of the reporter (this is described in more detail later). We also observed strong fluorescence of the pupal cuticle that surrounds the pupae (asterisks in Fig. S2A, Fig. S3H, Fig. S4). This was seen regardless of the time when the reporter was expressed. This observation was expected as the secreted reporter would be able to bind to the pupal cuticle being deposited from about 12-18 h awp. Using the reporter in both in vivo imaging and in fixed material we often observed a swirling banding pattern in the pupal cuticle sac that surrounds the wing (Fig. S2A). Similar swirling of fibers in abdominal pupal cuticle was also seen in TEM images (Fig. S2B). These observations seem likely to be a result of the pupal cuticle chitin fiber bundles being less densely packed than in the stronger and thicker adult cuticle, allowing us to see individual fiber bundles.

As a control to confirm that the chitin reporter was indeed binding to chitin, we also utilized the construct that lacked a CBD (Fig. 1). We detected no evident fluorescence of this control protein on hairs or bristles. Indeed, the bristles appeared as dark areas (Fig. 3G, arrow) surrounded by faint fluorescence throughout the liquid between the pupal cuticle and partially deposited adult cuticle. This was true at a variety of developmental stages (Fig. S3J,K, arrows).

As a further and more definitive test that the reporter was detecting chitin, we generated clones of cells that were homozygous for a mutation in kkv, the chitin synthase that synthesizes chitin found in cuticle. Because of the weakness of cuticle lacking in chitin, only animals with small kkv clones are viable. In adult wings, kkv hairs are almost invisible under a light microscope because of a lack of pigment, and the mutant hairs being flaccid and lying directly on the wing blade (Fig. 4C) (Adler et al., 2013). We previously published an experiment where we stained pupal wings with a fluorescently tagged chitin binding protein and observed a loss of staining that was restricted to the mutant cells (Adler et al., 2013). Such experiments are complicated by flattening of the wing cells beginning prior to the start of chitin deposition, and problems with immunostaining as a result of cuticle blocking access of the antibody to the clone marker. To lessen the accessibility problem, we used minimal fixation in experiments to test the specificity of the in vivo chitin reporter. This resulted in less than optimal, but still interpretable, morphology (e.g. hairs sometimes appeared to fall apart during the protocol, resulting in what appeared to be ‘doubled’ hairs). When we expressed the chitin reporter, but did not use immunostaining to mark the clone boundaries, we observed ‘missing’ hairs that were not detected by the reporter (Fig. 4A, asterisks). When we also used immunostaining to mark clone boundaries, we found that this lack of detection of the reporter in the hair was only seen in mutant cells (Fig. 4B,D, asterisks). This established that the chitin reporter specifically detects chitin.

As a test of whether ChtVis-Tomato could bind directly to chitin, we generated extracts of ap>ChtVis-Tomato, ap>ChtVis-Tomato and Ore-R wing discs and incubated the extracts with chitin magnetic beads. We then assayed the proteins bound to the beads by western blotting using an anti-Tomato monoclonal antibody. We found that ChtVis-Tomato bound, but ChtVis-Tomato-control did not (Fig. S5). Thus, we concluded from this test that the CBD present in ChtVis-Tomato is functional and leads to the reporter binding chitin.

A number of studies have used cuticle autofluorescence to analyze cuticle structure (Klaus et al., 2003; Zill et al., 2000; Haug et al., 2011). To determine if autofluorescence was impacting our studies we compared the fluorescence of a variety of ChtVis-Tomato-expressing cells and tissues to wild type. In all cases we found that autofluorescence was negligible compared with the fluorescence of ChtVis-Tomato (Fig. S3).

The previous experiments established that the chitin reporter was able to specifically detect chitin in developing Drosophila adult cuticle. The staining observed could represent the reporter stably binding to chitin or it could represent an equilibrium condition with reporter dynamically binding and releasing. To distinguish between these two possibilities we used the FRAP (fluorescence recovery after photobleaching) technique. If the reporter was stably bound, we predicted that after bleaching, recovery would be minimal as inactive reporter would block access of newly synthesized reporter to the chitin. By contrast, if the reporter was dynamically binding and releasing we predicted that recovery would be rapid and complete as the bleached reporter was replaced by unbleached molecules. When we bleached the reporter on thoracic bristles, fluorescence substantially recovered in the bleached area within 18 s (Fig. 5, Movie 1). However, there was never a complete recovery of fluorescence. This implies that there are two pools of reporter in the bristle; one stably bound and the second continually binding and releasing from chitin (Fig. 5). Similar results were obtained in other FRAP experiments on developing bristles but we did not quantitatively analyze how the results might vary as a function of developmental stage or the position and size of the bleached region along the proximal distal axis.

To determine if the chitin reporter was suitable for long term time lapse observations we examined the developing notum where the chitin reporter was expressed (Fig. 6). In previous experiments where we stained fixed tissue with a tagged chitin binding protein, we first observed chitin staining of bristles around 42 h awp (Nagaraj and Adler, 2012). For a brief time the staining was much brighter proximally but shortly afterwards we observed it in all positions along the bristle shaft. We also noted that hair staining began earlier than 42 h awp. In preliminary experiments we failed to see evidence of bristle fluorescence in pupal dorsal notum at 33 h. In such pupae the bristles could sometimes be detected in confocal optical sections by a lack of fluorescence in a field of faint fluorescence. These resembled images obtained when the control reporter lacking the CBD was expressed. We were able to observe hair fluorescence and there was a hint of bristle fluorescence in the notum of 39.5 h pupae (Fig. 6A,A′). The brightness of the reporter increased rapidly and by 42 h awp the hairs and by 44 h awp the bristles could be clearly observed (Fig. 6B,C,B′,C′). As development proceeded fluorescence intensity increased until reaching a maximum around 63 h (Fig. 6D,D′). In 63 h pupae we also observed reporter fluorescence in the notum cuticle apart from the hairs and bristles. By 80 h awp the brightness of the reporter decreased (Fig. 6E,E′). At this time the adult cuticle begins to accumulate pigment and we suspect the absorbance of both excitation and emission light plays a key role in decreasing fluorescence. The loss of td-Tomato activity because of age or the developing cuticle environment might also be contributing to the decreased fluorescence. Our observations established that the chitin reporter is suitable for long term in vivo imaging of Drosophila pupae.

In previous experiments where we stained fixed pupal wings with a fluorescently tagged chitin binding protein we frequently observed stained fibrous material loosely attached to pupal wings. Their location was not random and they were primarily found in a number of stereotypic locations (see Fig. 7 for examples). This staining of fibrous material could also be seen when the chitin reporter was expressed (Fig. 7B,C). These fibrous deposits were not seen in 25 h pupal wings (Fig. 7A) so they are unlikely to be produced during the deposition of the pupal cuticle. By 29 h they could be seen in pupal wings and remained visible in 44 h pupal wings. Thus, the fibrous deposits could be detected ∼13 h prior to the time we first observed chitin staining of the hairs at 42 h. They could be detected for several hours after the start of cuticle deposition in the hairs but were lost by 54 h awp. We also observed similar fibers near the notum during in vivo imaging experiments in 34 h pupae (Fig. 6A′,B′,C′,D′), and as we followed such pupae we noticed the fluorescence decreased slightly from 39.5 to 44 h and was lost by 63 h awp. In z-sections it was clear that the fibers were apical to the apical surface of the wing cells but below the pupal cuticle (Fig. S4, arrows). Consistent with the fibrous material being chitin, it was not detected by our control reporter that lacked a CBD.

The fibrous material did not appear to be highly dynamic, at least during the time period when it was brightest. For example, in the in vivo imaging experiment shown in Fig. 6 the fibrous material appeared to be rather stable between 39.5 and 42 h. These observations suggested that chitin fibers of unknown function are formed in the space between the pupal cuticle and the wing cells prior to the start of chitin deposition in adult cuticle.

As an additional test to determine if the fibrous material was in fact chitin we tested if it was sensitive to chitinase (see Materials and Methods). In these experiments we expressed ChtVis-Tomato in pupal wings using ap-Gal4. Pupae were fixed prior to the start of hair chitin deposition and the wings dissected. Some of the wings were then treated with chitinase a, whereas controls were simply incubated in enzyme buffer. We imaged the wings at various times after the start of the experiment. The control wings did not show a substantial loss of fluorescence (Fig. 7E,E′,F), with the maximum fluorescence of the reporter decreasing ∼20% over more than 2 h of incubation. By contrast, in the wings treated with chitinase there was a dramatic loss of fluorescence (Fig. 7D,D′,F). This result is as expected if the chitin reporter signal results from it binding to chitin in the fibrous material. We cannot rule out the alternative explanation that proteases contaminating the enzyme preparation reduced the staining by digesting the reporter and not the chitin. However, we think this explanation is less likely as we did not see disintegration of the wing as we have in other experiments where we treated pupal wings with proteases for long periods of time.

The chitin reporter was designed to be used to study exoskeleton chitin. It was for this reason that we used the dyl gene, which functions in the deposition of the cuticular exoskeleton, as a source of the amino and carboxy terminal segments. In addition to the epidermis, the trachea is a second well-studied tissue that forms a chitin-containing cuticle (Devine et al., 2005). We first tested if staining in the trachea by the reporter was specific by comparing white prepupae that expressed either ChtVis-Tomato or the control construct lacking the CBD. Staining of the trachea was only seen when the CBD was present (Fig. S6A,B). Hence, it is also specific for tracheal chitin. We examined the reporter in trachea in more detail in an experiment where we used the btl-Gal4 driver to simultaneously express the reporter and a cytoplasmic GFP. We dissected out trachea from btl-Gal4 UAS-GFP/UAS-ChtVis-Tomato 3rd instar larvae and examined these by confocal microscopy. ChtVis-Tomato labeled the apical edge of the tracheal cells as expected if it were binding to tracheal cuticle chitin (Fig. S6C,E). The taenidial folding pattern of the trachea (Matusek et al., 2006) was clearly revealed by reporter fluorescence (Fig. S6C,E, arrow). Observations of wild-type trachea that were not expressing ChtVis-tomato (Fig. S6D,F) established that tracheal cuticle autofluorescence was not significant in this experiment.

The chitin reporter we described here should be useful for a wide variety of experiments involving cells and tissues that synthesize chitin. In addition to the cuticle synthesized by the adult epidermis, limited experiments indicate that the reporter will also work for examining the synthesis of larval and pupal cuticle and the morphogenesis of the tracheal cuticle. It could be very useful in screening experiments to discover unknown genes that have a function in chitin deposition or in the assembly of cuticle. With the chitin reporter, potential mutant animals could simply be examined live under a fluorescence microscope−a much easier and quicker screen than one that relies on dissecting and staining tissues prior to microscopic screening. Alternatively, the reporter could be used as a rapid way to characterize mutants identified using an alternative phenotype.

The reporter described in this paper was not our first attempt to develop a chitin reporter. Earlier reporters constructed in our lab differed in two ways. Our original reporter used m-Cherry as the fluorescent protein and we used the amino terminal segment of the cuticle protein encoding gene dsx-c73A (Andrew and Baker, 2008) to provide a signal sequence. This first generation reporter appeared to work but we did not extensively characterize it. The second generation reporter was much brighter and more photostable, making it much more convenient to use – particularly for time lapse in vivo imaging. We suspect that this was primarily as a result of the substitution of td-Tomato for m-Cherry but we did not test reporters that contained only one change. There are other possible future changes that might also result in improvements to the reporter. Our FRAP data suggested the protein binds and releases from chitin with a short half-life. This was done at an early time in bristle chitin deposition and it is possible that at a later stage the tight packing of the cuticle would result in the reporter not being able to diffuse away from the chitin it was bound to even if it was released. We suspect that a reporter that contained multiple CBDs would bind more tightly and be less likely to be released and bind again. A stably bound reporter would be preferable for some types of experiments where temporal aspects of chitin deposition were important. Another way it might be valuable to modify the reporter would be to include sequences that facilitate detection under an electron microscope (Gaietta et al., 2002; Lam et al., 2015; Shu et al., 2011). Our data cannot rule out the possibility that ChtVis-Tomato could bind to other carbohydrates, although the lack of staining of cuticle synthesized by kkv mutant wing cells argues that such carbohydrates are not present in fly wing cuticle. Our success in the development and use of ChtVis-Tomato for chitin detection in vivo suggests that it would be useful to pursue similar strategies for detecting other carbohydrate polymers, such as cellulose.

In a number of experiments we noted that the pupal cuticle was labeled by ChtVis-Tomato and that we were able to see spirals of reporter that presumably represented spirals of chitin bundles. Consistent with this interpretation, chitin spirals were prominent in images of pupal cuticle obtained by TEM. In our experience with Drosophila, adult cuticle layers of chitin are found but the density of the cuticle makes the morphology less clear cut. Based on this we suggest that pupal epidermis and cuticle would be a favorable model system for studying chitin deposition in insect cuticle.

We developed the chitin reporter for use in Drosophila but expect that it would be useful in a wider range of organisms. We used the dyl backbone to provide two key components of the system. One was a signal sequence that would guide apical secretion, as chitin and cuticle are secreted apically. The second was the putative furin cleavage site. If the reporter was tethered to the plasma membrane it might interfere with the proper deposition of cuticle. As dyl is conserved in a wide variety of insects, we expect that ChtVis-Tomato would work in other insects. There are dyl homologs in non-insect arthropods such as Daphnia, making it probable that the reporter would also work in arthropods. Although there are dyl homologs in nematodes, they are thought to function in the formation of the hypoderm cuticle which does not contain chitin (Sapio et al., 2005). There is chitin in the egg shell and pharynx (Zhang et al., 2005) but it is not known if one of the dyl homologs is expressed in the tissues that deposit this chitin or if our reporter would be secreted apically in these tissues. Thus, it is possible that a suitable chitin reporter for nematodes might need to be modified from the one we have described.

The detection of fibrous material stained by both the chitin reporter and by a fluorescently tagged chitin binding protein was not expected and to our knowledge has not been described previously. It was lost after treatment with chitinase and thus seems likely to be composed of chitin. It was located between the pupal cuticle and the wing/thoracic cells and was detected in wing preparations ∼13 h prior to the time when we first detected chitin in wing hairs. It was gradually lost as development proceeded. The function of this fibrous material, if any, can only be guessed at. It might serve to sequester chitinase found in the liquid between the pupal cuticle and the epidermal cells and hence help to protect chitin deposited early in adult cuticle deposition. It would not be surprising if chitinase was found there as it could be needed earlier to facilitate the release of the epidermal cells from the pupal cuticle (Turner and Adler, 1995). It is uncertain if this fibrous material is found in insects other than Drosophila melanogaster, although that seems likely. Given that chitin is considered insoluble, its presence in a few locations suggests either that it is synthesized by cells only in those locations or that there is a chitin binding protein that both renders it soluble and targets it to those locations. Further studies will be needed to answer these questions.

The UAS/Gal4 system was used to direct the expression of the chitin reporter (Brand and Perrimon, 1993). In most experiments reported in this paper we generated and examined w, UAS-ChtVis-Tomato/ap-Gal4 pTub-Gal80^ts (McGuire et al., 2004) flies. They were grown for most of their development at 25°C or 21°C, temperatures at which the expression of the reporter is low. White prepuae were collected and aged for the desired time and then shifted to 29.5°C for 4 h to inactivate the ts Gal80 protein, allowing high level expression of the reporter. The pupae were then prepared for imaging. Several other Gal4 drivers were used in some experiments. These include AyGal4 where a heat shock-mediated recombination at FRT sites resulted in the expression of Gal4, ptc-Gal4, en-Gal4 and btl-Gal4. We also used ap-Gal4 without pTub-Gal80^ts in some experiments.

In vivo imaging was done as described previously (Nagaraj and Adler, 2012). Aged pupae were placed onto double-sided sticky tape on a microscope slide. Silicone rubber spacers were placed onto the slide and the pupal cuticle was removed locally or entirely. A coverslip with a small amount of halocarbon oil was placed onto the supports. The halocarbon oil created a good optical connection between the pupae and the coverslip. In some experiments the animals were incubated for various lengths of time at 25°C before imaging. In some of the in vivo imaging experiments the animals were moved back to a moist chamber in a 25°C incubator between imaging sessions. Most of the in vivo imaging experiments carried out in this study were done on the notum (dorsal thorax) as this is both a convenient tissue to examine and it allows us to image the development of hairs (trichomes), bristles and simple thoracic cuticle at one time.

The majority of imaging, and all of the in vivo imaging, was performed on a Zeiss 780 confocal microscope in the Keck Center for Cellular Imaging. A few preliminary experiments utilized a Zeiss 510 confocal microscope.

Adult wings were mounted in Euparal and examined on a Zeiss Axioskop 2 microscope and images obtained using a Diagnostic Instruments Spot-RT camera. This microscope and camera were also used to obtain low magnification images of stained pupal wings.

TEM was done at the University of Virginia Advanced Microscopy Center using standard techniques.

Pupal wings (or thoraces) were fixed with 3.7% paraformaldehyde, dissected in PBS, and then stained using standard procedures (Adler et al., 2004).

Pupae (28-34 h awp) were fixed in 3.7% paraformaldehyde and pupal wings dissected in PBS, then placed into a shallow depression slide in chitinase buffer (50 mM potassium phosphate, pH 6.0). The pupal wings were imaged at 10× magnification by epifluorescence on a Zeiss Axioskop 2 microscope equipped with a Diagnostic Instruments Spot RT digital camera. The same exposure time was used for all wings in an experiment. After the original imaging, the slides were moved to a moist chamber and chitinase (Sigma-Aldrich, C6137, 0.5 units/μl) in buffer was added to the experimental group, but not the control group. At various times after the start of the experiment additional images were taken to follow the loss of the chitin staining fluorescence. The images were analyzed using ImageJ (National Institutes of Health). The pupal wings were floating in buffer in the depression slide and their movement in and out of the moist chamber caused changes in position so that not all images of any particular wing were taken at identical orientations. We analyzed changes over time both for maximum fluorescence intensity and integrated intensity in the fibrous area. Although similar results were obtained for both, we have reported the maximum intensity values as we felt this would be less sensitive to the orientation issues.

Wing discs were dissected from ap>ChtVis-Tomato, ap>ChtVis-Tomato-control and Ore-R third instar larvae. One hundred discs were homogenized in 1.1 ml CBD column binding buffer (500 nM NaCl, 10 mM Tris-HCl, 1 mM EDTA, 0.05% Trition X-100, pH 8.0 at 25°C) and centrifuged to remove debris. Washed chitin magnetic beads (New England Biolabs E8036S), or His-6-binding magnetic beads as a control (Anti-His tag magnetic bead: MBL D291-9), in 50 µl aliquots were incubated with 500 µl of disc extract for 4 h at 4°C. The beads were collected in a magnetic binding rack and washed several times. Proteins were eluted with 100 µl of SDS sample buffer at 100°C for 10 min. The samples were then analyzed by western blotting using an anti-tdTomato monoclonal antibody (1:2000; ORIGENE, TA180009) and goat anti-mouse IRDye 800CW (1:10,000; LI-COR, 926-32210) secondary antibody. The blots were visualized with a LI-COR Odyssey.

pWALIUM10-moe was used as a vector for the reporter. The signal sequence from dsc73 was PCR amplified using oligonucleotides that incorporated EcoRI and NdeI restriction sites. The mCherry sequence was amplified to incorporate NdeI and BglII restriction sites. The CBD of the Bacillus circulans chitinase A1 gene (New England Biolabs) was amplified to incorporate BglII and XbaI restriction sites. The sequences were double-digested overnight using the respective enzymes, purified and ligated at 4°C at a 1:1:1 molar ratio. Subsequent PCR was performed with 1 µl of the ligation mixture and primers specific to the sequence ends (EcoRI and XbaI). This specific band was then purified from agarose gel and double digested. The insert was ligated to double-digested pWALIUM10-moe. This initial reporter plasmid was modified to make ChtVis-Tomato (Fig. S1). The tdTomato sequence (from FUtdTW, Addgene) was amplified with the same restriction sites as previously used for mCherry, and it was then used to replace the mCherry sequence using standard cloning techniques. A signal sequence from dyl was amplified from cDNA with the addition of NdeI and BglII restriction sites. TdTomato-CBD was amplified with BglII and XhoI restriction sites. The dyl tail sequence was amplified, truncating at the putative furin cleavage site with XhoI and XbaI restriction sites. For the control construct, tdTomato was cloned to contain NdeI and BglII restriction sites, and the dyl tail to contain BglII and XbaI sites. The sequences were assembled and subcloned using PCR as described above. Where necessary short linkers were added to keep the gene in frame. The constructs were integrated into the fly genome at the VK00001 site (Venken et al., 2006). The sequences of the final plasmids are provided in Fig. S7 and the sequences of the primers used in the construction are found in Table S1. pWalium-ChtVis-Tomato is available from Addgene (72102).

We thank FlyBase and the Bloomington Drosophila Stock Center, which were essential for our research. We thank the S. Segrist lab (University of Virginia, VA, USA) for the btl-Gal4 stock. We also want to thank the Keck Center for Cellular Imaging where the confocal imaging experiments were carried out and the University of Virginia Advanced Microscopy facility where the electron microscopy was done.