The C-type lectin Schlaff ensures epidermal barrier compactness in Drosophila

The stability of extracellular matrices is in general ensured by cross-linking of its components. Previously, we had shown that the integrity of the layered Drosophila cuticle relies on the presence of a covalent cuticular dityrosine network. Production and composition of this structure remained unstudied. In this work, we present our analyses of the schlaff (slf) gene coding for a C-type lectin that is needed for the adhesion between the horizontal cuticle layers. The Slf protein mainly localizes between the two layers called epicuticle and procuticle that separate from each other when the function of Slf is reduced or eliminated paralleling the phenotype of a cuticle with reduced extracellular dityrosine. Localisation of the dityrosinylated protein Resilin to the epicuticle-procuticle interface suggests that the dityrosine network mediates the adhesion of the epicuticle to the procuticle. Ultimately, compromised Slf function is associated with massive water loss. In summary, we propose that Slf is implied in the stabilisation of a dityrosine layer especially between the epicuticle and the procuticle that in turn constitutes an outward barrier against uncontrolled water flow. Summary statement Extracellular matrices adopt a stereotypic organisation for function during development. The lectin Schlaff assists adhesion reactions to ensure compactness of the epidermal cuticle in Drosophila.


Introduction 37
Extracellular matrices (ECM) contribute to tissue shape and function. Their integrity 38 depends on covalent and non-covalent interaction of their components. Collagen 39 crosslinking in the articular cartilage by lysyl oxidases, for example, enhances tissue 40 stability against physical wears (Saito and Marumo, 2010). Another prominent 41 example is the apical extracellular layered network of lipids and proteins that 42 constitutes the epidermal stratum corneum (Harding, 2004;Nishifuji and Yoon, 2013;43 Rogers et al., 1996). A defective stratum corneum in patients suffering different types 44 of ichthyoses provokes a dry and scaly skin (Akiyama, 2017). Lamellar ichthyosis is 45 caused by mutations in the gene encoding the central cross-linking enzyme 46 transglutaminase that introduces covalent glutamine-lysine bonds. Extracellular 47 localisation does not detach and marks the body surface (Fig. 2). 104 Taken together, mutations in slf affect the integrity of the larval body. Especially, the 105 adhesion between the TwdlD-dsRed and CPR67B-RFP domains depends on Slf. We 106 assume that the observed liquid in the egg space of slf mutant embryos is the 107 haemolymph that leaks out because through the loss of its integrity the cuticle has 108 become permeable. 109

Slf is not needed for the function of the inward barrier 110
To further inspect cuticle barrier integrity, we performed a dye penetration assay that 111 we had developed recently (Wang et al., 2017a;Wang et al., 2016). Incubation of 112 wild-type, slf and alas mutant ready-to-hatch embryos with bromophenol blue does 113 not result in dye uptake, while snurstorr snarlik (snsl) mutant animals with a defective 114 envelope (Zuber et al., 2018) do so (Fig. S2). Thus, Slf is not needed for protection of 115 xenobiotic penetration through the cuticle. 116 The cuticle of slf mutant embryos is delaminated. 117 In order to understand the defects at the cellular level, we analysed the ultrastructure 118 of the body cuticle of slf mutant larvae by transmission electron microscopy ( Fig. 1). 119 The wild-type body cuticle is composed of three biochemical distinct horizontal 120 layers, the envelope, the epicuticle and the procuticle. The upper envelope consists 121 of alternating electron-dense and electron-lucid films. The middle epicuticle is a 122 bipartite matrix of cross-linked proteins and lipids. The procuticle contacting the 123 apical surface of the epidermal cell is characterised by a helicoidal stack of chitin-124 protein sheets (laminae). In the cuticle of slf mutant ready-to-hatch larvae 125 unstructured regions of various sizes disrupt the organisation of the laminae. The 126 procuticle is occasionally separated from the above epicuticle. The upper tier of the 127 epicuticle is not smooth. The envelope is continuous and its ultrastructure appears to 128 be normal. In summary, cuticle compactness in slf mutant larvae is lost. 129

Mutations in slf do not affect septate junctions 130
Loss of barrier function is observed in Drosophila embryos that have mutations in 131 genes coding for septate junction (SJ) components (Izumi and Furuse, 2014). To test 132 whether slf mutations affect SJ integrity, we investigated ultrastructure of the SJ in slf 133 mutant embryos by transmission electron microscopy (Fig. S3). In the wild-type larva, 134 SJs connect neighbouring epidermal cells. In the slf mutant larva the SJ 135 ultrastructure is unchanged. The correct assembly of SJs does not exclude that they 136 may have nevertheless lost their barrier function. We performed dye injection assays 137 to analyse SJ barrier function (Fig. S3). When wild-type stage 16 embryos were 138 injected with 10 and 3 kDa dye-conjugated dextran, the epidermal cells retained 139 dextran within the body cavity. In slf mutant stage 16 embryos dextran retention is 140 normal. Hence, we conclude that the loss of barrier function in slf mutant larvae is not 141 due to defective SJ. 142 Slf is a C-type lectin expressed in the epidermis 143 In order to understand the molecular defects caused by mutations in slf, we identified 144 the gene affected by these mutations. The slf gene was initially mapped to the 145 cytological position 25A to 25C on the left arm of chromosome 2 (see flybase.org). 146 By deficiency mapping, we localised the mutations in an interval uncovered by 147 Df(2L)BSC225 containing 10 loci. To narrow down the slf region, we attempted to 148 reduce the number of candidate genes in a transgenic rescue experiment. Due to the 149 cuticle defects observed in slf mutant larvae, we suspected that the factor affected 150 might be associated with the apical plasma membrane or extracellular. A good 151 candidate is CG3244 (Clect27) that was reported to be needed for wing cuticle 152 integrity (Shibata et al., 2010). We recombined an insertion of the Pacman CHS322-153 140E11 (20233bp) that includes CG3244 and the neighbouring gene CG3294, 154 coding for a putative zinc-finger RNA-binding protein to the chromosome harbouring 155 the slf 2L-199 mutation (Fig. 3). Homozygous slf 2L-199 larvae carrying the CHS322-156 140E11 insertion do not display the slf mutant phenotype. In in situ experiments, we 157 detect the CG3244 transcript in the developing epidermis during late embryogenesis 158 when the cuticle is formed (Fig. 3). 159 According to the SignalP software, CG3244 possesses a signal peptide suggesting 160 that it may be secreted. CG3244, a Ca 2+ -dependent lectin (C-type lectin), has 161 recently been proposed to be a target of the transglutaminase that catalyses the 162 cross-linking of proteins in the cuticle (Shibata et al., 2010). We sequenced the 163 genomic DNA of the candidate CG3244 isolated from the two EMS alleles slf IJ83 and 164 slf 2L-199 and identified in each sequence a single point mutation that leads to an exchange of an amino acid (Fig. 3). These amino acids are highly conserved 166 between CG3244 and homologous sequences. A rabbit antiserum produced against 167 CG3244 failed to recognise an antigen in protein extracts from slf mutant larvae, 168 while a 25 kDa protein was present in protein extracts from wild-type first instar 169 larvae (data nit shown). Moreover, we were able to phenocopy the slf-mutant 170 phenotype by RNA interference (RNAi) through the expression of UAS-driven 171 CG3244 RNA hairpin constructs in the epidermis (Suppl. Fig. S1). Thus, mutations in 172 CG3244 are responsible for the slf mutant phenotype described above. 173 The Slf protein contains 231 amino acids and is composed of an N-terminal signal 174 peptide and a C-type lectin domain (Fig. 3). The motif QPD especially within the C-175 type lectin domain is found in galactose binding lectins (Zelensky and Gready, 2005). 176 Closely related sequences, probably Slf orthologs are found in other arthropods. A 177 weak homology is detected to L-Selectins from vertebrates, which actually do not 178 seem to have true counterparts in Drosophila. In order to determine the sugar moiety 179 recognised and bound by Slf, we studied the binding capacity of Slf to mannose, 180 galactose, lactose or N-acetyl-glucosamine (GlcNAc, the chitin monomer) in binding 181 assays using agarose columns exposing the respective sugar. Slf extracted from 182 stage 17 wild-type embryos was able to bind to mannose and galactose but not to 183 lactose or GlcNAc (Fig. S4). 184 Taking all these data together, we conclude that slf encodes the C-type lectin 185 CG3244, which potentially binds extracellular sugars but not chitin. 186

Slf defines a new zone within the epidermal cuticle 187
Loss of cuticle compactness suggests that Slf is a coupling link between cuticle 188 components. In order to examine the cuticular localisation of Slf we generated a C- probably depicting intracellular vesicles. Thus, Slf localisation within the procuticle is 198 necessary for cuticle compactness. We speculate that its accumulation in the apical 199 region of the procuticle may define a new cuticle zone. 200

Slf is required for soft cuticle integrity. 201
In slf mutant larvae, the soft body cuticle is disorganised, the head skeleton, by 202 contrast, that consists of a melanised and hard cuticle is unaffected (Fig. 1) and die within the pupal case. The overall anatomy of their head appears to be 208 normal (Fig. 5). However, the ptilinum, a soft and elastic cuticle that expands to break 209 open the pupa case, is ruptured. We thus reckon that soft cuticle integrity requires Slf 210 function. To corroborate this interpretation, we down-regulated slf activity in the 211 whole body of developing pupa by RNAi (Fig. S5). We observed that the cuticle in the 212 leg joints, wing hinges, ventral abdomen and ptillinum were necrotic. The body parts 213 with the hard cuticle appeared to be unaffected. These flies died in the pupal case or 214 shortly after eclosion. In summary, our genetic experiments suggest that Slf is 215 especially required in the unsclerotised, soft cuticle of larvae and adult animals. 216 Slf cooperates with heme synthesis pathway in dityrosine layer formation. 217 Defects provoked by mutations in slf are reminiscent of those caused by mutation in 218 alas, a gene encoding the delta-aminolevulinate synthase, which initiates the 219 synthesis of heme (Figs. 1 & S1) (Shaik et al., 2012). Is there a genetic and 220 molecular relationship between Slf and heme synthesis pathway? In order to answer 221 this question, we performed a series of genetic and histological experiments. First, 222 we examined embryos double-mutant for alas and slf mutations. The phenotype of 223 these embryos was comparable to the ones provoked by mutations in either of the 224 genes (Fig. S1). Assuming that both mutations represent loss-of-function situations, 225 this observation suggests that these genes act in a common pathway. Consistently, 226 reduction of larval alas or slf expression by RNAi caused a similar lethal phenotype 227 (Fig. S6). Second, we tested whether Slf localisation may depend on Alas function. 228 Using our anti-Slf specific antiserum, we find that Slf localises to the cuticle (Fig. 6). 229 However, the thin L1 cuticle does not allow a more detailed localisation. 230 The phenotype of alas mutant larvae has been linked with the breakdown of the 231 dityrosine barrier (Shaik et al., 2012). Using a DT specific antibody (α-DT), we tested 232 whether the dityrosine network may depend on the presence of Slf. We observed that 233 dityrosine signal intensity is reduced in these animals in the integumental cuticle, but 234 not in the tracheal cuticle (not shown). This suggests that Slf might be either involved 235 in dityrosine network formation or needed for the localisation i.e. stabilisation of 236 dityrosinylated proteins to form a network. A well-known substrate protein modified 237 by dityrosine links is Resilin (Andersen, 1964). We generated a Venus-tagged 238 version of Resilin and co-expressed it with Slf-RFP in the cuticle of L3 larvae (Fig. 7). 239 These chimeric proteins co-localise at the apical domain of Slf. Hence, Slf seems to 240 be associated with dityrosinylated proteins. To further elucidate the relationship 241 between Slf and Resilin, we expressed Resilin-Venus in third larvae with RNAi-242 induced reduced slf expression. We observed that Resilin-Venus is mislocalised in 243 these larvae (Fig. 7). This suggests that Slf might be responsible for either the 244 delivery or the stabilisation of dityrosine-forming proteins to the correct position in the 245

cuticle. 246
A well-known peroxidase involved in dityrosine formation in insects including fruit flies 247 is the membrane-inserted Dual Oxidase Duox (Anh et al., 2011;Edens et al., 2001). 248 Using a-DT specific antibody, we tested the presence of dityrosines in homozygous 249 mutant embryos deficient for duox. In these animals, the dityrosine signal was 250 comparable to the signal in wild type embryos (Fig.6). This suggests that either Duox 251 is not involved in the formation of a larval dityrosine network, activity of maternally 252 provided Duox is enough to catalyse the formation of the cuticle dityrosine network or 253 another peroxidase may compensate decreased Duox activity. 254 Taken together, we conclude that Slf is a part of the dityrosine layer in the cuticle and 255 the localisation of the dityrosinylated proteins to this layer depends on Slf activity. 256

The Slf homologous CG6055 is needed for tracheal air-filling 257
Not only the epidermis, but also the tracheal tubes produce a cuticle that lines their 258 surface (Moussian et al., 2006a). Slf is not present in the tracheal system (Fig. 3). 259 According to the BDGP FlyExpress database (Konikoff et al., 2009), the two Slf 260 homologous C-type lectins CG4115 and CG6055 are expressed in the tracheal 261 tubes. To test whether these two factors may play a barrier role in the tracheae, we 262 knocked-down the expression of CG4115 and CG6055 by RNAi. We expressed the respective UAS-driven stem loops by the tracheal-specific Gal4 driver btl-Gal4. RNAi 264 provoked down-regulation of CG6055 resulted in a subset of larvae that fail to air-fill 265 their tracheae. Failure to air-fill has been repeatedly associated with loss of tracheal 266 barrier function (Moussian et al., 2015;Tsarouhas et al., 2007;Wang et al., 2015). 267 We therefore conclude that reduction of CG6055 function in the tracheae causes loss 268 of barrier in the tracheal system.

Slf is a cuticular C-type lectin 287
Analyses of the Slf protein sequence suggest that it is a secreted galactose-binding 288 C-type lectin. Our sugar binding data confirm the prediction that Slf is able to bind 289 among others galactose. In D. melanogaster, galactose residues are found on side 290 branches of N-glycans and on a tetrasaccharide that links glycosaminoglycans 291 (GAGs) to serine residues of certain membrane-bound proteins such as glypicans 292 and syndecans (Nakato and Li, 2016). Cuticle proteins have not been reported yet to 293 harbour sugar moieties. Moreover, Slf is detected within the procuticle in D. 294 melanogaster stage 17 embryos, especially accumulating at a distinct sheet at the apical border of the procuticle between the two zones marked by the cuticle proteins 296 TwdlD and CPR67b. Based on these data, we assume that Slf is a cuticular C-type 297 lectin contributing to late cuticle differentiation. Presumably, Slf exerts its function by 298 binding an extracellular protein that carries a galactose. In principle, this finding is in 299 line with data demonstrating the Slf (Clect27) is a cuticle protein that is essential for 300 survival and needed for wing formation (Shibata et al., 2010). Moreover, it was 301 shown that Slf is a substrate of the cross-linking enzyme transglutaminase that 302 mediates covalent glutamine-lysine bonds. Down-regulation of transglutaminase 303 expression, however, causes a mild cuticle phenotype compared to the strong slf 304 mutant phenotype. Thus, taken together, Slf is a component of a composite 305 extracellular network including non-essential covalent (glutamine-lysine bridges) and 306 essential non-covalent (galactose binding) interactions. 307 We find that Slf is present in other insects. Thus, the role of Slf in the soft cuticle of 308 other insects is probably conserved. According to information from the beetle base 309 on the putative orthologue of Slf in the red flour beetle Tribolium castaneum 310 (http://ibeetle-base.uni-goettingen.de/details/TC013911), injection of double-stranded 311 RNA into larvae is 100% lethal. A phenotype has not been reported. However, this 312 result underlines that Slf is also essential in other insects than D. melanogaster. 313

Slf function is independent of the envelope 314
Classically, the outermost cuticle layer called envelope has been considered to be 315 the bona fide desiccation barrier. In a recent work, we demonstrated that the 316 extracellular protein Snsl and the ABC transporter Snu contribute to the 317 establishment of the envelope in turn ensuring desiccation as well as penetration 318 resistance (Zuber et al., 2017). The function of Snu is obviously conserved in other 319 insects (Broehan et al., 2013;Yu et al., 2017). The envelope of slf mutant larvae is 320 normal at the ultrastructural level. In addition, cuticle impermeability to xenobiotics is 321 maintained in these larvae indicating that Slf is dispensable for an inward barrier. 322 Furthermore, the procuticle is not disrupted in snu or snsl mutant larvae. Based on 323 these evidences, we conclude that Slf and Snu/Snsl act in different pathways or 324 mechanisms designed to establish a cuticular barrier preventing especially water 325 loss.
Elimination or reduction of Slf function especially affects the integrity of soft cuticle 328 types including the larval body cuticle, the joint cuticle and the ptilinium. By contrast, 329 hard cuticle types are largely unaffected. The major difference between hard and soft 330 cuticles is the presence of an elaborate exocuticle in the hard cuticle that, as at the 331 upper portion of the procuticle, consists of a sclerotised chitin-protein matrix. Based 332 on this histological difference, we hypothesise that the region between the 333 unsclerotised procuticle -called endocuticle in the hard cuticle -and the epicuticle is 334 a region where components are cross-linked either by catecholamines (sclerotized 335 exocuticle) or by dityrosines (soft cuticle). This region is apparently needed to 336 prevent massive water loss through the cuticle. 337 Slf is involved in organising cuticle compactness through production or stabilisation of 338 the dityrosine network 339 Mutations in slf are embryonic lethal. Loss of Slf function entails massive water loss. 340 By fluorescence microscopy, we show that the outer TwdlD-layer of the cuticle 341 detaches from the inner CPR67b-layer of the cuticle in respective ready-to-hatch 342 larvae. In addition, by transmission electron microscopy, we show that the procuticle 343 of these larvae is loose. Thus, Slf is needed for compactness in the procuticle as well 344 as the attachment of the TwdlD-to the CPR67b-layer within the cuticle. 345 The detachment of parts of the larval cuticle from the body is reminiscent of the alas 346 mutant phenotype (Shaik et al., 2012). This suggests that Slf and Alas may 347 contribute to the same structure in the cuticle. Alas is involved in the production of 348 heme that is a co-factor of a yet unidentified oxidase catalysing the formation of a 349 dityrosine network within the cuticle at the end of embryogenesis. We find that the 350 cuticular dityrosine signal is reduced in slf mutant embryos and that the 351 dityrosinylated cuticle protein Resilin is mislocalised in these animals. We conclude 352 that Slf is required either for production or stabilisation of the dityrosine network that 353 constitutes a barrier against water loss (Fig. 8). 354 Similarly, in vertebrates, galectin-3 forms an impermeable 500 nm thick lattice 355 through the interaction with mucins at the surface of the ocular epithelium (Argueso 356 et al., 2009). The presumed association of Slf with galactose-residues in a group of analogous manner stabilise extracellular proteins required for cuticle integrity and 359 barrier function. Slf is, hence, an adapter-like protein that glues different cuticular 360 networks. Overall, we suspect that lectins may play a key role in ECM organisation. 361

Materials & Methods 362
Fly work and microscopy 363 were put on a glass slide into a drop of Halocarbon oil 700 (Sigma) and covered with 392 a coverslip. Cuticle detachment was monitored using a Leica DMI8 fluorescent 393 microscope. For live imaging of third instar larvae, larvae were anesthesized with 394 ether, mounted in halocarbon oil on a glass slide and covered with a coverslip. 395 Fluorescence was observed on a Zeiss LSM 880 microscope. Images were prepared 396 using Adobe Photoshop and Illustrator CS6 software. 397

Generation of homozygous slf mutant clones 398
The slf 2L199 allele was induced on a chromosome carrying FRT (Flipase Recognition 399 Target) sequence (Luschnig et al., 2004). These flies were crossed to flies carrying a 400 lethal mutation on a 2nd chromosome with FRT sequence and expressing Flipase in 401 head driven by the eyeless promoter (eye>flipase). The progeny carried slf, FRT on 402 one second chromosome, FRT on another homologous second chromosome and 403 expressed Flipase in head of developing flies. As a consequence of the Flipase 404 activity, slf homozygous clones were generated in the head of developing pupae. 405

RNA interference 406
To generate flies expressing hairpin RNA against slf (slf RNAi ) in the epidermis of 407 pupae the UAS/Gal4 system was used (Brand and Perrimon, 1993). Flies carrying 408 slf RNAi under the control of the UAS promoter (UAS>RNAi-slf, from NIG-Fly, Kyoto, 409 Japan) were crossed with flies harbouring Gal4 under the control of the promoter of 410 the knickkopf gene (knk>gal4). The progeny eclosing of the pupae was observed. 411 RNAi experiments to suppress the expression of CG4115 and CG6055 were 412 conducted using appropriate UAS-driven hairpin constructs (Dietzl et al., 2007) under 413 the control of the tracheal-specific driver btl-Gal4. 414

Molecular Biology 415
Standard molecular techniques (PCR, sequencing) were applied to identify and 416 characterise the slf gene as presented in figure 3. 417

Author contributions 424
RZ, KSS, FM, HH, AS, NG performed the experiments. RZ, SB, HS and BM 425 analysed data. RZ and BM wrote the manuscript.