Developmental, cellular, and biochemical basis of transparency in the glasswing butterfly Greta oto

Numerous species of Lepidoptera have transparent wings, which often possess scales of altered morphology and reduced size, and the presence of membrane surface nanostructures that dramatically reduce reflection. Optical properties and anti-reflective nanostructures have been characterized for several ‘clearwing’ Lepidoptera, but the developmental basis of wing transparency is unknown. We apply confocal and electron microscopy to create a developmental time-series in the glasswing butterfly, Greta oto, comparing transparent and non-transparent wing regions. We find that scale precursor cell density is reduced in transparent regions, and cytoskeletal organization differs between flat scales in opaque regions, and thin, bristle-like scales in transparent regions. We also reveal that sub-wavelength nanopillars on the wing membrane are wax-based, derive from wing epithelial cells and their associated microvillar projections, and demonstrate their role in enhancing-anti-reflective properties. These findings provide insight into morphogenesis of naturally organized micro- and nanostructures and may provide bioinspiration for new anti-reflective materials.


45
The wings of butterflies and moths (Lepidoptera) have inspired studies across a variety of 46 scientific fields, including evolutionary biology, ecology, and biophysics (1)(2)(3). 47 Lepidopteran wings are generally covered with rows of flat, partially overlapping scales . 56 In contrast to typical colorful wings, numerous species of butterflies and moths 57 possess transparent wings that allow light to pass through, so that objects behind them can 58 be distinctly seen ( Fig. 1A-H, 8-10). This trait has been interpreted as an adaptation in the 59 context of camouflage, in which some lineages evolved transparent wings as crypsis to  (Fig. 2B,C). We can therefore infer that early into 148 wing development, SOP cell patterning is differentially regulated between clear and 149 opaque regions, which impacts the adult wing scale density and the amount of wing 150 membrane surface exposed in different parts of the wing. projections showed alternating sizes. In the opaque region similar budding scales at a 169 higher density were found, with larger buds corresponding to future cover scales, and 170 smaller, shorter buds corresponding to future ground scales (Fig. 2D,E).

171
By 48 hours APF, scale cell extensions have grown and elongated (Fig. 2F,G). what has been described in other colorful butterfly and moth species, with the ground 188 scales being shorter and wider than the cover scales (Fig. 2G).

189
By 60 hours APF, scale projections are even more elongated (Fig. 2H,I). The  (Fig. 2H). In the opaque region, scales were longer, wider, flatter, and had developed serrations at the tips (Fig. 2I). F-actin bundles extended all the way to 194 the distal tips of these serrations, which is necessary to produce finger-like projections at 195 the tips of scales (6). Phalloidin staining also revealed that actin bundles were arranged in 196 more symmetrical patterns around the periphery of the bristle-like scale morphologies, 197 forked scales showed modified actin organization at the branching points, and actin 198 bundle asymmetry was greatest in developing flat opaque scales, with larger bundles 199 present on the adwing side (Fig. 2H,I).

223
In our TEM cross sections we also observed scale types that appeared more 224 triangular in shape, suggesting that these corresponded to developing forked scales within 225 the clear wing region (Fig. 3D,E). We observed that these scales were ringed by   In all scale types we observed the presence of numerous internal organelles and 243 vesicles, including mitochondria, electron dense vesicles and free ribosomes (Fig. 3, fig. 244 S1). We also observed that the actin bundles contained dense, hexagonally packed F-actin  Ontogeny of wing membrane nanostructures 253 The clear wing regions of G. oto contain nanopillars that cover the surface of the 254 membrane ( Fig 1I, Fig 4A). These nanopillars were previously characterized in adult corneal surface nipple arrays (e.g. 9, 17), which we refer to throughout the text now as 271 "nipple nanostructures", and an upper layer containing pillar-like protrusions, which we 272 refer to as "nanopillars", that featured a more irregular height distribution (Fig. 4F).

273
These results show early subcellular processes of developing nanopillars within the 274 clear wing region, which arise distal to microvillar extension in epithelial cells. To test this hypothesis, we extracted the surface layer of G. oto clear wing regions 294 with either hexane or chloroform and analyzed the chemical composition by gas 295 chromatography-mass spectrometry (GC-MS). We found that the chemical profile 296 generated by both hexane and chloroform extracts yielded similar results (Fig. 5D). In all 297 extracts, we identified two straight-chain alkanes that made up approximately 2/3 of the 298 compounds detected: 41.64 ± 5.75% pentacosane (C 25 H 52 ) and 23.32 ± 5.35% 299 heptacosane (C 27 H 56 ) ( Table S1). The remaining compounds were primarily composed of   306 To address whether the wax-based nanopillars play a role in wing reflection, we measured 307 the reflectance spectra of untreated and hexane-treated wings (Fig. 6). Additionally, we 308 measured nanostructure geometries and membrane thickness from wing SEM cross 309 sections (n = 6), and determined the average distance between two nanostructures as d = low total diffuse reflection of about 2%, which is in line with previous reflectance 321 measurements of this species (Fig. 6F, 10). By contrast, the hexane treated wings without 322 the upper layer of wax nanopillars had about 2.5 times greater reflectance relative to the 323 untreated wings, and generated an iridescent thin film spectra, even though they harbored 324 dome-shaped nipple nanostructures (Fig. 6D,F).

325
For simulated data, the overall reflectance ratio of the hexane treated wing to that 326 of the untreated was approximately three, similar to experimental reflectance data ( Fig.   327 6F, Table S2). Most importantly, the simulated results for the untreated wing with wax-328 based irregular nanopillars make reflectance more uniform across wavelengths, which 329 reduces the iridescent effect of the wing membrane. Finally, we simulated a thin film 330 membrane without any nanostructures, which showed reflectance (averaged from all 331 wavelengths) of the membrane itself to be 8.81 ± 3.46%, whereas the treated and 332 untreated wing reflections were 5.78 ± 2.82% and 1.93 ± 0.77%, respectively (Fig. 6F).

333
While treated wings harboring dome-shaped nipple nanostructures reduced the overall 334 reflectance relative to the membrane only, their effect was not strong enough to reduce 335 reflectance spectra oscillation. The wax-based irregular nanopillars on top introduced a 336 more gradual transition between refractive indices to lessen the oscillation by 337 approximately five-fold, in addition to reducing overall reflection (Fig, 6F). Additionally,   In the present study, we found that G. oto serves as an excellent model to study protein, Fascin, connects filaments together into hexagonally packed bundles. Our TEM 391 of actin bundles, along with previous studies, support a similar mechanism of hexagonally 392 packed F-actin bundles in Lepidoptera (Fig. 3, fig. S1) (6, 7). 393 In animal cells, microtubules have been frequently observed in arrangements

418
Chitinous wing membrane has a higher refractive index than air, so as a 419 mechanism that reduces glare, some clearwing species have evolved sub-wavelength anti-420 reflective nanostructures (9, 10). In this study, we identified the early developmental 421 processes of nanostructures that arise in the wing epithelium. We also note interesting   Scanning electron microscopy 541 We cut 2mm square pieces from dry wings, coated them with a 10 nm layer of gold using 542 the BIO-RAD E5400 Sputter Coater, and imaged with a Hitachi TM-1000 SEM at 5 kV.

543
Top-view and cross section SEM images were analysed with ImageJ 1.52 to measure 544 membrane thickness and nanostructure dimensions (n = 6).

577
Chemical data processing was carried out using the software "Enhanced Chemstation" 578 (Agilent Technologies, USA). We retained peaks with abundances greater than 0.25% of 579 the total and compounds were identified according to their retention indices, diagnostic 580 ions, and mass spectra, which are provided in Table S1. For some peaks, it was not 581 possible to narrow the identity to a single specific compound because (1) some low 582 abundance substances produced poor quality mass spectra, (2) multiple compounds could  Table S2. The reflectance of the wing membrane before and after chemical treatment by hexane was 599 analytically modeled using effective medium theory and transfer matrix method (10, 18). 600 First, the effective volume fraction of the nanoprotuberances before and after the chemical 601 treatment were based on measurements taken from SEM micrographs of the wings. We