B. anynana butterflies were raised in lab conditions at 27°C, 60% humidity, and 12-12 hrs day-night cycle. Larvae were fed young corn leaves and adults were fed mashed bananas.

Optix CRISPR experiments were performed based on the protocol described in Banerjee and Monteiro (2018). Two guide RNAs were synthesised in vitro to target a region of the CDS and 300 ng/ul of both were mixed with Cas9 enzyme and Cas9 buffer, then microinjected into eggs within 3 hours after laying. Adults were frozen at -20° C and imaged using a Leica DMS1000 microscope.

A fine metal needle was used to pick scales from adult WT and Optix crispant wings. The scales were then mounted on a carbon tape fixed to a SEM stub, platinum coated using JEOL JFC-1600 Auto Fine Coater, and imaged under a JEOL JSM-6701F Field-Emission SEM (voltage 5 kV) by detecting secondary electrons.

Individual scales were placed on a glass slide and immersed in a chitin-matching refractive index liquid (clove oil). The scales were then covered with a coverslip and the absorbance spectrum was measured using an uSight-2000-Ni microspectrophotometer (Technospex Pte. Ltd, Singapore). Spectra with a usable range between 400 and 700 nm were collected using the 20X lens (NA = 0.5) with an integration time of 100 ms and an averaging of 100. Ten samples were measured for each scale type, with measurements taken from three distinct areas on each scale. Measurements were analyzed and plotted using the R package pavo 2 (Maia et al., 2013).

Scales were mounted in clove oil, then imaged using the 20X lens of a uSight-2000-Ni microspectrophotometer (Technospex Pte. Ltd, Singapore) and a Touptek U3CMOS-05 camera. For each scale type, measurements of five scales from one individual were averaged. First, we extracted the grey values along a line, perpendicular to the ridges, for ten ridges, then we did the same for ten crossribs and intercrossrib areas (= proxy for the lower lamina) along a line perpendicular to the crossribs, using the Profile tool in Fiji (Schindelin et al., 2012). Second, we normalized each measurement of grey value against the value of the background, and we converted the grey value in the percent of pigmentation (grey value [white] = 255, grey value [black] = 0).

FIB-SEM was used to measure the lower lamina thickness of scales. Lower lamina and crossrib height were exposed by milling with gallium ion beam in FIB-SEM on a FEI Versa 3D microscope with the following parameters: beam voltage 8 kV, beam current 12 pA, tilt 52°. Image acquisition was performed in the same equipment by detecting secondary electrons with the following settings: beam voltage 5 kV, beam current 13 pA, tilt 52°. Thicknesses were measured using the Line tool implemented in Fiji (Schindelin et al., 2012), and corrected for tilted perspective (measured thickness/sin52°) (Villinger et al., 2012). For each scale type, ten measurements were taken per scale with ten scales sampled from one individual.

Measurements of thicknesses and distances were measured from SEM images using the Line tool implemented in Fiji (Schindelin et al., 2012). To calculate the percentage of the area of the open upper lamina, a large region (∼ 750 μm²) was selected in the center of the scale. The image was converted to 8-bits, and thresholding was applied based on the values of the bright (ridges, crossribs, and upper thin film) and dark (windows) areas of the SEM image. A separate region of interest (ROI) was made for each dark region using the Analyze Particles tool implemented in Fiji, and then the area of each ROI was measured. The total combined dark area of the open upper lamina was converted into a percentage.

For all parameters, repeated measurements were taken per scale with ten scales sampled from one individual. Due to the multilevel context of the datasets, we ran linear mixed-effects (LME) models using the R package nlme (Pinheiro et al., 2023) that allows coefficients to vary with respect to one or more grouping variables. The scale type was treated as the fixed factor, and the scale nested within individual as a random factor. The lack of homogeneity of variances among scale types prompted us to use the varIdent() function in the nlme package. Akaike information criterion (AIC) was used to compare different possible models and determine which one is the best fit for the data. Adjusted p-values for different pairwise comparisons were obtained by the Bonferroni post-hoc test with Tukey contrasts using the R package multcomp (Hothorn et al., 2008).

We previously showed that Optix crispants lost the orange color in the eyespot’s outer ring, but we did not characterize the color of the transformed (brownish) scales nor the type of pigments that they contained (Banerjee and Monteiro, 2023). Here we show that orange scales were transformed into brownish scales that strongly resemble the brown scales in more peripheric rings around the eyespots ( Figure 1B ; Supplementary Figure S1 ). To quantify this color change, we measured the absorbance spectra of the WT orange, WT brown, and Optix orange-to-brown scales. From 400 to 520 nm, WT orange scales absorbed more light than WT brown scales, whereas brown scales became more absorbent at wavelengths above 520 nm ( Figure 1B ). The absorbance of the Optix transformed scales was qualitatively and quantitatively similar to the spectrum obtained for WT brown scales ( Figure 1B ). The absorbance of black and white scales was not significantly affected as expected, as Optix is not expressed in these scales ( Supplementary Figure S2 ). Thus, knocking-out Optix resulted in a change in the nature of the pigments within the orange ring scales, with the brown pigments, likely melanins, replacing the orange pigments, likely ommochromes.

To more precisely determine where the pigment changes localized within the scales, we imaged the three scale types ( Figures 1C, E-G ; Supplementary Figure S3 ). We observed that the pigments were predominantly located in the ridges and the crossribs, and we confirmed these observations by quantifying the percent of pigmentation in the ridges, the crossribs, and the lower lamina for the three scale types ( Figure 1H ). We found that brown scales are more pigmented than WT orange scales, and only slightly more pigmented (and reddish) than Optix transformed scales, indicating some possible mixture of ommochromes and melanins ( Figure 1H ). Moreover, the disruption of Optix led to an increase in pigmentation in the ridges, the crossribs, and the lower lamina, which became darker (and greyer) than the WT orange scales ( Figure 1H ).

In addition to color changes, knocking-out Optix resulted in the loss of a thin cuticular film that is present in the orange scales ( Figures 2A, B ; Supplementary Figure S4 ). This modification made these scales resemble the morphology of WT brown scales ( Figure 2C ; Supplementary Figures S4A-F ). On average, WT orange scales had 28% percent of their windows closed by this thin film. In Optix transformed scales, only 2% of the windows were closed, resembling WT brown scales ( Figure 2D ). The structures of black and white scales, however, were not affected ( Supplementary Figures S4G-Q ). Optix disruptions affect thus, the type of pigments and the morphology of the orange scales in the eyespots.

To examine the morphological differences more precisely between the three scale types, we measured a series of parameters in these scales: the distance between ridges, the distance between crossribs, and the area of the windows (the area in between a set of ridges and crossribs) ( Figure 3A ). The distance between ridges was increased in Optix transformed scales (average d_ridge-ridge = 2055±84 nm) compared to WT orange scales (average d_ridge-ridge = 1824± 149 nm) ( Figure 3B ). The same trend was found for the distance between crossribs with a slightly higher spacing in Optix crispants (average d_{crossrib-crossrib} = 870±104 nm) compared to WT orange scales (average d_{crossrib-crossrib} = 734±105 nm) ( Figure 3C ). The increase in both inter-ridge and inter-crossrib distances led to expanded window areas in Optix transformed scales ( Figure 3D ), resembling the morphology of WT brown scales.

We hypothesized that these three distinct scale morphologies could derive solely from variation in the pigment composition between the three scale types ( Figure 1C ). WT orange and brown scales having both melanins and ommochromes, but in inverse amounts (Matsuoka and Monteiro, 2018; How et al., 2023), and crispant orange scales having primarily melanins, but no ommochromes.

To try and gauge whether specific pigments have specific effects on scale morphology, we 1) examined the morphology of a different scale type, the silver scale, where the function of Optix’ appears to be confined to repressing melanin synthesis, rather than promoting ommochrome synthesis (Prakash et al., 2022b); and 2) examined the morphology of orange scales in previous melanin enzyme mutants, such as yellow, ebony and DDC mutants (Matsuoka and Monteiro, 2018), where ommochromes are expected to be at WT levels, but melanin levels have changed. We examined how specific scale parameters such as inter-ridge and inter-crossrib distances, lower lamina, ridge, and crossrib thicknesses, and window areas, changed across all these scale types with different amounts of each type of pigment. We tested whether the presence of specific pigments (or pigmentation in general) was consistently correlated with specific scale morphologies.

In B. anynana silver scales, which are almost transparent, Optix is required for repressing melanin pigments, and for closing the windows of the scales. When Optix is disrupted, these coupling scales are transformed into brown scales containing melanin and open windows (Prakash et al., 2022b). In order to investigate how the presence of melanins in these scales affects ridge-ridge distances, we measured and compared this distance between WT silver coupling scales and Optix silver-to-brown transformed scales. In WT silver scales ridge-ridge distance is ~2800nm, but in Optix crispants, with extra melanin, this distance is narrower, ~2100nm ( Supplementary Figure S5 ). This indicates that melanin is associated with tighter inter-ridge distances. In orange scales, where Optix represses melanins but promotes ommochrome synthesis, inter-ridge distances are even narrower ~1800nm ( Figure 3B ). This suggests that ommochromes are potentially more powerful ridge distance tighteners than melanins. Correspondingly, when Optix is disrupted in orange scales, replacing ommochromes with melanins, windows become the largest ( Figure 3D ). A similar effect is observed for inter-crossrib distances, which become narrower with increasing ommochrome pigmentation, decreasing window area ( Figure 3C, D ). In summary, the upper surface of scales becomes a tighter grid when there are more pigments in the scale, and ommochromes appear to have a stronger effect than melanins; the upper cuticular film is retained when melanins are absent, or present in very low amounts (as in the silver and orange scales), or when the grid of the upper lamina is tighter (as in the case for orange WT scales).

To examine the effect of Optix on the lower lamina, we measured scale thickness in the three scale types of B. anynana ( Figure 4A ). We found that WT brown and Optix transformed scales, with the darker lower laminae ( Figure 1H ), and also with the highest melanin content, have the thickest lower laminae relative to WT orange scales ( Figure 4B ). This suggests that the presence of melanins, in particular, leads to thicker lower laminae.

To test whether similar effects are visible in the context of a disrupted melanin biosynthesis pathway, we measured the thickness of the lower lamina of orange scales in yellow, ebony and DDC loss-of-function mutants generated in a previous study (Matsuoka and Monteiro, 2018). In yellow crispants (with less melanin), the lower lamina is significantly thinner than WT orange scales, whereas the lower lamina is significantly thicker in ebony crispants (with more melanin) ( Figure 5A ; Supplementary Table S3 ). We found no significant differences in lower lamina thickness in DDC crispants ( Figure 5A ; Supplementary Table S3 ). These results suggest that the thinner lower lamina of orange scales is due, at least in part, to the little melanin present in these scales, relative to the other scales.

To examine the effect of Optix on the morphology of finer structures in the upper surface, we measured ridge thickness, crest-thickness (the only part of the ridge that is visible when surrounding windows are closed by an upper lamina), as well as crossrib height, in the three scale types. We found that in the two darker scales, Optix transformed orange scales and WT brown scales, with the largest amount of melanin and smaller amount of ommochromes, the ridges are thicker but the crossribs are narrower than the lighter orange scales ( Figure 4C ). We found no significant differences in crest-thickness across scale types ( Figure 4E ), but the lighter orange scales have thicker and deeper crossribs than the darker scales ( Figures 4D, F ). These results suggest that the presence of melanins in the ridges leads to thicker ridges, whereas the presence of ommochromes, or absence of melanins in the crossribs, leads to sheet-like crossribs.

To test whether removing melanins from the orange scales accentuates the sheet-like crossribs, we measured the height of crossribs in loss-of-function melanin pathway mutants (yellow, ebony, and DDC) generated in a previous study (Matsuoka and Monteiro, 2018). We found that crossrib height was only moderately or not at all affected in these mutants ( Figure 5B ; Supplementary Table S3 ). This indicates that little melanin is likely present in the crossribs of orange scales.

To test whether removing ommochromes from the orange scales and/or adding melanins impacts the shape of the crossribs, we measured the height of crossribs in Optix knockouts. We found that crossribs became shorter ( Figure 5B ; Supplementary Table S3 ). This indicates that either extra melanin or less ommochromes shortens the crossribs. Given that removing melanins from black scales (with DDC disruptions) leads to very tall sheet-like crossribs (Matsuoka and Monteiro, 2018), we propose that the moderate sheet-like cross-ribs of WT orange scales are primarily due to small amounts of pigments, especially melanin in these scales.

In the present study we examined how Optix, a transcription factor that differentiates the ommochrome-bearing orange scales (Banerjee and Monteiro, 2023; How et al., 2023), alters the color and morphology of these scales when disrupted. We show that in Optix knockouts the orange pigment is substituted by a brown pigment, likely melanins (Wasik et al., 2014; Matsuoka and Monteiro, 2018). This result is consistent with the proposed function for Optix as a switch gene, promoting ommochrome synthesis and repressing melanin synthesis, as previously shown in other butterfly species (Zhang et al., 2017). We show that pigments are primarily localized in the ridges and crossribs, but are also present at low levels in the lower lamina. And we show that pigmentation increased in all scale subcompartments, i.e. ridges, crossribs and lower lamina, following Optix disruption in B. anynana orange scales. We propose that the changes in pigmentation driven by Optix are sufficient to alter how chitin is polymerized within the scale, changing its morphology (see summary Figure 6 ). The observed effects on nanomorphology are summarized below.

We showed scales that express Optix are characterized by a thin lower lamina and by remnants of a thin cuticular film in the upper surface. Optix knockout scales have thicker lower laminae and lose the thin film in the upper surface. Similar concurrent changes in pigmentation and thickness of the lower lamina were identified in Junonia orange scales, when Optix was disrupted (Thayer et al., 2020). Optix was proposed to be a pleiotropic gene that controls pigmentation pathways, and, separately, chitin deposition in the lower lamina (Thayer et al., 2020). Here we propose that Optix may merely act to promote ommochrome and repress melanin synthesis in scales, and that pigments, by themselves, can alter the way that chitin polymerizes to produce the scale’s skeleton.

Changes in lower lamina thickness, not only in an Optix mutant context but more generally, are often accompanied by changes in pigmentation (Prakash et al., 2022a). In the silver scales of B. anynana, where Optix represses melanins, Optix disruptions increase melanin content. We propose that this increase in melanin content thickens the lower lamina, and disrupts the thin film of the scales, opening windows (Prakash et al., 2022a). In addition, our investigation of melanin pathway mutants in orange scales showed that ebony mutants, with increased melanin content, had even thicker lower laminae, whereas yellow mutants, with less melanin, had thinner laminae. These yellow mutants also retained more of the upper thin film that closed the windows (Matsuoka and Monteiro, 2018). Taken together, these data suggest that the presence of melanins, on their own, are sufficient to promote thicker lower laminae, and disrupt upper thin films.

Within the upper surface, we found that Optix keeps ridges thin, but promotes thicker and sheet-like crossribs. We believe that thin ridges and sheet-like crossribs result from the gene’s effect on repressing melanins, whereas the thick crossribs may be due to the gene’s effect on promoting ommochromes. Optix transformed orange scales exhibit thicker ridges, as in WT brown scales, suggesting that the extra melanin in these scales promotes chitin polymerization around the ridges. Melanin pathway mutant orange scales, however, do not show changes in crossrib thickness (Matsuoka and Monteiro, 2018), nor height (this study), compared to WT orange scales, suggesting that little melanin is present in the ridges and crossribs of orange scales. We propose instead that ommochromes promote thicker crossribs, and when ommochromes are missing, via disruptions of Optix, crossribs become thinner.

These results in orange scales support those obtained for black scales, where disruptions of DDC, required for the synthesis of the brown pigment dopamine-melanin, led to melanin loss and to the polymerization of chitin in vertical sheets below crossribs, instead of in round pillars (Matsuoka and Monteiro, 2018).

The upper surface of scales becomes a tighter grid when there are more pigments in the mass of the scale, and ommochromes appear to have a stronger effect than melanins in tightening inter-ridge and inter-crossrib distances. Previous disruptions of melanin-promoting genes showed that inter-ridge and inter-crossrib distances were not much affected in orange scales (Matsuoka and Monteiro, 2018). On the other hand, Optix disruptions, with the loss of ommochromes, produced the largest windows. The link between ridge and/or crossrib spacings, and the changes in pigment levels is still unclear. Buckling has been proposed as a plausible mechanism for the formation of ridges, where the growth of the scale surface meets mechanical constraints (Ghiradella, 1974). We propose that adding or removing pigments in the upper surface of scales impacts cuticle mechanical properties, e.g. elasticity, and this might impact buckling and ridge and crossrib spacing.

Melanization has previously been shown to control the mechanical properties of insect cuticles. In D. melanogaster, loss of yellow affects cuticle melanization and this leads to structural changes in bristles (Massey et al., 2019) and in wing shape rigidity (Riedel et al., 2011). In the red flour beetle Tribolium castaneum, changes in melanin pigmentation pathway genes affect both cuticle pigmentation and sclerotization (rigidity) (Arakane et al., 2009; Noh et al., 2016). A link between pigmentation and cuticle rigidity has also been shown in the moth Bombyx mori (Xiong et al., 2017) and Manduca sexta (Xu et al., 1997; Kerwin et al., 1999; Suderman et al., 2006). Future studies focused on the interaction between pigmentation and the structural properties of cuticle are of interest.

In addition to previous studies on melanized scales (Matsuoka and Monteiro, 2018) and nearly unpigmented, silver scales (Prakash et al., 2022b), our investigation of ommochrome-containing scales in B. anynana brings further evidence that pigments alone might explain multiple features of scale morphology, as previously proposed by Matsuoka and Monteiro (2018). A recent ptychographic X-ray computed tomography-based analysis showed that low pigmentation correlates with high density cuticle, consisting mostly of crystalline alpha-chitin, in the lower lamina of B. anynana and Junonia orythia scales (Prakash et al., 2022a). The authors proposed that pigments, in addition to absorbing specific wavelengths, affect the way chitin polymerizes into more of less organized/crystalline structures. They also proposed that low pigment levels are associated with sheet-like structures, and high pigment levels are associated with rod-like chitin structures.

Such structural constraints might explain differences in nanostructure heights and thicknesses, where chitin polymerization is likely involved. Future dissection of enzymes downstream of Optix, especially those in the ommochrome pathway, will help confirm several of the inferences from this work. Mutations in these enzymes will have fewer pleiotropic effects relative to regulatory genes such as Optix, and will help pinpoint exactly, which aspect of a scale’s morphology is due to ommochrome pigments being present in the mass of the scale. Future work should also examine the role of cuticular proteins in scale morphology. In B. mori, the silencing of several of these proteins led to malformed ridges and crossribs (Liu et al., 2021). Future studies are needed to reveal the interconnected roles of chitin, pigments, and cuticular proteins in determining scale nano-morphologies.