View all hidden authors and organizations

Edited by Lynn M. Riddiford, University of Washington, approved on October 22, 2021 (reviewed on June 29, 2021)

Many organisms exhibit functional micro- and nano-scale materials with structural definitions and properties, challenging synthetic manufacturing techniques, but we know very little about the process of their formation. Using butterfly scales as a model system for functional biological materials, we established a timetable for the formation of scales and quantified the relevant structural parameters for the development of painted butterflies. We overcome the challenges of previous efforts by directly imaging structure formation in living organisms, which allows us to continuously observe the fine details of evolving wing tissue and single-scale cells. The visualization of the formation of scale structures in living butterflies forms the basis for modeling underlying biomechanical processes and opens the way for their translation into advanced manufacturing strategies.

In the process of metamorphosis, the butterfly's wings grow hundreds of thousands of scales. These scales have complex microstructures and nanostructures. These scales determine the optical appearance, wetting properties, thermodynamic properties and aerodynamic behavior of the wings. Although the functional properties of scales are well known and proved to be desirable in various applications, the dynamic processes and time coordination required for many structural features of engraving scales are still poorly understood. The current knowledge about scale growth mainly comes from in vitro studies of fixed scale cells at discrete time points; in order to fully understand the formation of scales, it is important to characterize the morphological changes over time during the entire development process. Here, we report the continuous, in vivo, label-free imaging of squamous cells grown in Vanessa cardui using speckle-related reflection phase microscopy. By capturing time-resolved volume tissue data and nano-level surface height information, we established the morphological timeline of wing scale formation and gained quantitative insights into the potential processes involved in the pattern formation and growth of scale cells. We determined the early differences between the overlay and ground-scale patterns on the young wing, and quantified the geometric parameters of the growth-scale features, indicating that surface growth is critical to the formation of structures. Our quantitative and time-resolved in vivo imaging of butterfly scale development provides a basis for decoding the processes and biomechanical principles involved in the formation of functional structures in biological materials.

The functional structure of butterfly wing scales is formed during pupal development: scale cells grow protrusions, which serve as templates for finely carved nano-scale cuticle morphology (1⇓ –3). By adjusting the shape of these scales, the butterfly produces a unique visual appearance (4⇓ ⇓ –7), ensuring thermal regulation (8) and water resistance (9), or producing beneficial acoustics (10) and aerodynamic effects (11). The interdisciplinary interest in the functions of these materials has led to scientific progress in the comprehensive evaluation of the multifunctional material properties of scales (12), the design of next-generation biomimetic functional materials (13, 14), and the identification of key genes in patterns and structural colors. . 15⇓ ⇓ ⇓ –19), and the evaluation of the impact of ecological factors on biodiversity (20, 21). Although the enviable functions of butterfly wings depend to a large extent on the precise structure of wing scales, little is known about the dynamics, processes and phenomena involved in scale development (22).

Each scale on a butterfly's wing is formed by a single cell, which secretes a chitin cuticle that forms a single-celled exoskeleton. In many butterflies, these scales are further organized into rows of alternating cover and ground scales (1). The mature scales of the painted ladybug (Vanessa cardui) are examples of skeletal scale blueprints, which are widely reflected in the simple and complex wing scale morphology found in Lepidoptera (Figure 1A-D). Generally speaking, the upper surface of the scale is composed of ridges extending along its length; these ridges are composed of overlapping flakes connected by transverse ribs (23). The supporting trabeculae bridges the upper feature and the lower scale surface, which is essentially a thin layer with a thickness of the order of 100 nm. The rich and varied scale morphology of other butterflies and moths may be considered as a modulation of the basic structure found in this general scale structure. Therefore, the easy-to-breed V. cardui is a favorable model system for in-depth understanding of the biological formation process and mechanical phenomena of functional micro- and nano-structures (15, 24, 25).

Image the structure of fully formed butterfly scales. (A) Painted female butterfly, V.cardui. (B) Optical micrographs of orange and black wing scales. (C) Scanning electron micrograph of a single adult scale, with ridges extending along the length of the scale. (D) The scale refers to the magnification, showing that the ridge is composed of stacked sheets and connected by transverse ribs. (E) Adult-scale volumetric images obtained by speckle-related reflection phase microscopy (red, top of volume data stack; green, bottom). (F) A single phase data slice of the same scale. (Scale bars: B, 200 µm; C, 20 µm; D, 5 µm; E and F, 20 µm.)

The key insights into the formation of these structures come from the analysis of anatomical and stained wing tissues at discrete developmental time points. About a century ago, the sequence of cell division, scale protrusion, growth, and ridge formation was recorded in the flour moth (26). Since then, electron microscopy has been used to elucidate the nanoscale structure in wing tissue and provide a glimpse of the growth of the stratum corneum on squamous cells (27⇓ –29). Groundbreaking research emphasized the optical function of scales and provided hypotheses for the formation of flakes on the ridges of scales by mechanical corrugation (30) and the formation of three-dimensional (3D) structures by the internal membrane template (23). This basic principle has been used to explain how gyroscopes and other cubic structures are formed on a scale (31). Recently, confocal imaging allows more careful examination of material distribution, albeit in fixed-wing tissue due to the lack of endogenous labeling methods in Lepidoptera. These studies described the signal factors involved in the formation of scales (32), explored the role of actin in the formation of scales and ridges (24), and quantified the spacing of actin bundles and chitin during and after development. Distribution (25). Together with discrete snapshots of wings fixed at different developmental stages (33), the cuticle structure of mature scales may suggest their formation: in some adult scales, the internal tops increase in size according to the length of the scales, which may indicate the onset and growth of time ( 34).

Although these time-discrete imaging works provide a glimpse of scale development, only by continuously observing the spatiotemporal progress of living-scale cells can we fully understand the process of scale structure formation (22). Recent exogenous fluorescence imaging of live lepidopteran pupae captured the initial protrusions of young squamous cells and scales (35, 36). Despite this progress, visualizing the subcellular characteristics of living cells throughout development remains an unresolved challenge due to the complications associated with tissue imaging with heterogeneity and significant micro- and nano-scale refractive index changes . In addition, long-term imaging using fluorescence technology is susceptible to photobleaching and photodamage; in addition, genetic constructs that are fluorescently labeled in living organisms are still limited for butterflies.

Here, we report the continuous, in vivo, label-free imaging of the developing scale cells in the living V. cardui butterfly using speckle-related reflection phase microscopy (see Figure 1 for comparison with scanning electron microscopy data). This quantitative phase imaging technique provides a general strategy for observing the growth of functional materials in the body, with high temporal and spatial resolution. We have captured the critical moments in the formation of Lepidoptera structures in organisms on a continuous timeline. In particular, we determined the two-step process of tissue patterns in the early epithelial sheets and quantified the morphological changes that occurred as the scale cells grew on different length scales. The insights into continuous imaging of scale formation form the basis for understanding the biomechanical processes involved in functional stratum corneum morphogenesis.

In order to image butterfly wing scales during metamorphosis, we developed a sample preparation and maintenance protocol that enables quantitative phase imaging in vivo (Figure 2). We use various surgical techniques (Figure 2A and B; see Materials and Methods and Reference 35 for time and technology) by replacing part of the pupae's cuticle with glass windows during development to obtain optical access to the wing tissue. Even at a macro level, the development of wing tissue can be monitored through the observation window. Initially, the wings are translucent epithelial flakes; as the scales produce a chitin cuticle, the wings become reflective and eventually show a mature pigment pattern (Figure 2 BE). For long-term quantitative observations with high spatial resolution, speckle-related reflection phase microscopes (37, 38) (Figure 2F and SI appendix, Figure S1) were used to image the optically exposed wings through the windows in the pupae. This interferometric method produces phase and amplitude data from live and unlabeled specimens, has excellent axial slices, and rejects out-of-focus information (Figure 2G and I; SI appendix, Figure S2; and Movie S1). The amplitude data captures the change in refractive index, which is usually related to the material interface (Figure 2G). By scanning the depth of the tissue, three-dimensional images of living bodies, developing scales and wing tissues can be reconstructed. The volume is 75 × 75 × 200 µm3, the maximum lateral resolution is 490 nm, and the maximum axial resolution is 1.03 µm. We visualize this 3D volume amplitude data by color-coding each data slice according to the height in the image volume (Figure 2H and Movie S2). The phase data captures the height of the material interface within each optical section (SI appendix, Figure S2). The direction of the phase gradient encodes the local slope of the scale surface, revealing the critical scale feature (Figure 2I). The individual line profile quantifies the scale surface height, is sensitive to changes in the 10 nanometer scale, and provides quantitative insights into the formation of ridges and flakes throughout the scale development process (Figure 2J and K).

Optical window and imaging development scale of intravital microscope. (A) Create an optical window in the pupa: Lift a small piece of cuticle and forewings to expose the hindwings, and seal the exposed area with a cover glass and biocompatible adhesive (blue). (B) The exposed wing is developed at 2%. The red asterisk marks the position in A. (CE) Optically reachable wing area at 3%, 73% and 100% development (a few minutes before emergence). (F) Schematic diagram of the in-vivo imaging device. QWP, quarter wave plate; PBS, polarization beam splitter. (G) A single piece of deconvoluted reflected light amplitude data, showing the tip of the forewing scales in 83% of the pupae (this specimen is 8.12 days old). (H) A stack of volume amplitudes showing overlapping scales (red, highest slice; green, lowest slice, 6 µm in depth). (I) Visualize micro- and nano-scale features on a scale through the phase gradients related to the amplitude data in G. The colored cones indicate the direction of inclination (red, to the south; blue, downhill to the north). (J and K) The surface profile along the line indicated in G shows the height of the ridge (J) and the ridge layer (K). (Scale bars: BE, 1 mm; GI, 20 microns.)

We track the development of individual specimens from the first few hours after pupation, until the organism begins to close; the duration of the pupal stage (100% development) is usually about 10 days, up to 2 days, and the development time depends on the wing Scale type and location (24, 39). We observe the wing tissue as it changes from a simple folded epithelial monolayer to a mature wing with a fully formed scale structure (Figure 3; Movie S3; and SI appendix, Figure S3). In about 1% of development, morphologically homogeneous epithelial cells are densely packed on the surface of the wing (Figure 3A); they are more sparse internally through the evolving network of intercellular connections (SI Appendix, Figure S4 and Movie S4) Ground connections, these networks have previously been described as feet and cytokines (36, 40). It is easy to see that the mitosis of generalized epidermal cells occurs in the plane of the tissue (movie S4). Then, select cells-squamous precursor cells-to expand in volume (red area in Figure 3B). These cells undergo two divisions: after the first division, one daughter cell degenerates, while the other continues for the second division, producing a squamous cell and a fossa cell (SI appendix, Figure S5) (26). The scale unit protrudes through the socket, away from the surface of the wing (Figure 3C). Although the raised membrane is initially rough—perhaps due to the formation of the underlying microvilli from the upper stratum corneum (27, 29)—the surface quickly becomes smooth (SI appendix, Figure S6). Then the scale expands (Figure 3D), and finally reaches its final length and width before 60% of the pupae develop. When the scale reaches its final length, the leading edge splits into fingers, and the longitudinal structure of the ridge appears (Figure 3E). In previous studies, scanning electron microscopy showed that the anterior stratum corneum-a matrix composed of chitin and protein-grows on these ridges (27, 29). The ridges become more defined and form flakes, while the lower flakes gradually expand to cover all areas except the edges of the lower surface (Figure 3F).

Continuous imaging of scale growth of individual V. cardui pupae. (A) The pupa develops 5% of the wing tissue, showing cell division before the precursor cells differentiate. (B) Enlarged, raised precursor cells (red, surrounded by dashed lines) are identifiable at 15% development. (C) Scales (two highlighted by dashed outline) have begun to grow from the nest on the wings (one highlighted by dashed outline), and the development rate is 34%; coverage scale (arrow) and ground scale (arrow) are easy to distinguish . (D) Scale expansion and the beginning of finger formation at 44% development. (E) Scaled to the final size with articulated fingers at 62% development. The longitudinal structure is faintly visible. (F) The scale bar shows the lower level (green area) below the 99% developed ridge (red area). (G) Timeline of important events observed in developmental tissues, allowing for changes between wing positions. Shown below G are representative color bars representing the volumetric image depth of all images: 0 µm to 6 µm (A, B, and D–F) and 0 µm to 16.8 µm (C). (Scale bars: A and B, 10 µm; C–F, 20 µm.)

By tracking the gradual changes in scale growth over time, we can clearly define the timing of specific developmental events in individual pupae and begin to understand the process that guides morphogenesis (Figure 3G). Here, we have solved two aspects of wing-scale development: the spatial pattern of mesoscale precursor cells in wing tissue and the evolution of ridge spacing and height during scale growth.

In V. cardui and other butterflies, the scales are arranged neatly, and the ground and covering scales alternate. The first step in this model is attributed to lateral inhibition via the Notch signaling pathway, which leads to loose rows of isolated precursor cells. This attribution is based on comparison with the development of Drosophila bristles, where Notch signaling between adjacent cells creates a feedback loop, causing spatially isolated low Notch cells to become bristly precursor cells (32, 41, 42). Similarly, in the wings of juvenile lepidopteran pupae, cells with low Notch expression were observed in a loose row pattern (32). Although cell migration and rearrangement have been proposed as hypotheses (32, 40), the way in which scale precursor cells organize into their final neat row pattern has not yet been determined. Little is known about when and how the precursor cells transition to the ground or coverage.

Our data in V.cardui showed that a portion of epithelial cells differentiated into loose large, isolated precursor cells (Figure 4A), consistent with previous observations of early morphological differentiation (32, 39, 40). Although these precursor cells are only roughly aligned initially, they will then move to more defined rows (Figure 4B and C) without touching each other. Then, the smaller cells located between the precursor cells will also differentiate and grow to the size of the larger precursor cells (Figure 4C and D and Movie S5). Since the first and second groups of precursor cells alternate in a given row, each group must produce coverage or ground scale. Once the scales grow up and the shape of the ground and covering scales can be distinguished, we can track them in time to determine their relative positions. The distal position and later appearance of the second group indicate that they are ground-scale precursor cells, and the first group are overlay-scale precursors (SI appendix, Figure S7).

Squamous precursor cells appear in two stages. The coverage-scale precursor cells (raised red spheres) are labeled 1, 2 and 3, and more than 13% to 18% of the development is tracked in a single pupa. (A) The precursor cells are initially aligned roughly in rows. (B) The same precursor cell moves slightly. (C) A new precursor cell (arrow "a") begins to expand between cell 2 and cell 3; all precursor cells move in a larger direction. (D) The new ground-scale precursor cells (arrows) are now the same size as the old cells; all the precursor cells are arranged in a row. (A'–D') Schematic diagram of precursor cells (magenta). Please note that the borders of surrounding cells (green) are not very obvious anywhere in the data, they are drawn for illustrative purposes only. Volumetric image depth: 0 to 6 µm. (Scale bar: 20 µm.)

Therefore, the formation of densely arranged scales is mainly achieved by the differentiation of the second group of scale precursor cells arranged alternately with the first group in space. What determines the spatial positioning and differentiation time of the second group of precursor cells? During the development of the Drosophila bristles-which are often described as homologous to the lepidopteran scales (24, 32)-the lateral inhibitory behavior of Notch kinetics can produce temporary striped patterns of continuous cells that evolve into rows of disparate patterns. Adjacent bristle precursor cells (42). However, our data shows that the first group of scale precursor cells are no longer adjacent in V. cardui, similar to the Notch pattern on Heliconius erato butterfly (32).

In order to determine the geometric constraints and infer the biomechanical processes that affect the development of scale morphology, we tracked and quantified the evolution of various structural parameters in individual butterflies (Figure 5; please note that the precise time of scale development may vary, and the scale map The developments in 5 are slightly earlier than those shown in Figure 3, but follow the same event progression). A relatively short time window, approximately 35% to 40% of development (approximately 100 hours to 114 hours), contains many key moments for scale formation: scales reach their maximum size, fingers are formed, and ridges appear (Figure 5A and B). The length and width of the scales develop on different timelines, and even before the scale reaches its terminal width, the length shrinks slightly (Figure 5C and D). During most of the scale expansion process, the already very thin scales will gradually become thinner (Figure 5E). However, despite this anisotropic growth, estimates of scale volume and surface area will increase simultaneously (Figure 5F).

Quantification of scale shape evolution and ridge growth. (A) Scale morphology at 35% pupal development. The volume data (top) shows a medium scale. The height profile (h) of the phase data (middle) from the area marked in the volume data and the phase data (bottom) taken along the white dashed line represents a relatively smooth scale surface. (B) The pupae develop 41% scale morphology. The scale is obvious in the volume data. The phase data from the marked area (middle) represents the regular spacing of the original ridges, with a height of approximately 100 nm (bottom contour extracted from the dashed line). (CF) Measured and average scale shape parameters during development: length (C), width (D), thickness (E) and volume (red) and surface area (black) as a percentage scale of the maximum value estimated from the bounding box ( F). The shadow time span is specified in H. (G and H) Ridge parameters during development: the spacing between forming ridges at multiple scales (G). Each curve connecting the data points shows continuous measurements on a single scale. (H) Characterization of ridge height (purple) profile and ridge periodicity (blue-green). The bar represents SD; the curve profile qualitatively illustrates the development of general characteristics. The white background indicates the relatively smooth time span of the scale surface; the dark gray shading indicates the time span of forming irregular structures and changing the spacing distribution; light gray indicates the time span of determining the periodicity and height growth of the ridge. The data in AF and H are from a single pupa; for G, different pupas are used to track ridge development on a single scale over a longer period of time. The error bars correspond to the measured values ​​of the three scales. Volumetric image depth: 0 to 6 µm. (Scale bars: A and B, top, 20 µm; A and B, middle, 10 µm.)

In about 39% of the development, a long strip of material extending along the length of the scale becomes visible in the amplitude data (Figure 5G). Since the phase data shows a smooth scale surface when these features first appear, these spars are likely to be actin bundles, which will eventually form a template for the ridges, as described previously (24, 25, 28). As the scale widens, the spacing of actin bundles widens until it reaches a 1.8 µm spacing. After a while, the surface of the membrane was no longer smooth, and it was difficult to determine whether the longitudinal stripes in the amplitude data were actin as a ridge template or the ridge itself. Previous work has shown that the spacing of actin bundles is closely related to the spacing of epidermal ridges and appears to be unchanged during 40% to 95% of development (25). We tracked this spacing on various scales in continuous data, and confirmed that after the scale surface is no longer smooth, the periodic spacing on each scale does remain unchanged during most of the ridge development process (for the measured scale , At least until 89% of pupae develop).

Our phase data reveal how the surface morphology changes at the beginning of ridge formation (Figure 5H and SI appendix, Figure S8). During most of the time when the scales expand, the upper stratum corneum on the surface of the scales is smooth. After the scale reaches its maximum length but before it reaches its full width, the onset of ridge formation appears to begin at the end of the scale expansion (approximately 37% of the development of this specimen). The surface height will fluctuate, but the spacing on the surface of the scales is irregular and uneven. This heterogeneity in space and time may be due to the initial changes in the stratum corneum. Then, in about 39% of the development, the periodicity becomes very regular and keeps regular as the height of the new ridge increases. The phase and amplitude data together show that from the earliest moment of ridge appearance and growth to the later stage of scale development, most of the fine structure of the ridge has been established, and the ridge spacing remains almost unchanged; we have not yet determined whether the 89% development through eclosion and scale drying will occur. There is any change in spacing.

Other studies have shown that ridges appear between regularly spaced actin filament bundles (25, 28), which are necessary for proper ridge formation (24). As the ridges grow, they form a fine layered structure. A hypothesis established nearly 40 years ago proposed that laminar formation is a wrinkling phenomenon driven by the required stress caused by the mechanical flexion of the epidermal layer and the reduced distance between the actin bundles (30).

However, our data indicates that the ridge spacing remains constant after the initial ridge appears, which means that the actin bundle spacing is unlikely to decrease and therefore cannot drive ridge formation. Nevertheless, buckling is still a reasonable mechanism for the formation of ridges and flakes, possibly due to the growth of the surface of the scale that is potentially mechanically constrained. The growth surface area of ​​the membrane and the deposited stratum corneum may be limited by the space of the actin bundles, resulting in out-of-plane flexion; or, the differential growth between the membrane and the stratum corneum may cause stress, which induces the appearance of ridges (43). Further research is needed to reveal the biomechanical phenomena behind the interaction of actin, cell membrane, and deposited stratum corneum that drive the formation of squamous ridge structures.

Using speckle-related reflection phase microscopy and lepidopteran surgical techniques, we demonstrated the label-free continuous visualization and quantitative characterization of the micro-scale structure formation of living lepidopteran insects, and obtained information about the development of cellular tissue and subcellular characteristics during the entire pupal development process. Quantitative insights. At the tissue level in the early stages of development, we observe a continuous pattern of precursor cells covering and on the ground scale. Our observations raise questions about the signal transduction that differentiates the terrestrial scale, because previous work showed that the first group of precursor cells exhibited low Notch and inhibited the formation of transcription factors on the scales in its neighbors (32, 42), including the first Two groups of precursor cells. Further research is needed to understand how ground-scale patterns between coverage scales are organized. At the single-cell level, our quantification of the spacing and height of scale ridges shows that the ridge spacing remains constant from the first appearance of about 47% of development to at least 89% of development. This suggests that the formation of ridges is unlikely to be caused by the reduced spacing of actin bundles assumed in earlier work (30), but may be driven by surface growth and flexion when the actin bundles are constrained. Future studies using our imaging methods may further characterize the growth of scale materials and provide information for the quantitative mechanical models required to determine the biophysical mechanisms behind the formation of scale ridges and stripes. We expect that our imaging strategy may accelerate the development of other Lepidoptera species and even other purposes, because it bypasses the need to develop endogenous markers or continuous exogenous markers. The formation process of the Lepidoptera structure of V. cardui may have obvious similarities with other Lepidoptera insects. The comparison between the scale structures of different species and their respective development schedules will help determine the processes responsible for creating and fine-tuning structural features. Future work on pushing the resolution limit of the proposed phase imaging method, quantitative mapping of refractive index distribution, and standardization of key system components may promote the implementation of our method in other studies of biomaterial formation and cell and tissue development . In addition, our in vivo imaging can be combined with genome editing programs (15, 16) or molecular suppression (24) to clarify the genetic and molecular mediators of structure formation. We believe that label-free quantitative phase imaging of whole living organisms is an important tool for exploring the interaction of genetics, proteomics, and biomechanics to achieve structural phenotypes with specific functions. An in-depth understanding of the formation principles of the multifunctional material structure used by insects may also enable us to collect specific manufacturing strategies by controlling the material structure of all functionally relevant length scales.

The V. cardui butterfly is bred in our laboratory with original stocks obtained from Carolina Biological Supply Company. Monitor the larvae in a separate container to make sure that the molting time of the larval skin is within 15 minutes. The average duration of pupal development of other pupae of the same generation was used to estimate the percentage of ontogeny after surgery. Other detailed information is contained in the SI appendix.

We used two surgical methods to expose the forewing and hindwing tissues in the pupae. In order to expose the forewing imaged at a single time point (Figure 2A), we removed part of the cuticle from above the forewing. The stratum corneum can be removed at any time after lysis, which occurs in about 20% of development (SI appendix, Figure S3). We anesthetize the pupa in a small room flushed with carbon dioxide for 5 to 10 minutes to immobilize the organism, while using a scalpel (feather; micro scalpel) to make a small (~1 × 1 mm) shallow incision in the pupa to remove the cuticle . Then, we placed a 150 µm thick (VWR; No. 1) glass cover glass on the exposed tissue and used a hand-held dental curing lamp (NSKI; LY-) to cure the dental composite (Pentron; Flow-It ALC) with light Seal it 02 LED light curing lamp).

In order to expose the hind wings for continuous imaging, we adjusted the surgical strategy described by Otaki and colleagues (35, 44). We grab the pupa epidermis of the recently molted pupae and forewings and fold it forward toward the head (these young pupae do not need to be fixed). In order to limit the spread of any potential melanizing immune response, we use dental composite strips to separate the hind wings from the forew wings. Again, we placed the glass coverslip on the exposed wing and sealed it with dental composite. Other detailed information is contained in the SI appendix.

We use the previously published interferometry techniques (37, 38) (SI appendix, Figure S1). In short, a supercontinuum laser (NKT Photonics; SuperK Extreme EXR-4) generates dynamically changing speckle illumination through a rotating diffuser. For the band-pass filter, choose a center wavelength of 800 nm and a bandwidth of 40 nm. Then use a polarization beam splitter to distribute the illumination to the objective lens on the reference arm and the sample arm (Olympus; LUMPLFLN 60XW, 1.0NA immersion in water); the quarter wave plate keeps the reference and sample signal polarizations orthogonal to each other. Off-axis holography uses a grating to divide the combined signal, and then a combination of a polarizer and a quarter-wave plate to filter and align the interference signal. The interferogram is then captured by a camera (Point Grey; Flea3) at up to 100 fps and processed in Fourier space to collect amplitude and phase information from the sample. The motorized stage in the sample stage scans the vertical position of the sample to obtain a volume image. The theoretical lateral resolution of 490 nm and the theoretical axial resolution of 1.03 µm have been previously confirmed by measurements (37). When imaging under the surface of the body, the resolution is expected to decrease due to multiple scattering from the tissue and spherical aberration, and if the reference arm drifts, it may gradually decrease over time.

We usually start imaging the pupae within 1 to 3 hours after surgery. We placed the pupae under the reference arm of the speckle-correlated reflection phase microscope and scanned along the optical axis (z-axis) in 400 nm steps to create a large number of image slices. The appropriate imaging depth is determined by algorithm by locating the depth of the maximum area integral intensity. Adjust the reference arm to match the change in optical path length. A cover glass with a reflective gold coating on the far side is used as a reference mirror to eliminate the dispersion and path length delay introduced by the pupa cover glass.

For continuous imaging, the program automatically circulates the focusing and image acquisition process every 15 minutes, and personally checks the alignment of the microscope once a day; the syringe pump regularly replenishes the water lost due to the evaporation of the immersion lens. For a single point in time image (for example, Figure 2G-J), carbon dioxide flows through the pupa for 5 minutes to ensure immediate fixation before imaging. Use MATLAB (MathWorks) to perform calculations and interface control. Other detailed information is contained in the SI appendix.

We found that using the theoretically estimated point spread function (45) for 10 iterations of Lucy-Richardson deconvolution is effective for visualizing scale features in amplitude data, especially in mid-to-late scale structures (Figure 1 and Figure 2) ).

The color representation of the 3D amplitude data is based on the Temporal-Color Code plug-in provided by Kota Miura for Fiji/ImageJ. The image slice is colored according to its depth position, and then the average intensity is obtained along the volume depth of each RGB channel.

The initial phase data is wrapped on 2π, which can be described as (n) as an integer to ensure packaging; that is, -π<Wφ(n)≤π. Itoh (46) describes the unwrapped phase data of one-dimensional wrapped phase data at n=0,1,…,N points −π<φwrapped(n)≤π: φunwrapped(m)=φwrapped(0)+∑n=1m ( WΔϕwrapped(n)), [3] where Δ is the differential operator, Δϕ(n)=ϕ(n)−ϕ(n−1). We use this method to analyze the one-dimensional contours in the phase data.

In 2D, unfolding may be affected by noise and cliffs, and these effects are more complicated when considering two dimensions (47). We use the integrand function WΔφwrapped(n) of the expansion scheme, [4] to represent the local phase behavior, which represents the local phase gradient along a specific dimension. The combination of the horizontal and vertical components of the gradient allows us to determine the direction of inclination of the surface. This visualization strategy allows fine features to be identified in greater depth changes. Other detailed information is contained in the SI appendix.

We use Dragonfly (Object Research System) to measure the length, width, and thickness of the overlay in the deconvoluted 3D amplitude data (SI appendix, Figure S9). For scales that are partly outside the imaging window, the length is extrapolated (SI appendix). Use the measured length, width, and thickness to estimate the surface area and volume to define an assumed bounding box of proportions.

To measure the ridge height and spacing in the phase data (SI appendix, Figure S10), we unwrap and rotate the one-dimensional profile from the phase data slice. We found the main spatial frequency through Fourier analysis. We define the height of a peak as the distance from the peak to the line connecting the troughs on both sides of the peak. Calculate the average peak height and SD peak height for each profile; then use all configuration files to determine the combined average and combined SD for each time point.

To measure the actin/ridge spacing in the amplitude data (SI appendix, Figure S11), we first rotate the data to capture the scale plane and define the mesoscale region of interest in the dragonfly. Then we find the main spatial frequency through two-dimensional Fourier analysis. Other detailed information is contained in the SI appendix.

The butterfly scale data and data analysis code are publicly available on Zenodo (https://doi.org/10.5281/zenodo.5532941) (48).

Thanks to Youngwoon Choi (Korea University) for insights on speckle-related reflection phase microscopy, Yong Zhang (Massachusetts Institute of Technology [MIT]) for support of gold-plated coverslips, Bodo Wilts (University of Fribourg) for comments on the manuscript, Cindy Lu discusses cell differentiation, Meera Singh (Massachusetts Institute of Technology) and Julia Kudryashev (Massachusetts Institute of Technology) assist in breeding butterflies and early surgical trials. This work was approved by the National Science Foundation through the "Designing Materials to Innovate and Design Our Future" program (DMREF-1922321) and the CBET program (Grant 1804241) on "Particles and Multiphase Processes" (to ADM and MK) support. SK, ZY and PTCS recognized the support of NIH grants P41EB015871, R21GM140613, R01HL158102 and R01DA045549. PTCS further thanks the support of DE-FOA-0002359 Grant U01CA202177 of the Ministry of Energy; Hamamatsu Corporation; Samsung Advanced Institute of Technology; Singapore-MIT Research and Technology Center Alliance-Manufacturing Personalized Medicine Interdisciplinary Research Group (SMART CAMP IRG) Key analysis; Hardy Holdings Co., Ltd.; Fujikura Co., Ltd.; and Hong Kong Innovation and Technology Alliance.

Author contributions: ADM, SK, ZY, PTCS and MK design research; ADM and SK conducted research; ADM, SK, ZY, PTCS and MK contributed new reagents/analysis tools; ADM, SK and MK analysis data; ADM and MK wrote the paper; ADM, SK, ZY, PTCS, and MK discussed and reviewed the manuscript.

The author declares no competing interests.

This article is directly contributed by PNAS.

View relevant content online, such as comments.

This article contains online support information at https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.2112009118/-/DCSupplemental.

This open access article is distributed under the Creative Commons Attribution-Non-Commercial-No Derivative License 4.0 (CC BY-NC-ND).

Thank you for your interest in advertising on PNAS.

Note: We only ask you to provide your email address so that the people you recommend the page to know that you want them to see it, and that it is not spam. We do not capture any email addresses.

Feedback privacy/legal

Copyright © 2021 National Academy of Sciences. Online ISSN 1091-6490. PNAS is a partner of CHORUS, COPE, CrossRef, ORCID and Research4Life.