Samples of the two cicada species for this study, namely the lesser bronze cicada Kikihia scutellaris (Fig. 1a) and the clapping cicada Amphipsalta cingulata (Fig. 1b), were collected in a dead state in New Zealand (Baracchi and Baciadonna 2020).

Both cicada species for this study are endemic to New Zealand and were collected in a dead state. ©  2021, Alexander M. Bürger. Scale bar is 1 cm. a Kikihia scutellaris, the lesser bronze cicada. b Amphipsalta cingulata, the clapping cicada

Both New Zealand cicadas selected for the study have nanopillars. We focused the replication on K. scutellaris due to higher nanopillars (“Wing topology analysis section) and abundant availability. Employing casting and curing of PVS at 4\(^\circ \text \) enabled the creation of cicada wing surface replicas with nanoscale precision (Bürger 2022; van Nieuwenhoven 2022; Koch et al. 2008; Zobl 2018).

The wings were carefully removed from the thorax and used without any further cleaning steps. A window-like rigid vein structure supports the membrane of the cicada wings. The more prominent veins (due to their thickness and aspect ratio) reduce the nanopillar-covered area and the field of view, especially during microscopy. A biopsy punch (Kai Medical Biopsy Punch 5 mm, Kai Industries Co. Ltd., Seki, Japan) was used to cut comparable 5 mm discs out of the wings. The more prominent veins were excluded during sectioning.

An atomic force microscope (AFM) was used in tapping mode to minimize destructive interactions of the AFM tip with the sample.

Most AFM investigations were performed with an MFP-3D-BIO AFM (Asylum Research, Oxford Instruments plc) because the more accessible chamber enables a faster switching time between different samples. However, some investigations used the Cypher ES (Asylum Research, Oxford Instruments plc, Abingdon, United Kingdom) Atomic Force Microscope (AFM). The tips used in all scans were BudgetSensors® Tap300-G (resonance frequency 300 kHz; force constant 40 N/m, the cantilever was made of uncoated monolithic Silicon).

The Scanning Electron Microscopy (SEM) micrographs used in this research were taken with a Scios 2 DualBeam and an FEI Quanta 250 FEG (from Thermo Fisher Scientific GmbH, Waltham, Massachusetts, U.S.) at the University Service Centre for Transmission Electron Microscopy, Vienna University of Technology, Austria. The cicada wings were washed in distilled water ultrasonic bath for 10 min and afterward put under a stream of nitrogen to remove the water residuals. In addition, the wings of both species with attached E. coli bacteria were studied by air-drying the wings after immersing them in a bacterial suspension OD\(^=0.7\) for one hour. All samples were mounted using double-sided adhesive carbon tape (Thermo Fisher Scientific GmbH, Waltham, Massachusetts, U.S.) and sputtered with a 4 nm gold/palladium layer.

Testing protocols for analyzing the efficiency of antibacterial nanostructured surfaces are not readily available, and previous studies used diverse techniques and protocols to show the antibacterial properties (Ivanova et al. 2012, 2013a, b, 2017; Hasan et al. 2012; Kelleher et al. 2015; Román-Kustas et al. 2020; Pogodin et al. 2013; Tripathy et al. 2017; Bandara et al. 2017). The testing protocol of this study was designed based on the experiences of these previous studies.

For the antibacterial tests, as representative of Gram-negative bacteria Escherichia coli (E. coli, DSM 5698/ATCC 25404) and as representative of Gram-positive bacteria Staphylococcus aureus (S. aureus, DSM 1104/ATCC 25923) were used, both obtained from the Leibnitz Institute (DSMZ-German Collection of Microorganisms and Cell Culture GmbH, Braunschweig, Germany).

The initial cultivation was performed in Lysogeny broth (LB-medium), which was produced after the standard protocol developed by Giuseppe Bertani  Bertani (1951). Stock bacteria solutions were mixed in a 1:1 ratio with a 50% glycerol solution in phosphate-buffered saline (PBS, Life Technologies Limited, United Kingdom). The suspension was mixed at 800 rpm in a thermal shaker (Thermomixer Compact 5350, Eppendorf SE, Hamburg, Germany) before distributing 1 ml per Eppendorf tube® (Eppendorf SE, Hamburg, Germany) and stored at − 50 \(^\circ \text \).

Before each experiment, the bacteria were gradually thawed and suspended in LB medium and mixed at 800 rpm in the thermal shaker at 37 \(^\circ \text \).

The bacteria concentration was determined by measurement of the optical density (OD\(^\)) at wavelength 600 nm with a BioTek EL800 plate reader (The Lab World Group, Hudson, MA, U.S.). The bacteria were grown until they reached an optical density of OD\(^\) of 0.7. An OD\(^\) of 0.7 is commonly used in studies on antibacterial properties  (Sambrook and Russell 2001) as the bacterial growth is at its optimum at the end of the exponential growth phase (Hall et al. 2013; Buchanan 1918). Former studies on the antibacterial properties of cicada wings used OD\(^\) in the range of 0.1 and 0.3 (Hasan et al. 2012; Kelleher et al. 2015; Román-Kustas et al. 2020). This study stayed with the OD\(^\) value of 0.7 from the laboratory manual (Sambrook and Russell 2001).

Live/dead staining was performed with propidium iodide (PI) (Sigma-Aldrich, Merck SA, Germany) and bisbenzimide Hoechst 33343 (Hoechst) (Sigma-Aldrich, Merck SA, Germany) fluorescence stains. PI is a nucleic acid stain that does not permeate intact membranes and stains only dead cells with damaged membranes. In comparison, Hoechst is a DNA stain that can pass intact cell membranes. The PI was dissolved in distilled water to a stock concentration of 1 mg/ml. During experiments, the concentration of the stock solution was further reduced with PBS to 40 \(\upmu \)l/ml (Ciancio et al. 1988; Scientific 2021). Hoechst was dissolved in distilled water to a stock concentration of 0.5 mg/ml. Methanol-based fixation(Mangels et al. 1984) was done by adding 25% methanol to the Hoechst staining solution. Experimental concentration was reached by dissolving the stock solution in a PBS/methanol mixture (75/25) to 6 \(\upmu \)l/ml (Ciancio et al. 1988; Scientific 2021).

Five-millimeter diameter samples (cicada wing, 3D-printed disks) were glued into 3D-printed bottomless wells with nail polish onto a microscope slide. All 3D-printed parts were printed using photopolymer resin (Elegoo ABS-Like LCD UV photopolymer rapid resin, GE-EL-3D-005, Elegoo, Shenzhen, China) with an Elegoo Mars and a Phrozen Sonic Mini 8K. The sample was incubated in the well for one hour with 350 \(\upmu \)l bacterial suspension at room temperature. Blank poly-L-lysine coated microscope slides (Electron Microscopy Sciences, Hatfield, Pennsylvania, U.S.) were used as a control group for the cicada wings. The empty well was directly glued to the slide for these coated slides. To exclude the potentially toxic effects of the used resin residues on the viability of bacteria, a 3D-printed 5 mm diameter disk with 100x40x40 \(\upmu \)m\(^3\) pillars was used as a control specimen. Such large pillars do not have any mechanical effect on the bacteria.

In the first staining step, 110 \(\upmu \)l of the PI/PBS solution was pipetted in each well after gradually extracting the bacteria suspension and incubating for 15 min. For the dead control group, 25% of the PBS from this step was replaced by methanol. The samples were protected from light during the whole staining procedure. Afterwards, the samples were gently washed with 110 \(\upmu \) PBS to remove unbound PI stain and unattached bacteria. The samples were exposed to 110 \(\upmu \)l Hoechst/PBS/methanol for 30 min, again shielded from light. Then the suspension and the surrounding well were removed from the microscope slide, leaving the sample glued in the center. Fluoromount-G™(00-4958-02, Thermo Fisher Scientific GmbH, Waltham, Massachusetts, U.S.) was applied to the sample with a cover slip to enhance the stability of the fluorescence stains (PI/Hoechst). The margin of the cover slip was sealed with nail polish to prevent the evaporation of the Fluoromount-G™.

The fluorescence imaging of the wings was performed with an IX73 inverted microscope (Olympus Corporation, Tokyo, Japan) at the Department of Dermatology and a Zeiss LSM 880 Airyscan (Carl Zeiss AG, Oberkochen, Germany) at the Core Facility Imaging, both at the Medical University of Vienna. PI was excited at 561 nm, and the emission was measured at 640 nm, whereas Hoechst was excited at 405 nm and measured at 488 nm.

The fluorescence image color channels were analyzed separately with the software Fiji Schindelin et al. (2012). In the bright field channel of the image, clusters could be separated into single bacteria.

At the end of the 60-min incubation cycle and just before the staining started, 20 \(\upmu \)l of the bacteria-containing fluid was slowly (with as little disturbance of the solution as possible) extracted from the bottom of the container with a pipette. The concentration of the droplet collected from the bottom also represents the concentration of the 2 \(\upmu \)m above the nanopillars. The 20 \(\upmu \)l droplet was transferred to a \(15\times 15\) mm\(^\) cover slip.

Subsequently, a second cover slip was positioned on top of the droplet. The droplet completely filled the area between the two cover slips without spilling. Bright-field images of the plain between the cover slips were obtained with a Zeiss LSM 880 Airyscan (Carl Zeiss AG, Oberkochen, Germany) and were used for counting the bacteria with Fiji Schindelin et al. (2012). The counting was done automatically using the Fiji built-in function “find maxim” after twice applying the Fiji-filter-maximum with a radius of 5 pixels and the Fiji FFT bandpass filter (between 40 and 3 pixels) to remove the background shading.

The hydrophobic properties of the cicada wing were investigated with a 3 \(\upmu \)l deionized waterdrop pipetted on a veinless spot on the wing sample. The open-source software OpenDrop Huang et al. (2021) was used to analyze the contact angle in the images taken with a Thorlabs CMOS Camera (Thorlabs, Newton, New Jersey, U.S.) with a telecentric lens (0.5x, 65 mm WD CompactTL TM Telecentric Lense, Edmund Optics Inc., Barington, New Jersey, U.S.).