Resolving Bio-Nano Interactions of E. coli Bacteria-Dragonfly Wing Interface with Helium Ion and 3D-Structured Illumination Microscopy to Understand Bacterial Death on Nanotopography

Obtaining a comprehensive understanding of the bactericidal mechanisms of natural nanotextured surfaces is crucial for the development of fabricated nanotextured surfaces with efficient bactericidal activity. However, the scale, nature, and speed of bacteria-nanotextured surface interactions make the characterization of the interaction a challenging task. There are currently several different opinions regarding the possible mechanisms by which bacterial membrane damage occurs upon interacting with nanotextured surfaces. Advanced imaging methods could clarify this by enabling visualization of the interaction. Charged particle microscopes can achieve the required nanoscale resolution but are limited to dry samples. In contrast, light-based methods enable the characterization of living (hydrated) samples but are limited by the resolution achievable. Here we utilized both helium ion microscopy (HIM) and 3D structured illumination microscopy (3D-SIM) techniques to understand the interaction of Gram-negative bacterial membranes with nanopillars such as those found on dragonfly wings. Helium ion microscopy enables cutting and imaging at nanoscale resolution while 3D-SIM is a super-resolution optical microscopy technique that allows visualization of live, unfixed bacteria at ~100 nm resolution. Upon bacteria-nanopillar interaction, the energy stored due to the bending of natural nanopillars was estimated and compared with fabricated vertically aligned carbon nanotubes. With the same deflection, shorter dragonfly wing nanopillars store slightly higher energy compared to carbon nanotubes. This indicates that fabricated surfaces may achieve similar bactericidal efficiency as dragonfly wings. This study reports in situ characterization of bacteria-nanopillar interactions in real-time close to its natural state. These microscopic approaches will help further understanding of bacterial membrane interactions with nanotextured surfaces and the bactericidal mechanisms of nanotopographies so that more efficient bactericidal nanotextured surfaces can be designed, fabricated, and their bacteria-nanotopography interactions can be assessed in situ .


INTRODUCTION
The ability of bacteria to adhere, survive and subsequently form biofilms on abiotic surfaces is the leading cause for infection of prostheses after surgery, resulting in implant failure and revision surgery. [1][2] In recent years, there have been significant efforts to reduce infection by developing various nanotextured surfaces on implantable medical devices. [3][4] These nanotextured surfaces are often inspired by nature, and the fabricated topographies are expected to reduce initial bacterial adhesion or kill any bacteria attempting to attach to the surface. The investigation of bacteria-nanotopography interactions has, therefore, become a significant research interest for the efficient and effective development of advanced fabricated bactericidal nanotextured surfaces (NTS) for biomaterials application. [5][6][7][8] This is a growing field of research as nanotextured surfaces do not leach bactericidal chemicals, and therefore, the bactericidal property is conserved for longer periods compared to chemical approaches. [9][10] Furthermore, nanotextured surfaces are non-toxic and more effective at controlling bacterial strains that produce extracellular polymeric substance (EPS) secretions. [11][12] The mechanical bactericidal efficacy of nanotextured surfaces is dependent on the nanotopography, architecture of the nanostructures, surface chemistry, and the type of bacteria. 5,8,[13][14][15][16] However, the antibacterial mechanisms and effectiveness of surface parameters are not well understood due to the limited understanding of the bacteriananotopography interactions. Recently, various views and explanations have been proposed for understanding the mechanism of membrane damage. 5-8, 13, 17-21 Cicada wing-like nanostructures are mainly bactericidal against Gram-negative bacteria. 7,[22][23] The initial report of bacterial membrane damage upon interaction with cicada wing nanostructures used a differential DNA staining method with confocal microscopy to predict the membrane integrity. 6,24 This approach is an indirect measurement of the membrane condition, where two separate fluorescent markers are used to stain the nuclear material. As this approach only stains the DNA, it cannot be used to visualize the bacterial membrane directly. Furthermore, due to the limitations in spatial resolution, confocal microscopy can only resolve the color variation of the bacterium and cannot provide detailed information on its membrane integrity. Therefore, to obtain a comprehensive understanding of the effects on the bacterial membrane during the interaction with nanostructures, it would be better to directly visualize the bacterial membrane using super-resolution microscopy techniques such as 3D structured illumination microscopy (3D-SIM) which utilizes spatially patterned fluorescence excitation beams to achieve a resolution of ~100 nm. [25][26] Recently, the interfaces where bacteria and natural nanotopographies interact have been observed at nanoscale resolution through the use of some of the most advanced microscopic techniques available to-date including helium ion microscopy (HIM), transmission electron microscopy (TEM), focused ion beam (FIB) techniques and faster atomic force microscopy (AFM). 13-14, 17, 27-33 These approaches have furthered our understanding of bacteriananotopography interactions of natural surfaces. High-resolution TEM images of crosssectioned interfaces revealed that the natural nanopillars under the bacteria are bent during the interaction and that dead bacteria appeared to have produced EPS. 13 A recent study on the interactions of bacteria with Vertically Aligned Carbon Nano Tubes (VACNT) reported the bending of these nanostructures during bacteria-VACNT interactions. 19 This is in line with the previously reported bactericidal effects of natural nanopillar topography of dragonfly wing with E. coli. 13 In this study of bacteria-VACNT interactions, the energy stored during bending of multiple lengths of VACNT were quantified, and it was determined that shorter VACNT had more efficient bactericidal activity. 19 The bactericidal mechanisms of VACNT and black silicon (bSi) have been described in terms of a purely physical mechano-bactericidal basis, and significant differences in the two different surfaces were observed. In bSi, no significant difference in bactericidal activity was reported with changing nanopillar height or its surface chemistry. bSi nanostructure was identified as rigid structures that do not bend when interacting with bacteria. 24 A separate study suggested that adhesion is important but that the membrane is not necessarily pierced by the bSi nanostructures to achieve the bactericidal activity. 14 For VACNT, it was reported that the shorter nanopillars are more efficient and that VACNT bend during interaction with bacteria and in some cases, the bottom surface of bacteria is wrapped by VACNT. 19 Furthermore, when the surface chemistry of VACNT was changed using O2 or CF4 plasma, significant changes to bacterial adhesion on the surface and bactericidal activity of Pseudomonas aeruginosa were identified.
One plausible mechanism to explain bacterial membrane damage upon interaction with nanopillars is that the bacterial cell first attaches onto the nanopillars, and then while attempting to move parallel to the surface, the nanopillars are bent, and membrane damage occurs through shearing. 13 Another study has suggested that attempted bacterial cell division on the titanium nanotopography causes bacterial membrane damage. 34 When the membrane damage occurs, characteristic EPS production, bending of nanopillars, and leaking of cytoplasm are observed.
These post-membrane damage characteristics have recently been observed in HIM, TEM, SEM and confocal studies. 13-14, 17, 19, 28-32, 35 However, as these events occur rapidly, and only at the cell surface, the images obtained using these techniques commonly show the bacteria flattened on the nanopillars, apparently due to membrane damage. It should be noted that the interaction between bacterial membranes and nanostructures cannot be resolved with these microscopic tools due to the limitation in the time-scale or spatial resolution. For instance, extensive sample preparation steps are required prior to imaging the bacteria-nanotopography interaction by TEM, and for FIB, metal deposition is required to reduce charge accumulation for successful cross-sectioning. 13,35 As these steps are time-consuming, they prevent capturing the initial stages of the bacterial membrane damage process.
The different observations and interpretations in reports of the mechanobactericidal action of different nanotextured surfaces provide the rationality for further investigations. It remains of urgent and significant importance to evaluate the mechanism of bactericidal activity and bio-nanotopography interactions for both fabricated and natural nanotopographies towards gaining a deeper understanding of the bactericidal mechanism for the efficient design of bactericidal biomaterials.
A combination of approaches could be used to overcome the technical limitations encountered to-date in order to understand the membrane damage process more clearly. One approach would be to freeze the initial stages of bacteria-nanosurface interactions and utilize an imaging technique such as HIM, which has nanoscale resolving power with a larger depth of focus and microscale field of view to image the entire interaction. A second, complementary approach could be to utilize a super-resolution optical imaging technique such as 3D-SIM to image bacteria-nanotopography interactions in situ under physiological conditions in real-time.
The aim of this work, therefore, was to explore the possibilities of utilizing both HIM and 3D-SIM to provide further insights into E. coli bacteria-nanotopography interactions in real-time and thereby further our understanding of the mechanobactericidal activity of the natural nanotopography of dragonfly wings to Gram-negative bacteria. This work supplements previous studies carried out with Gram-negative Escherichia coli and dragonfly wings using HIM, AFM, FIB, and TEM. 13,35 In previous studies, we have used TEM and FIB to expose the interface. Here, we use the Ne + beam to cross-section and He + to image exposed interface using HIM; thereby, we avoid excessive sample preparation steps associated with TEM and FIB.
Special attention is given to the characteristics of the bacteria-nanopillar interface.

Interfacial characteristics of dragonfly wing nanopillars-Escherichia coli bacterium by
Helium Ion Microscopy (HIM). We used helium ion microscopy (HIM), to examine the bacteria-nanotopography interface of E. coli on dragonfly wings. Thin planes of bacteria attached to dragonfly wings were precisely milled using a Ne + beam, and the exposed interface imaged using a He + beam. 36 This cross-sectioning was carried out without any conductive metal coating, as is required for focused ion beam (FIB) or scanning electron microscopy (SEM). 13,17 Hence, the bacteria-nanotopography interface images acquired by HIM are not obscured and remain close to their natural state, limiting possible artifacts that might occur during the metal coating processes involved in FIB/SEM. Compared to TEM, this method is flexible for performing cross-sections at various positions, lengths, and directions, whereas TEM is limited to tiny sections of lamellas, and positions of cross-sections cannot be controlled.   [43][44] Therefore, in a favorable chemical environment, interactions between these components may be possible, and a partial or even full encasing of the nanopillars with bacterial membranes could account for the adhesion between Gram-negative bacteria and the nanopillars. Though we have used a strain without flagella in this study, a one contains flagella would show much favorable interactions with superhydrophobic surfaces. 45 The circled area of Figure 1e highlights a nanopillar flooded by membrane deformations, EPS, or cytoplasm. The bacterial membrane appears to be extended and stretched towards the nanopillar to form the connection while the nanopillar is slightly tilted or bent to the left from its vertical position. This bending and stretching could be a result of strong adhesion of bacteria and its attempts to move away from the surface.
The above observations and analysis suggest that a series of complex interactions could take place between E. coli and natural nanopillar topography before the bacterial membrane is damaged. Therefore, the biological effects of such interactions should also be considered when developing reliable models of the bactericidal effects of antibacterial surfaces.
Further chemical composition studies are required to determine the composition of the fingerlike connections we observed between bacteria and nanopillars by HIM. Although we were unable to characterize the nature of the extracellular bacterial-derived substances in this study, at this stage, our approach contributes toward a greater understanding of the bacterial death on nanopillars and will reduce future experimental bias towards just one single bactericidal mechanism.  Figure 2b shows the crosssection of the bacteria-wing interface, where unevenness of the wing is evident. The wing is indicated by a green arrow, and a membrane damaged bacterium by a white arrow.

Interfacial characteristics of dragonfly wing nanopillars-Escherichia
Interestingly, the bacterial cell membrane appears to be missing at the side of the bacterial cell that is in contact with the dragonfly wing (Figure 2a, c). The section views (Figure 2c) clearly show that the bacterial membrane is discontinuous when in contact with the wing surface, thereby allowing the DNA stain EthD-III to enter the cell. Interestingly, we observed that cells that had taken up the dead cell stain (EthD-III) presented different red fluorescence intensities (white arrows, Figure 2a), which suggests that membrane damage may have occurred at different times and/or that different levels of membrane damage have occurred. 3D-SIM utilizes spatially patterned excitation beams to achieve the resolution below the diffraction limit, thereby surpassing the classical spatial resolution limit of conventional microscopy. The 3D-SIM platform used in this study provides a lateral and axial resolution of 110 nm and 280 nm, respectively. 25 By using stains specific for membranes and DNA of membranecompromised cells, we were able to explore membrane damage in live bacterial cells interacting with dragonfly wings while maintaining physiological conditions. This is the first report of such live in-situ imaging of E. coli interacting with dragonfly wings with higher contrast between the membrane and nuclear material.
Elastic energy estimate of bent nanopillars. While our microscopic approach confirms the bending of nanopillars, bacterial membrane stretching, deformation, and damage, microscopy alone do not provide the mechanism that leads to these consequences. The lack of a comprehensive understanding of the bactericidal mechanism ultimately limits the translational potential towards real-world applications of nanotopography. Achieving such knowledge will ultimately lead to the efficient and effective development of real-world applications. A number of recent publications have indicated that the arrangement of nanopillars is not sufficient to achieve the bactericidal activity but that external factors, including the movement of bacteria and bending of nanopillars, also contribute to the bactericidal activity of dragonfly nanopillars. 21 To assess the contribution of bending of dragonfly nanopillars on the underlying mechanism of bactericidal activity, we estimated the amounts of the stored elastic energies of natural nanotopography of dragonfly wings during interactions with bacteria. This data could then be used to design artificial bactericidal surfaces and compare them with fabricated nanopillars. As we have identified that bending of dragonfly wing nanopillars occurs when the bacterial membrane is being damaged, 13 we assessed the storage of energy in the bending of dragonfly nanopillars. According to the model suggested by Linklater et al. 19 for a surface with VACNT, using Bernoulli's beam theory, a force, P (load) acting on the tip of a nanopillar parallel to the substratum when a bacterium attempts to move, predicts a deflection of the tip, δ as; where L is the length of the nanopillar, E is the material's Young modulus, and I the area moment. Young's modulus (E) and inversely related to L 3 . That means, for a given deflection, longer nanopillars stores smaller energy.
According to these storage energy calculations, both dragonfly wing nanopillars and 1 µm CNT store similar energies upon the same tip deflection (Figure 3). With the same tip displacement, short nanopillars of the dragonfly wing exert stronger forces than tall nanopillars, while 1 µm VACNT energy storage values are in between short and tall nanopillars. Consequently, the natural nanotopography of dragonfly wings may induce more damage to the bacterial cell membrane than CNT. We observed with HIM that the interacting nanopillars with bacteria are bent, maybe when bacterium attempts to move. The stored energy due to bending of the nanopillars may release, leading to further stretching of the bacterial membrane causing physical damage to the membrane. This study has utilized sophisticated imaging techniques with minimal sample processing that has important implications for the comprehensive understanding of the biointerface. By using a Ne + beam and electron flood gun, we eliminated the use of a conductive metal coating that is used in SEM and avoids the application of metal layer onto the sample prior to cross-sectioning, as reported previously in FIB/SEM and TEM studies. 13,17,19,27 Elimination of both metal coating and Pt deposition allowed us to image the bacteria-nanopillar interaction close to its natural state. Ne + beam cross-sectioning has allowed us to cross-section small amounts of the sample at specific positions, where TEM cross-sections are random. This has allowed us to observe the interfacial features, at precise locations with more clarity and without disturbance at nanoscale resolution. In E. coli, it was found that bacteria and nanopillars make contacts at specific anchor locations. Finger-like protrusions, possible secreted EPS, cytoplasmic material, and membrane deformations were also visible; however, they were not able to be distinguished using ion beam microscopy. Further, it was confirmed that the membrane is not pierced by the nanopillars, but is more likely stretched and deformed during the interaction.
With live 3D-SIM, bacterial membranes and the nuclear material of membrane-compromised cells were resolved. However, the nanopillars or finger-like protrusions that we observed with HIM could not be visualized with 3D-SIM due to the resolution limit of this technique. We were, however, able to observe that membrane damage occurred at the surface of the bacterial cell that was in contact with the dragonfly wing. Although neither of the imaging methods alone could fully resolve the interactions between the nanopillar and the bacteria, this study provides useful insight into the limits and possibilities of the imaging methods. It is evident from this study that multiple experimental approaches that involve minimal sample processing are necessary to comprehensively assess the bactericidal activity of nanotextured surfaces.
Further mechanical insights were obtained by estimating the bending energy storage of short and tall nanopillars of the dragonfly wings. These elastic energy storage estimations indicated that short nanopillars are slightly higher in energy storage compared to tall nanopillars of the dragonfly wing. It was also found that 1 µm VACNT energy storage is between that of the tall and short nanopillars of dragonfly wings. This indicates that both natural bactericidal nanotopography and VACNT are likely to mechanically behave quite similarly in achieving bactericidal activity. This suggests that soft materials can achieve bactericidal effects and that antibacterial surfaces do not necessarily need to be designed using hard solid materials.
The microscopy techniques that we utilized in this study distinguished the features of the interface between a Gram-negative bacterium and the dragonfly wing. Given that Gramnegative and Gram-positive bacteria have significant differences in their cell wall and membrane structures, in future studies, bacterial surface/nanotopography interfaces should be examined with techniques capable of nanoscale and molecular resolution with minimal processing. Our approach could be applied to examine interactions of Gram-positive and Gramnegative bacteria and eukaryotic cells with various nanotopographies to guide the design of efficient nanofabrication.

EXPERIMENTAL SECTION
Dragonfly wing collection and sample preparation. The dragonfly specimens were collected in Brisbane, Australia and wings were aseptically removed from the body, washed with flowing deionized water, and stored in the dark at 4°C in sterile containers.
Bacterial growth conditions and sample preparation for HIM. Dehydrated and dried wings were mounted using double-sided carbon tape onto an aluminum stab for imaging. Zeiss Orion NanoFab equipped with Ne + and He + beam was used for cross- Thermo Fisher Scientific) was placed on a microscope slide to create a chamber, the wing was attached inside the chamber using a UV glue (Dymax), and PBS was used as mounting media for imaging. A #1.5, 22x22 mm glass coverslip was ethanol sterilized and placed on top.
3D images were acquired with a z-step size of 0.125 µm for 3 µm total thickness. Raw images were processed with SoftWoRx (GE Healthcare) for image reconstruction and channel alignment and then prepared with IMARIS (v 9.1.2, Bitplane).