Nanoplasmonic Sensing Architectures for Decoding Membrane

6 hours ago - Received 2 March 2018. Published online 28 May 2018. +. Altmetric Logo Icon More Article Metrics. CURRENT ISSUELATEST NEWS...
2 downloads 0 Views 1MB Size
Subscriber access provided by UNIVERSITY OF TOLEDO LIBRARIES

Article

Nanoplasmonic Sensing Architectures for Decoding Membrane Curvature-Dependent Biomacromolecular Interactions Abdul Rahim Ferhan, Joshua A. Jackman, Bita Malekian, Kunli Xiong, Gustav Emilsson, Soohyun Park, Andreas B. Dahlin, and Nam-Joon Cho Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.analchem.8b00974 • Publication Date (Web): 28 May 2018 Downloaded from http://pubs.acs.org on May 28, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 16 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

Nanoplasmonic Sensing Architectures for Decoding Membrane CurvatureDependent Biomacromolecular Interactions Abdul Rahim Ferhan†, Joshua A. Jackman†, Bita Malekian§, Kunli Xiong§, Gustav Emilsson§, Soohyun Park†, Andreas B. Dahlin§, Nam-Joon Cho*,†,‡ †

School of Materials Science and Engineering and Centre for Biomimetic Sensor Science, Nanyang Technological University, 50 Nanyang Drive 637553, Singapore ‡ School of Chemical and Biomedical Engineering, Nanyang Technological University, 62 Nanyang Drive 637459, Singapore § Department of Chemistry and Chemical Engineering, Chalmers University of Technology, 41296 Göteborg, Sweden *Corresponding author E-mail: [email protected] Phone: +65-6592-7945 Abstract Nanoplasmonic sensors have emerged as a promising measurement approach to track biomacromolecular interactions involving lipid membrane interfaces. By taking advantage of nanoscale fabrication capabilities, it is possible to design sensing platforms with various architectural configurations. Such capabilities open the door to fabricating lipid membranecoated nanoplasmonic sensors with varying degrees of membrane curvature in order to understand how biomacromolecular interaction processes are influenced by membrane curvature. Herein, we employed an indirect nanoplasmonic sensing approach to characterize the fabrication of supported lipid bilayers (SLBs) on silica-coated nanowell and nanodisk sensing platforms, and to investigate how membrane curvature influences membrane-peptide interactions by evaluating the corresponding measurement responses from different spectral signatures that are sensitive to specific regions of the sensor geometries. SLBs were prepared by the vesicle fusion method, as monitored in real-time by nanoplasmonic sensing measurements and further characterized by fluorescence recovery after photobleaching (FRAP) experiments. By resolving different spectral signatures in the nanoplasmonic sensing measurements, it was determined that peptide binding induces membrane disruption at positively curved membrane regions, while peptide binding without subsequent disruption was observed at planar and negatively curved regions. These findings are consistent with the peptide’s known preference to selectively form pores in positively curved membranes, providing validation to the nanoplasmonic sensing approach and highlighting how the integration of nanoplasmonic sensors with different nanoscale architectures can be utilized to study the influence of membrane curvature on biomacromolecular interaction processes. Keywords: nanoplasmonic sensing, localized surface plasmon resonance, label-free biosensor, near field decay, membrane-peptide interactions

1 ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Introduction Nanoplasmonic sensing represents an emerging surface-sensitive measurement technique for studying biomacromolecular interactions that occur at lipid membrane interfaces.1-5 Structural transformations involving adhered lipid vesicles at solid-liquid interfaces as well as biomolecule- and nanoparticle-induced interaction processes have been characterized in detail using nanoplasmonic sensor arrays comprised of noble metal nanostructures on glass substrates.6-16 These measurement capabilities have been aided by the growing maturity of nanofabrication technologies to produce nanostructured arrays, leading to robust sensing performance with a high level of reproducibility.17-20 In typical cases, the interaction of nonperiodic, well-separated metallic nanostructures with incident light results in the excitation of localized surface plasmons that leads to maximum light extinction at a resonant wavelength that is known as the extinction peak (max). Changes in the local dielectric environment in the near vicinity of the nanostructures lead to slight alterations in the corresponding extinction spectra, including a shift in the peak wavelength (max). Taking into account the surface sensitivity of the measurement platform and experimental conditions, interpreting such data can provide a wealth of information about interaction processes occurring across the entire sensor surface. As fabrication advances continue, it has become possible to tailor the architecture and surface chemistry of nanoplasmonic sensing platforms.18,21-28 This tailoring enables greater control over the fabrication of conformal supported lipid membranes on sensor surfaces and ultimately expands the capability of nanoplasmonic sensors to explore different aspects of membranerelated interaction processes. To this end, we and others have successfully employed oxidecoated gold nanodisk arrays with different surface sensitivities to quantitatively characterize vesicle adsorption12,16,29 and surface-induced shape deformation12,30 as well as supported lipid bilayer (SLB) formation.10,11,29 One area of particular interest has been the fabrication of nanoplasmonic arrays comprised of more complex nanoscale architectures such as nanoholes3133 and nanopores34-36 that give rise to optical extinction spectra with multiple spectral features that not only have distinct electromagnetic field enhancements but are also sensitive to different geometrical regions of the sensor surface.37 Altogether, the capability to fabricate conformal supported lipid membranes across nanostructured arrays with different geometries together with positional sensing opens the door to studying how membrane curvature affects biomacromolecular interaction processes. Indeed, membrane curvature is one of the most important factors influencing biomacromolecular interaction processes while also proving one of the most difficult to study.38-47 It also provides the basis for biotechnology and medical applications such as utilizing an amphipathic, α-helical (AH) peptide to selectively form pores in small lipid vesicles (~150 nm diameter or smaller) with high positive membrane curvature.48 Above a certain density of AH peptide-induced pores in a lipid vesicle, membrane lysis occurs and these capabilities can be translated into the development of antiviral strategies that work against enveloped viruses.49 Such possibilities highlight the importance of developing quantitative measurement strategies to probe membrane curvature effects on biomacromolecular interaction processes. To date, the influence of membrane curvature on AH peptide-induced membrane disruption has been mainly investigated using intact vesicles.47-52 While this measurement approach has shed light on membrane curvature-dependent processes, the current approaches are limited to studying positive membrane curvature with intact vesicles and it is challenging to tightly control the vesicle size distribution, and hence the degree of membrane curvature. To overcome these challenges, the development of surface-templated supported lipid membranes, 2 ACS Paragon Plus Environment

Page 2 of 16

Page 3 of 16 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

whereby membrane curvature can be precisely controlled by leveraging the favorable energetics of lipid-substrate interactions to coat nanostructured sensor arrays with conformal SLBs, represents a promising goal. To this end, it was recently demonstrated that nanoplasmonic sensing measurements can detect preferential analyte binding to regions of negative membrane curvature in SLB-coated nanowell configurations.37 Extending such approaches to decipher how biomacromolecular interaction processes are influenced by the particular type of membrane curvature (positive or negative) would greatly advance measurement capabilities. Herein, we employed an indirect nanoplasmonic sensing approach to characterize the fabrication of supported lipid bilayers (SLBs) on silica-coated nanowell and nanodisk sensing platforms, and to investigate how membrane curvature influences membrane-peptide interactions. The silica coating provided a suitable interface for fabricating conformal SLBs on nanoplasmonic sensing platforms with different architectural features. Spectral signatures in the measurement responses of each sensing platform are sensitive to specific regions of the sensor geometries and hence interpreting particular spectral features, and combinations thereof, can provide insight into how membrane curvature – the type and extent of which is determined by nanoscale architecture – influences biomacromolecular interaction processes, as exemplified by the binding interaction of AH peptide. Experimental Section Nanowell Array Fabrication. The nanowell arrays were fabricated via colloidal lithography on niobia fused silica substrates, as previously described.36 Briefly, the colloidal pattern was created in the form of a short-range ordered monolayer of monodisperse polystyrene particles ~150 nm in diameter, after which the diameter was reduced to ~100 nm by O2 plasma treatment.53 Gold and alumina were deposited by physical vapor deposition to a thickness of 30 nm and 20 nm, respectively. After removing the colloids, anisotropic dry etching was employed to create nanowells with a depth of ~200 nm. This was followed by the removal of alumina by cleaning with H2O2 and NH3.36 Finally, a conformal layer of silicon nitride was deposited to a thickness of 20 nm via plasma-enhanced chemical vapor deposition. Although the contrast in the relative peak and dip shifts (see below) is higher for nanowells in Nb2O5,37 it is still clearly observed also for silicon nitride nanowells and the fabrication is much simplified.36 Indirect Nanoplasmonic Sensing. Experiments were performed using fabricated silicon nitride-coated gold nanowell arrays, as well as silicon nitride-coated nanodisk arrays (Insplorion AB, Gothenburg, Sweden). The sensor chips were assembled within the measurement cell of the Insplorion XNano instrument (Insplorion AB) and ensemble-averaged nanoplasmonic sensing measurements were conducted in transmission mode, as previously described.30 Prior to each experiment, the sensor chips were thoroughly rinsed with 1 wt% sodium dodecyl sulfate (SDS) in water and ethanol and dried with a stream of nitrogen gas. The chips were then treated with oxygen plasma (Harrick Plasma, Ithaca, NY) before being loaded into the measurement cell. Of note, although silicon nitride was the base material for the conformal coatings, oxygen plasma treatment results in the formation of a silica layer on the sensor surface and hence the coatings are referred to as silica coatings.54 A peristaltic pump (Reglo Digital, Ismatec, Glattbrugg, Switzerland) was used to introduce liquid sample into the measurement cell at a constant flow rate of 100 µL/min. All measurement data were collected using the Insplorer software package (Insplorion AB) with a time resolution of 1 Hz. The

3 ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

resolution of particular spectral features, and their centroid positions (denoted as λ), was determined by high-order polynomial fitting.55 Results and Discussion Sensor Characterization To fabricate lipid-coated nanoplasmonic sensors with varying degrees of membrane curvature, two sensor arrays with distinct architectural configurations were utilized, and were comprised of randomly arranged silica-coated gold nanodisks and nanowells fabricated via hole-mask colloidal lithography (HCL)19 and colloidal lithography followed by dry etching, respectively.36 The nanodisk structure represents a shallow protrusion measuring 100 nm in diameter and 20 nm in height, with a sputter-coated conformal silica overlayer of ~10 nm thickness. The overlayer provides a solid support that is conducive to forming conformal SLBs with mainly planar geometries and a low degree of positive curvature (i.e., slightly convex) around the edges of the nanodisks (Figure 1a). On the other hand, the nanowell structure represents a deep cavity measuring 150 nm in diameter and 200 nm in depth, with a silica overlayer of 20 nm thickness. The sensor geometry is conducive to forming conformal SLBs with planar geometries as well as SLBs with sharp positive (i.e., highly convex) and sharp negative (i.e., highly concave) membrane curvatures37 (Figure 1b). Owing to the different sensor geometries, illumination with incident light yields slightly distinct optical phenomena arising from different plasmon modes. While the excitation of surface plasmons from non-interacting nanodisks on the sensor surface contributes to a localized surface plasmon resonance (LSPR) mode with a single peak in the extinction spectrum, the excitation of propagating and localized surface plasmons on the nanowell array results in an asymmetric resonance feature manifested by a peak and a dip in the extinction spectrum.33,36 In the latter, case, the near-field distribution and refractometric sensitivity at the peak wavelength corresponds to the planar surface and the upper rim of the nanowell. In contrast, the dip is mainly sensitive to refractive index changes inside the cavity, and the interior base exhibits negative membrane curvature. As the response sensitivities of the modes vary gradually along the vertical axis of the nanowell, molecular binding events can be spatially resolved. On the other hand, the nanodisk array consists of a mainly planar geometry (with a relatively low extent of positive curvature) to rule out sharp (highly convexed) curvaturedependent interactions. For both sensing platforms, the measurements were conducted in transmission mode and the corresponding extinction spectra yielded discernable measurement signatures. The extinction spectra from the nanodisk array in buffer exhibited a single peak located at around 720 nm (Figure 1c). On the other hand, the extinction spectra from the nanowell array in buffer exhibited two dips located at around 510 nm (short dip) and 835 nm (long dip) as well as a peak at around 675 nm (Figure 1d). The peak mainly arises from Bloch wave coupling to surface plasmon modes with symmetric charge distribution across the gold film.33 By contrast, the long dip corresponds to a resonance with a field enhancement that is more focused on the nanowell interior.33,53 The shorter wavelength dip is merely an effect of increasing absorption by gold in the blue spectral region. Following this line, we next evaluated the bulk refractive index sensitivities of the gold nanodisk peak as well as the gold nanowell peak and dips. This approach allowed us to evaluate sensing performance while also facilitating measurement comparison across sensing 4 ACS Paragon Plus Environment

Page 4 of 16

Page 5 of 16 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

Figure 1. (a) Schematic illustration of AH peptide interaction with conformal supported lipid bilayer on a silica-coated plasmonic nanodisk, accompanied by a representative top-down SEM image of a gold nanodisk array. (b) Schematic illustration of AH peptide interaction with conformal supported lipid bilayer on a silica-coated plasmonic nanowell, accompanied by a representative tilt-angle cross-sectional SEM image of a gold nanowell array; inset shows the magnified view of a single nanowell. Evolution of the extinction spectra during buffer-glycerol titration measurements (0 – 30 wt% glycerol in buffer) on (c) gold nanodisk and (d) gold nanowell arrays. (e) Bulk refractive index sensitivity plots of nanodisk peak responses from panel (c). (f) Bulk refractive index sensitivity plots of nanowell peak (black square) and dip (red circle) responses from panel (d). 5 ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

platforms. Indeed, we have previously shown that the normalization of absolute plasmonic peak shifts with the bulk refractive index sensitivity of different sensor surfaces enables the direct comparison of mass adsorbed across different surfaces.30 The bulk refractive index sensitivity of the gold nanodisk peak was determined to be around 101 nm/RIU (Figure 1e). While the short dip of the gold nanowell is insensitive to any change in bulk refractive index, the bulk refractive index sensitivities of the peak and long dip were found to be 136 nm/RIU and 220 nm/RIU, respectively (Figure 1f). These values are in good agreement with previous results, which showed that the bulk refractive index of the dip is generally higher than the peak.36,37 It was also determined that the different spectral signatures have sufficiently high resolution for detecting biomacromolecular interaction processes occurring at the sensor surface (Figure S1). Finite-difference time-domain (FDTD) simulations further revealed a strong field enhancement around the nanowell structure (Figure S2). While it is challenging to implement the exact structure geometry in simulations, the results clearly showed that the field enhancement inside the nanowells is higher for the dip than for the peak. This finding supports that the dip is sensitive to interactions occurring not only at the base of the nanowell but across a significant region of the nanowell sidewall as well. While the nanodisk and nanowell sensing platforms have distinct nanostructured geometries and involve excitation of different plasmon modes, it should also be noted that distortion to the electromagnetic field around different nanostructures and at different regions of the nanowell is likely to be minimal and that the decay lengths are relatively similar for the two platforms.56 Supported Lipid Bilayer Formation We then assessed SLB formation on the silica-coated nanodisk and nanowell arrays via vesicle fusion by tracking the plasmonic peak and dip shifts in real time. Optical transmission experiments were conducted using the same microfluidic flow chamber with equivalent diffusion flux (flow rate) for both substrates. Upon vesicle addition, the time-resolved wavelength shift showed a rapid increase in peak shift during the initial stage on the nanodisk array (Figure 2a). On the contrary, the increase in wavelength shifts occurred at a slower rate on the nanowell array, which is expected considering the geometrical differences between the two arrays. In the case of the gold nanodisk array, it is straightforward for lipid vesicles to contact active regions of the sensor surface as the nanodisk features are protruding and fully exposed.57 However, in the case of a nanowell array, planar areas at the top are relatively insensitive to adsorption of lipid vesicles and it is more challenging, and hence slower, for adsorbing vesicles to infiltrate into the nanowells58,59 where the sensitivity is higher. By comparing the size of lipid vesicles, which were around 60 nm diameter (Figure S3), to the dimensions of the nanowells (i.e., diameter of 100 nm and depth of 200 nm), it is clear that apart from accessing the nanowell, achieving optimal packing inside the nanowells leading to SLB formation is spatially constrained as well. The difference in vesicle adsorption behaviors on the planar surface regions versus in the interior of the nanowells also led to slightly varied peak and dip shift kinetics. Nonetheless, a kink in both time-resolved peak and dip shifts confirmed SLB formation across the nanowell array.8,11,29,60 The final peak shift arising from SLB formation on the nanodisk array was lowest at around 3.0 nm, compared to peak and dip shifts from the nanowell array, which were around 3.4 nm and 4.4 nm, respectively. This trend is expected since the bulk refractive index sensitivity of the nanodisk was also the lowest among the three tracked signals. After normalization with the respective bulk refractive index sensitivities, the normalized wavelength shifts of the 6 ACS Paragon Plus Environment

Page 6 of 16

Page 7 of 16 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

Figure 2. (a) Time-resolved peak and dip shifts tracking SLB formation on silica-coated nanowell array, and time-resolved peak shift tracking SLB formation on silica-coated nanodisk array via vesicle fusion using 100 mol% DOPC vesicles. (b) Comparison of final wavelength shifts before and after normalization with the respective bulk refractive index sensitivities. (c) Corresponding first-order derivative of the time-resolved peak and dip shifts. (d) Corresponding time-resolved normalized peak-to-dip shift ratio for SLB formation on silicacoated nanowell array. (e) Schematic comparison of SLB formation on nanodisk versus nanowell arrays. nanodisk peak, nanowell peak, and nanowell dip after SLB formation were around 0.030 RIU, 0.024 RIU, and 0.021 RIU, respectively (Figure 2b). The highest normalized wavelength shift observed from the nanodisk array may arise from a slightly more focused electromagnetic field of the nanodisk array compared to the nanowell array and, in all cases, the final measurement responses are consistent with SLB formation. The adsorption kinetics are also consistent with SLB formation, and were further analyzed by plotting the first-order derivative curves of the time-resolved peak shifts (Figure 2c). As expected, the initial rate of change of the peak shift on the nanodisk array was highest, followed by the initial rates of change of the wavelength shifts from the peak- and dip-sensitive regions of the nanowell. The onset of vesicle rupture also occurs at the earliest time point for the 7 ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

nanodisk array. Of note, while the initial rate of increase in wavelength shift is lower in the dip-sensitive region of the nanowell, the onset of rupture happened slightly earlier there compared to the peak-sensitive region. This suggests that a smaller amount of vesicles is required to initiate rupture inside the nanowell than on the top, planar surface. For example, when a vesicle enters a nanowell, it might form multiple contact points with the sidewalls/bottom and therefore experience more extensive vesicle-surface interactions, as opposed to when a vesicle adsorbs onto a planar surface.7 Ultimately, vesicle rupture and SLB formation occurred across both regions of the sensor surface. We further evaluated the relative mass uptake at the peak- and dip-sensitive regions of the nanowell by calculating the normalized peak-to-dip shift ratio. Uniform mass accumulation can be inferred when the normalized peak-to-dip shift ratio is equal to unity. Any significant deviation above unity would imply greater mass accumulation in the peak-sensitive region relative to the dip-sensitive region, or vice versa. Based on the time-resolved peak-to-dip shift ratio, we can conclude that vesicles quickly adsorbed on the top (i.e., planar surfaces) of the nanowell arrays in the initial stage when the peak-to-dip shift ratio rapidly increased to around 1.8 (Figure 2d). However, as vesicles started to fuse to form an SLB, the ratio eventually decreased to around 1.2, pointing to the formation of a conformal SLB across the entire array. The slight deviation above unity is mainly contributed by a slightly more focused electromagnetic around the peak-sensitive region. The overall SLB formation process on the nanodisk and nanowell arrays described above is depicted in Figure 2e. We also performed fluorescence recovery after photobleaching (FRAP) experiments to verify SLB conformality on both arrays. Fluorescence recovery was quickest on a flat silica surface without nanostructures, followed by nanodisk and nanowell arrays (Figure 3a). Full recovery

Figure 3. (a) FRAP measurements for SLBs formed on flat silica surface, silica-coated nanodisk array, and silica-coated nanowell array. (b) Corresponding FRAP recovery curves for SLBs formed on flat silica surface, silica-coated nanodisk array, and silica-coated nanowell array. (c) Comparison of diffusion coefficients for SLBs formed on the three substrates. (d) Comparison of corresponding mobile fraction percentages for SLBs formed on the three substrates. 8 ACS Paragon Plus Environment

Page 8 of 16

Page 9 of 16 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

on flat silica was obtained after 60 s while recovery on nanowell and nanodisk arrays only occurred after 90 s. The trend was quantified via normalized fluorescence intensity curves, which revealed distinct recovery profiles on the different surfaces. While a higher initial rate of fluorescence recovery was observed on flat silica, the final rates of recoveries were comparable and the fluorescence intensity curves on nanowell and nanodisk arrays overlap closely (Figure 3b). Based on the kinetics of fluorescence recovery, the lateral diffusivities of SLBs on flat silica, nanodisk, and nanowell arrays were calculated using the Hankel transform method61 and determined to be 1.64, 1.52, and 1.36 µm2/sec, respectively (Figure 3c). The corresponding mobile fractions were 77.18, 67.81, and 60.73%, respectively (Figure 3d). The magnitudes of the differences in diffusivities and mobile fractions are reasonable because the nanostructures present a larger effective area (i.e., compared to the planar equivalent) for diffusion, leading to an effective drop in the diffusion coefficient.8 The deep cavity nanostructures also lead to SLB regions with higher aspect-ratio features because the SLB does not lie contiguously on the same plane and these bent features make the SLB less mobile. In any case, these results provide evidence supporting the formation of relatively conformal SLBs on both arrays across planar as well as positively and negatively curved regions, for subsequent membrane-peptide interaction studies. AH Peptide Interaction with Supported Lipid Bilayer After establishing conformal SLBs on both arrays, we next proceeded to investigate membrane-peptide interactions by employing AH peptide. We injected AH peptide with successively increasing concentrations from 500 nM to 16 µM into the measurement chamber for both sensing arrays. In general, subtle changes to the time-resolved peak shift were observed at low concentrations and a noticeable increase in peak shift was detected at 8 µM and higher AH peptide concentrations (Figure 4a). To analyze the concentration-dependency effect, we normalized the absolute wavelength shifts with the bulk refractive index sensitivities of the respective signals and plotted the normalized wavelength shift versus AH peptide concentration (Figure 4b). For the interaction of AH peptide with the SLB-coated nanodisk array, a brief dip in the normalized wavelength shift was observed when the AH peptide concentration was increased up to 2 µM, above which the wavelength shift consistently increased up to a concentration of 16 µM. Interestingly, a similar trend was also observed for the AH peptide interaction with the SLB in the dip-sensitive region of the nanowell array. On the contrary, a gradual decrease in normalized peak shift was observed for AH peptide interactions with the SLB in the peak-sensitive region of the nanowell up to a concentration of 8 µM, above which a sharp decrease in wavelength shift was observed. The measurement responses from the peak- and dip-sensitive regions of the nanowell array moved in opposite directions, indicating striking differences in the membrane-peptide interaction behavior at regions of positive and negative membrane curvature (Figure 4c). While slight differences in the electromagnetic field distribution of the nanodisk and nanowell arrays might influence the magnitude of the signal responses, the varying direction of the measurement responses provides clear evidence that the membrane-peptide interaction strongly depends on membrane curvature. Since the peak shift signal from the shallow nanodisk (i.e., mainly arising from planar regions and those with mild positive curvature) displayed a similar trend to the nanowell dip shift signal, we can infer that the decrease in the nanowell peak shift is mainly arising from the highly positive curvature area of the peaksensitive region. In other words, by comparing the three wavelength shifts, we have factored out the contribution from planar and negatively curved regions. Taken together, these observations suggest that, at planar regions and those with mildly positive membrane 9 ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 4. (a) Time-resolved peak and dip shifts tracking AH peptide interaction with SLB on the nanowell array and time-resolved peak shift tracking AH peptide interaction with SLB on the nanodisk array, with stepwise increase in AH peptide concentration from 500 nM to 16 µM. (b) Corresponding plots of normalized extinction wavelength shifts versus AH peptide concentration. (c) Comparison of normalized extinction wavelength shifts after the addition of 16 µM AH peptide. (d) Corresponding time-resolved normalized peak-to-dip shift ratio arising from AH peptide interaction with SLB on the nanowell array. (e) Schematic illustration of AH peptide interaction with SLB at the peak- and dip-sensitive regions of the nanowell array as a function of AH peptide concentration. curvatures, AH peptide generally adsorbs onto the lipid bilayer with minimal to zero membrane disruption, as evidenced by modest positive wavelength shifts of the nanodisk peak and nanowell dip. However, at regions of highly positive membrane curvature, the peptide binds to the bilayer causing concentration-dependent membrane disruption up to a concentration of 8 µM AH peptide, above which significant membrane disruption occurs, as inferred from a negative wavelength shift of the nanowell peak.

10 ACS Paragon Plus Environment

Page 10 of 16

Page 11 of 16 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

To closely follow the relative mass changes at different regions of the nanowell array, we also analyzed the normalized peak-to-dip shift ratio over time. We found that upon addition of AH peptide at low concentrations, the normalized peak-to-dip shift ratio remained relatively constant at around 1.3 (Figure 4d). The peak-to-dip shift ratio then significantly decreased to around 1.1 upon addition of 16 µM AH peptide, suggesting a loss of mass at the peak-sensitive region. However, upon switching to buffer, the normalized peak-to-dip shift ratio increased again to slightly above 1.3, implying that there is significant reorganization of the SLB to occupy the disrupted regions. This also indicates that AH peptide only disrupts the membrane at concentrated regions of highly positive curvature, leaving the surrounding bilayer regions intact, and subsequently ensured a sufficient amount of lipid molecules for bilayer reorganization once excess peptide molecules are rinsed away. The process of AH peptide’s interaction with SLBs at different regions of the nanowell array and, at varying peptide concentrations as described above, are summarized in Figure 4e. Collectively, our findings demonstrate that AH peptide significantly disrupts the membrane only at regions of highly positive membrane curvature, and only at sufficiently high concentrations above 8 µM, even though the peptide still binds to other regions. As shown through previous works employing adsorbed vesicle layers48,49,62 as well as tethered single vesicles,47,51 the curvature-dependent AH peptide activity against the conformal lipid bilayer can be attributed to strain-dependent pore formation.47 At the same time, the concentration threshold observed in this work matches well with results from the work by Wang et al. whereby it was observed that AH peptides adsorbed onto SLBs up to a critical concentration at which it began to aggregate leading to membrane pore formation and SLB solubilization. 63 The agreement with previous works emphasizes the advantages of using SLB-coated arrays to study the effect of biomolecule concentration vis-à-vis membrane curvature. As opposed to utilizing intact vesicles of different sizes whereby there would be limited spatial control over the binding of biomolecules, the use of SLBs that are conformal on the nanostructured arrays ensures that biomolecular binding leading to membrane solubilization would only occur at sensitive regions of the measurement platform, ensuring sharper measurement responses. Conclusions In this work, we have utilized nanoplasmonic sensor arrays with different nanostructure geometries to investigate the effects of membrane curvature on membrane-peptide interactions. SLBs served as model membranes and were formed on silica-coated gold nanodisk and nanowell arrays via vesicle fusion. The bilayer formation process was tracked in real-time and the conformality and fluidic properties of the SLBs on the nanostructures were characterized. The effect of positive and negative membrane curvatures, along with planar regions, on membrane-peptide interactions was then systematically investigated using AH peptide as a curvature-sensitive probe. The introduction of increasing concentrations of AH peptide to the bilayer-coated nanodisk array resulted in a gradual increase in the LSPR peak shift. On the contrary, a net decrease in peak shift but a gradual increase in dip shift was observed using the bilayer-coated nanowell array. These measurement responses suggest that, for planar and negatively curved membrane regions, AH peptide binds to the bilayer with negligible membrane solubilization across all tested peptide concentrations. On the other hand, at membrane regions of highly positive curvature, AH peptide binds to the bilayer with minimal disruption up to a concentration of 8 µM, above which significant membrane disruption occurs. These findings highlight the potential of developing real-time nanoplasmonic sensing strategies that are composed of multiple plasmonic signals with high sensitivity to distinct sensing regions with different nanostructured curvatures. Taken together, our results 11 ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

demonstrate how nanoplasmonic arrays with different geometries can be employed to overcome challenges associated with quantifying the effects of membrane curvature at the nanoscale. At the same time, our work highlights the excellent potential of integrating highprecision nanofabrication techniques with nanoplasmonic sensing approaches for quantitative biointerfacial science. Acknowledgements This work was supported by the National Research Foundation of Singapore through a Competitive Research Programme grant (NRF-CRP10-2012-07) and a Proof-of-Concept grant (NRF2015NRF-POC0001-19) as well as through the Center for Precision Biology at Nanyang Technological University. The authors thank G.J. Ma for technical assistance with experiments. Supporting Information The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/ Additional data is provided for FDTD simulation results (Figure S1), measurements of baseline signal stabilities (Figure S2) and size distribution of lipid vesicles (Figure S3). References (1) Anker, J. N.; Hall, W. P.; Lyandres, O.; Shah, N. C.; Zhao, J.; Van Duyne, R. P. Nat. Mater. 2008, 7, 442. (2) Jackman, J. A.; Rahim Ferhan, A.; Cho, N.-J. Chem. Soc. Rev. 2017, 46, 3615-3660. (3) Szunerits, S.; Boukherroub, R. Chem. Commun. 2012, 48, 8999-9010. (4) Bruzas, I.; Unser, S.; Yazdi, S.; Ringe, E.; Sagle, L. Anal. Chem. 2016, 88, 7968-7974. (5) Unser, S.; Bruzas, I.; He, J.; Sagle, L. Sensors 2015, 15, 15684. (6) Pfeiffer, I.; Petronis, S.; Köper, I.; Kasemo, B.; Zäch, M. J. Phys. Chem. B 2010, 114, 46234631. (7) Pfeiffer, I.; Seantier, B.; Petronis, S.; Sutherland, D.; Kasemo, B.; Zäch, M. J. Phys. Chem. B 2008, 112, 5175-5181. (8) Jonsson, M. P.; Jönsson, P.; Dahlin, A. B.; Höök, F. Nano Lett. 2007, 7, 3462-3468. (9) Larsson, E. M.; Edvardsson, M. E. M.; Langhammer, C.; Zorić, I.; Kasemo, B. Rev. Sci. Instrum. 2009, 80, 125105. (10) Ferhan, A.; Ma, G.; Jackman, J.; Sut, T.; Park, J.; Cho, N.-J. Sensors 2017, 17, 1484. (11) Ferhan, A. R.; Jackman, J. A.; Cho, N.-J. Anal. Chem. 2016, 88, 12524-12531. (12) Ferhan, A. R.; Jackman, J. A.; Cho, N.-J. Phys. Chem. Chem. Phys. 2017, 19, 2131-2139. 12 ACS Paragon Plus Environment

Page 12 of 16

Page 13 of 16 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

(13) Ferhan, A. R.; Jackman, J. A.; Cho, N.-J. Anal. Chem. 2017, 89, 4301-4308. (14) Jackman, J. A.; Ferhan, A. R.; Yoon, B. K.; Park, J. H.; Zhdanov, V. P.; Cho, N.-J. Anal. Chem. 2017. (15) Jackman, J. A.; Yorulmaz Avsar, S.; Ferhan, A. R.; Li, D.; Park, J. H.; Zhdanov, V. P.; Cho, N.-J. Anal. Chem. 2017, 89, 1102-1109. (16) Jackman, J. A.; Zhdanov, V. P.; Cho, N.-J. Langmuir 2014, 30, 9494-9503. (17) Lopez Gerardo, A.; Estevez, M. C.; Soler, M.; Lechuga Laura, M. In Nanophotonics, 2017, p 123. (18) Spackova, B.; Wrobel, P.; Bockova, M.; Homola, J. Proc. IEEE 2016, 104, 2380-2408. (19) Langhammer, C.; Larsson, E. M.; Kasemo, B.; Zorić, I. Nano Lett. 2010, 10, 3529-3538. (20) He, J.; Boegli, M.; Bruzas, I.; Lum, W.; Sagle, L. Anal. Chem. 2015, 87, 11407-11414. (21) Nathan, C. L.; Prashant, N.; Kevin, M. M.; David, J. N.; Sang-Hyun, O. Rep. Prog. Phys. 2012, 75, 036501. (22) Dahlin, A. B.; Wittenberg, N. J.; Höök, F.; Oh, S.-H. Nanophotonics 2013, 2, 83-101. (23) Tatsuno, T.; Okamoto, T.; Ezaki, T.; Isobe, T.; Nakajima, A.; Matsushita, S. Bull. Chem. Soc. Jpn. 2016, 89, 369-374. (24) Ohno, T.; Wadell, C.; Inagaki, S.; Shi, J.; Nakamura, Y.; Matsushita, S.; Sannomiya, T. Opt. Mater. Express 2016, 6, 1594-1603. (25) Oliverio, M.; Perotto, S.; Messina, G. C.; Lovato, L.; De Angelis, F. ACS Appl. Mater. Interfaces 2017, 9, 29394-29411. (26) Yangyang, W.; Jianhua, Z.; Jinghong, L. Small Methods 2017, 1, 1700197. (27) Špačková, B.; Lynn, N. S.; Slabý, J.; Šípová, H.; Homola, J. ACS Photonics 2018, 5, 10191025. (28) Hackett, L. P.; Ameen, A.; Li, W.; Dar, F. K.; Goddard, L. L.; Liu, G. L. ACS Sensors 2018, 3, 290-298. (29) Zan, G. H.; Jackman, J. A.; Kim, S.-O.; Cho, N.-J. Small 2014, 10, 4828-4832. (30) Jackman, J. A.; Spackova, B.; Linardy, E.; Kim, M. C.; Yoon, B. K.; Homola, J.; Cho, N.J. Chem. Commun. 2016, 52, 76-79. (31) Im, H.; Lee, S. H.; Wittenberg, N. J.; Johnson, T. W.; Lindquist, N. C.; Nagpal, P.; Norris, D. J.; Oh, S.-H. ACS Nano 2011, 5, 6244-6253. (32) Junesch, J.; Sannomiya, T.; Dahlin, A. B. ACS Nano 2012, 6, 10405-10415. (33) Sannomiya, T.; Scholder, O.; Jefimovs, K.; Hafner, C.; Dahlin, A. B. Small 2011, 7, 16531663. 13 ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(34) Dahlin, A. B.; Mapar, M.; Xiong, K.; Mazzotta, F.; Höök, F.; Sannomiya, T. Adv. Opt. Mater. 2014, 2, 556-564. (35) Junesch, J.; Sannomiya, T. ACS Appl. Mater. Interfaces 2014, 6, 6322-6331. (36) Malekian, B.; Xiong, K.; Emilsson, G.; Andersson, J.; Fager, C.; Olsson, E.; LarssonLanghammer, E.; Dahlin, A. Sensors 2017, 17, 1444. (37) Junesch, J.; Emilsson, G.; Xiong, K.; Kumar, S.; Sannomiya, T.; Pace, H.; Voros, J.; Oh, S.-H.; Bally, M.; Dahlin, A. B. Nanoscale 2015, 7, 15080-15085. (38) Antonny, B. Annu. Rev. Biochem. 2011, 80, 101-123. (39) Iversen, L.; Mathiasen, S.; Larsen, J. B.; Stamou, D. Nat. Chem. Biol. 2015, 11, 822. (40) Jarsch, I. K.; Daste, F.; Gallop, J. L. J. Cell Biol. 2016, 214, 375. (41) Kegulian, N. C.; Sankhagowit, S.; Apostolidou, M.; Jayasinghe, S. A.; Malmstadt, N.; Butler, P. C.; Langen, R. J. Biol. Chem. 2015, 290, 25782-25793. (42) Koller, D.; Lohner, K. Biochim. Biophys. Acta, Biomembr. 2014, 1838, 2250-2259. (43) Larsen, J. B.; Jensen, M. B.; Bhatia, V. K.; Pedersen, S. L.; Bjørnholm, T.; Iversen, L.; Uline, M.; Szleifer, I.; Jensen, K. J.; Hatzakis, N. S.; Stamou, D. Nat. Chem. Biol. 2015, 11, 192. (44) McMahon, H. T.; Boucrot, E. J. Cell Sci. 2015, 128, 1065. (45) Strandberg, E.; Tiltak, D.; Ehni, S.; Wadhwani, P.; Ulrich, A. S. Biochim. Biophys. Acta, Biomembr. 2012, 1818, 1764-1776. (46) Sugiura, Y.; Ikeda, K.; Nakano, M. Langmuir 2015, 31, 11549-11557. (47) Tabaei, S. R.; Rabe, M.; Zhdanov, V. P.; Cho, N.-J.; Höök, F. Nano Lett. 2012, 12, 57195725. (48) Jackman, J. A.; Zan, G. H.; Zhdanov, V. P.; Cho, N.-J. J. Phys. Chem. B 2013, 117, 1611716128. (49) Jackman, J. A.; Goh, H. Z.; Zhdanov, V. P.; Knoll, W.; Cho, N.-J. J. Am. Chem. Soc. 2016, 138, 1406-1413. (50) Hatzakis, N. S.; Bhatia, V. K.; Larsen, J.; Madsen, K. L.; Bolinger, P.-Y.; Kunding, A. H.; Castillo, J.; Gether, U.; Hedegård, P.; Stamou, D. Nat. Chem. Biol. 2009, 5, 835. (51) Jackman, J. A.; Saravanan, R.; Zhang, Y.; Tabaei, S. R.; Cho, N.-J. Small 2015, 11, 23722379. (52) Manneville, J.-B.; Leduc, C.; Sorre, B.; Drin, G. In Methods Cell Biol., Di Paolo, G.; Wenk, M. R., Eds.; Academic Press, 2012, pp 47-71. (53) Xiong, K.; Emilsson, G.; Dahlin, A. B. Analyst 2016, 141, 3803-3810. 14 ACS Paragon Plus Environment

Page 14 of 16

Page 15 of 16 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

(54) Jiménez, C.; Perrière, J.; Vickridge, I.; Enard, J. P.; Albella, J. M. Surf. Coat. Technol. 1991, 45, 147-154. (55) Dahlin, A. B.; Tegenfeldt, J. O.; Höök, F. Anal. Chem. 2006, 78, 4416-4423. (56) Mazzotta, F.; Johnson, T. W.; Dahlin, A. B.; Shaver, J.; Oh, S.-H.; Höök, F. ACS Photonics 2015, 2, 256-262. (57) Lynn, N. S.; Homola, J. Anal. Chem. 2016, 88, 12145-12151. (58) Escobedo, C.; Brolo, A. G.; Gordon, R.; Sinton, D. Anal. Chem. 2010, 82, 10015-10020. (59) Zhao, Y.; Gaur, G.; Mernaugh, R. L.; Laibinis, P. E.; Weiss, S. M. Nanoscale Res. Lett. 2016, 11, 395. (60) Larsson, E. M.; Edvardsson, M. E. M.; Langhammer, C.; Zorić, I.; Kasemo, B. Rev. Sci. Instrum. 2009, 80, 125105. (61) Jönsson, P.; Jonsson, M. P.; Tegenfeldt, J. O.; Höök, F. Biophys. J. 2008, 95, 5334-5348. (62) Tabaei, S. R.; Choi, J.-H.; Haw Zan, G.; Zhdanov, V. P.; Cho, N.-J. Langmuir 2014, 30, 10363-10373. (63) Wang, J.; Liu, K.-W.; Biswal, S. L. Anal. Chem. 2014, 86, 10084-10090.

15 ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

For TOC Only

16 ACS Paragon Plus Environment

Page 16 of 16