www.acsnano.org
Monovalent and Oriented Labeling of Gold Nanoprobes for the High-Resolution Tracking of a Single-Membrane Molecule Downloaded via NOTTINGHAM TRENT UNIV on July 20, 2019 at 01:12:38 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
Yi-Hung Liao, Chih-Hsiang Lin, Ching-Ya Cheng, Wai Cheng Wong, Jz-Yuan Juo, and Chia-Lung Hsieh* Institute of Atomic and Molecular Sciences (IAMS), Academia Sinica, Taipei 10617, Taiwan S Supporting Information *
ABSTRACT: Single-molecule tracking is a powerful method to study molecular dynamics in living systems including biological membranes. High-resolution single-molecule tracking requires a bright and stable signal, which has typically been facilitated by nanoparticles due to their superb optical properties. However, there are concerns about using a nanoparticle to label a single molecule because of its relatively large size and the possibility of cross-linking multiple target molecules, both of which could affect the original molecular dynamics. In this work, using various labeling schemes, we investigate the effects using nanoparticles to measure the diffusion of single-membrane molecules. By conjugating a low density of streptavidin (sAv) to gold nanoparticles (AuNPs) of different sizes (10, 15, 20, 30, and 40 nm), we isolate and quantify the effect of the particle size on the diffusion of biotinylated lipids in supported lipid bilayers (SLBs). We find that single sAv tends to cross-link two biotinylated lipids, leading to a much slower diffusion in SLBs. We further demonstrate a simple and robust strategy for the monovalent and oriented labeling of a single lipid molecule with a AuNP by using naturally dimeric rhizavidin (rAv) as a bridge, thus connecting the biotinylated nanoparticle surface and biotinylated target molecule. The rAv−AuNP conjugate demonstrates fast and free diffusion in SLBs (2−3 μm2/s for rAv−AuNP sizes of 10−40 nm), which is comparable to the diffusion of dye-labeled lipids, indicating that the adverse size and cross-linking effects are successfully avoided. We also note that the diffusion of dyelabeled lipids critically depends on the choice of dye, which could report different diffusion coefficients by about 20% (2.2 μm2/s of ATTO647N and 2.6 μm2/s of ATTO532). By comparing the diffusion of the uniformly and randomly oriented labeling of a single lipid molecule with a AuNP, we conclude that oriented labeling is favorable for measuring the diffusion of single-membrane molecules. Our work shows that the measured diffusion of the membrane molecule is highly sensitive to the molecular design of the cross-linker for labeling. The demonstrated approach of monovalent and oriented AuNP labeling provides the opportunity to study single-molecule membrane dynamics at much higher spatiotemporal resolutions and, most importantly, without labeling artifacts. KEYWORDS: single-particle tracking, gold nanoparticle, monovalent labeling, lipid bilayer membranes, cross-linking, lipid diffusion
T
scopic membrane dynamics calls for measurements with high spatial and temporal resolution. Single-molecule tracking is a powerful method for studying membrane dynamics. By attaching optical probes to molecules of interest, the position of a single-membrane molecule can be determined by detecting and localizing the optical signal of the probe. Fluorophores are common optical probes for membrane molecules, which facilitate single-molecule tracking under fluorescence microscopy. Unfortunately, the limited photon
he biological membrane is a complex and dynamic system in which important cell activities occur. The fluidity of the membrane facilitates the lateral mobility of various kinds of lipids and proteins. Increasing evidence has shown that the cell membrane is heterogeneous at the nanoscale.1−3 Membrane proteins are found to assemble into nanoclusters.4,5 In addition, through lipid−lipid and lipid− protein interactions, nanosized membrane domains, such as lipid rafts, can form.6,7 Dynamic interplay between individual membrane molecules and association with membrane nanodomains are found to be essential for membrane functions.8−10 Deeper understanding of the governing principles of nano© XXXX American Chemical Society
Received: February 12, 2019 Accepted: June 24, 2019 Published: June 24, 2019 A
DOI: 10.1021/acsnano.9b01176 ACS Nano XXXX, XXX, XXX−XXX
Article
Cite This: ACS Nano XXXX, XXX, XXX−XXX
Article
ACS Nano
Figure 1. (a) Schematic diagram of tracking single lipid molecules in the SLB through AuNP labeling. The schematic is plotted to scale. A low density of sAv is conjugated to AuNPs of various sizes for targeting biotinylated lipid in bilayer membranes. Insets show the iSCAT optical images of single sAv−AuNPs of corresponding sizes; scale bars: 500 nm. (b) Diffusion trajectories of 20 nm AuNPs obtained from our SPT experiments.
budget of fluorophores (as a result of photobleaching) leads to a compromise between the data acquisition rate and total observation time. Moreover, saturation due to the fluorescence lifetime sets the maximum signal flux available for detection. To resolve the motion of membrane molecules at a smaller length scale and shorter time scale, nanoparticles with proper optical properties, such as gold nanoparticles (AuNPs)3,11−17 and quantum dots (Qdots),18−21 are commonly used as probes because of their strong and reliable optical signal. Scatteringbased single-particle tracking (SPT) is particularly promising for high-speed and high-precision measurements because of the indefinite and stable scattering signal. For example, using 40 nm AuNPs as labels, single-molecule dynamics were unveiled in the plasma membranes of living cells by SPT at 40000 frames per second (fps).3 More recently, interferometric scattering (iSCAT) microscopy has enabled ultrahigh-speed SPT of smaller AuNPs, as small as 20 nm, with nanometer spatial precision and microsecond temporal resolution.12−17 By contrast-enhanced microscopy, SPT on bilayer membranes has been demonstrated by using 10 nm AuNPs with a 1 ms time resolution.22 There are concerns when using nanoparticles as labels for membrane molecules.23,24 At first glance, the size of the nanoparticle is typically much larger than that of the target molecule, and therefore, the nanoparticle labeling may affect the motion of the target molecule.23 It is worth noting that the target molecule resides in the membrane with high viscosity, and the nanoparticle is surrounded by an aqueous environment with low viscosity.24 For example, the diffusion of a 40 nm nanoparticle alone in solution is approximately 10 μm2/s, faster than the diffusion of a lipid molecule in most bilayer membrane systems (typically 1−5 μm2/s). As a result, although without further supporting evidence, the hydrodynamic loading of the nanoparticle on the molecule has been
assumed to be insignificant.24 However, SPT measurements generally report a diffusion of the membrane molecule slower than that measured with small fluorophore labels. For example, in a simple model membrane system where a fluidic phospholipid bilayer is deposited on a solid support, dyelabeled lipids typically give a diffusion coefficient of 2−4 μm2/s,25,26 whereas nanoparticle-labeled lipids result in a diffusion coefficient of 75%) bind to two biotinylated lipids, and the rest mainly bind to only one biotinylated lipid. The three-lipid binding of sAv seems to be largely prohibited by conjugating sAv to the surface of a AuNP where one biotin binding site of sAv is occupied for particle attachment. We also note a consistent trend that the population fraction of very slowly diffusive particles (D < 0.5 μm2/s) grows when the size of the particle increases (e.g., 9% for 10 nm sAv−AuNP and 31% for 40 nm sAv−AuNP; see Table 1). This very slow population is attributed to multivalent binding to three or even more biotinylated lipids, possibly cross-linked by more than one sAv presented on the particle surface. A larger particle has a smaller surface curvature, and thus, this makes it easier for multiple sAvs to attach to the membrane simultaneously. Future investigation is needed to interpret the cause of the very slow diffusive population in a more quantitative manner. Effect of Particle Size to the Measured Diffusion of sAv−AuNPs. Given the diffusion coefficients of monovalent and bivalent sAv−AuNPs of various sizes, we estimate the effect of the size of the AuNP on the measured diffusion coefficient. Figure 3a plots the histograms of the measured diffusion coefficients of sAv−AuNPs of the five sizes. The diffusion coefficients of bivalent and monovalent binding modes, namely, Dbi and Dmono, of 10, 15, 20, 30, and 40 nm sAv−AuNPs are listed in Table 1, and they are plotted in Figure 3b as a function of the particle size. Both Dmono and Dbi can be fitted to straight lines, showing that the measured diffusion coefficient is indeed affected by the size of the AuNP within the size range we explored. To retrieve genuine dynamics of the targeted molecules, hydrodynamic loading of the probe has to be sufficiently small. In our case, the diffusion of the AuNP in the aqueous environment needs to be much faster than the diffusion of the target membrane molecule.24 Otherwise, the loading of the probe would slow down the measured diffusion coefficient.
midplane (the interface between the top and bottom membrane leaflets) is lost.30 In the SLB system, previous experimental data and theoretical calculations have shown that the diffusion coefficient of a lipid complex is roughly inversely proportional to the number of lipid molecules of the complex.30,36,37 Our data of 20 nm sAv−AuNP show that the diffusion coefficient of bivalent binding is approximately one-half of that of monovalent binding, which is in good agreement with the previous studies. We stress that our measured diffusion coefficient of the nanoparticle is highly sensitive to the number of attached lipids because there is a supporting substrate nearby.30 In SLBs, the bilayer is very close to the substrate, separated by a thin water layer of 1−2 nm.38 The opposite scenario would occur in freestanding membranes, in which both membrane leaflets face bulk aqueous solution where the molecular diffusion is faster25 and less sensitive to the size of membrane inclusion and crosslinking.39 To illustrate the significance of the substrate effect in enhancing the slowdown of diffusion caused by cross-linking, we perform SPT of sAv−AuNPs in multilayer SLBs where the substrate effect is diminished (Supporting Information).40 The sAv−AuNPs attach and diffuse on the top bilayer of the multilayer SLBs that is further away from the substrate, and thus, it experiences a reduced substrate effect. We measure the free and single model diffusion of 20 nm sAv−AuNP in multilayer SLBs with a greater diffusion coefficient of 3.81 ± 0.81 μm2/s (Figure S5). The faster diffusion in multilayer SLBs indicates that the coupling between the top bilayer and the substrate is indeed weaker than that in single-layer SLBs. Importantly, the unimodal diffusion of sAv−AuNPs in multilayer SLBs denotes that the diffusion coefficient is no longer sensitive to cross-linking, as the bivalent and monovalent bindings of sAv−AuNPs are also expected in the multilayer SLBs. Thus, having a substrate close to the bilayer is essential for detecting cross-linking from the measured diffusion coefficient. High Tendency of Bivalent Binding of sAv−AuNP to Biotinylated Lipids in the cis Configuration. Our data show that our sAv−AuNP tends to attach to two biotinylated lipids. To quantify the tendency of bivalent over monovalent labeling and their diffusion coefficients, we fit the histogram of measured diffusion coefficients with two Gaussian functions. For example, for the sAv−20 nm AuNP, we found the population ratio of monovalent to bivalent binding is approximately 1/3. The diffusion coefficients of bivalent and monovalent binding of the 20 nm sAv−AuNP are 1.24 ± 0.16 and 2.07 ± 0.28 μm2/s, respectively, whose ratio is 0.6. This ratio is important for understanding how our sAv−AuNP cross-links two lipid molecules. Camley and Brown have studied how the diffusion of lipids is slowed down by crosslinking in SLBs by computation.30 They found that the slowdown of the collective diffusion of two lipids associated by a protein (corresponding to sAv in our case) critically depends on the distance between the two lipids. In an extreme scenario where the two lipids are infinitely close, they diffuse together as one entity, and the slowdown effect is minor. When the distance between two associated lipids increases, the diffusion becomes slower. The slowdown ratio, defined as the ratio of the diffusion coefficient of two associated lipids (Dbi) to that of a single lipid (Dmono), reaches a value of 0.5 when the distance is greater than 6 nm.30 Based on the computational results and our experimental data of a slowdown factor of 0.6, the two lipids cross-linked by sAv are approximately 2.3 nm apart. This E
DOI: 10.1021/acsnano.9b01176 ACS Nano XXXX, XXX, XXX−XXX
Article
ACS Nano
This explains the results from our data that increasing the particle size, Dmono (black dots in Figure 3b), decreases more prominently than Dbi (red dots in Figure 3b). Quantitatively, Dmono of the 40 nm AuNP is 50% of that of the 10 nm AuNP, whereas Dbi of the 40 nm AuNP is 60% of that of the 10 nm AuNP. Our interpretation is that the sAv-labeled single lipid diffuses intrinsically faster than the sAv-labeled two-lipid complex in SLBs. Thus, to measure the true diffusion of a single lipid without slowdown, it requires a smaller AuNP (less loading) than the measurement of a two-lipid complex. Our data show that the 15 nm AuNP reports a Dmono slower than the 10 nm AuNP (2.34 ± 0.25 μm2/s versus 2.47 ± 0.25 μm2/s, which is statistically significant with a p value
