Yashashwa
Pandey
,
Andrea
Ingold
,
Naresh
Kumar
* and
Renato
Zenobi
*
Department of Chemistry and Applied Biosciences, ETH Zurich, Vladimir-Prelog-Weg 3, 8093 Zurich, Switzerland. E-mail: naresh.kumar@org.chem.ethz.ch; zenobi@org.chem.ethz.ch
First published on 6th May 2024
Supported lipid membranes are an important model system to study the phase separation behavior at the nanoscale. However, the conventional nanoanalytical tools often fail to provide reliable chemical characterization of the phase separated domains in a non-destructive and label-free manner. This study demonstrates the application of scanning tunneling microscopy-based tip-enhanced Raman spectroscopy (TERS) to study the nanoscale phase separation in supported d62-DPPC:
DOPC lipid monolayers. Hyperspectral TERS imaging successfully revealed a clear segregation of the d62-DPPC-rich and DOPC-rich domains. Interestingly, nanoscale deposits of d62-DPPC were observed inside the DOPC-rich domains and vice versa. High-resolution TERS imaging also revealed the presence of a 40–120 nm wide interfacial region between the d62-DPPC-rich and DOPC-rich domains signifying a smooth transition rather than a sharp boundary between them. The novel insights obtained in this study demonstrate the effectiveness of TERS in studying binary lipid monolayers at the nanoscale.
Supported lipid monolayers in particular, are a valuable model system to study membrane biophysics.22 Since a monolayer is half of a bilayer, it can be effectively used to explore various interfacial processes, such as interactions between proteins and membranes,10 drugs and membranes,23 as well as enzymatic reactions.24 Furthermore, they are an excellent biomimicking systems for studying pulmonary surfactant monolayer in the alveoli of lungs.25 Studying phase segregation of supported lipid monolayers at the nanoscale is crucial for the understanding of biological processes as well as for potential technological applications.16,26–33 However, the inherently heterogeneous nature of lipid domains and the limitations of the current analytical tools present significant challenges. Super-resolution fluorescence microscopy, nanoscale infrared spectroscopy, and nanoscale secondary ion mass spectrometry are techniques that in principle have the necessary capabilities. However, they suffer from serious limitations such as altering the native behavior of lipids due to the need for fluorophore labelling,34,35 insufficient sensitivity,36 or lack of detailed chemical information due to severe molecular fragmentation.37
On the contrary, tip-enhanced Raman spectroscopy (TERS) offers several advantages over other techniques for studying phase separation of lipid monolayers at the nanoscale level. TERS combines the high spatial resolution of scanning probe microscopy with the molecular specificity and sensitivity of surface-enhanced Raman spectroscopy, enabling the acquisition of chemically specific information with nanoscale spatial resolution.38–40 Importantly, TERS does not require fluorescent probes, thus allowing for a label-free analysis, and eliminating potential perturbations in the system.41 Hyperspectral TERS imaging has been successfully used to probe phospholipid species.7,42–44 For example, in our previous work, we demonstrated reproducibility of the hyperspectral TERS imaging on supported DPPC monolayers.42 Furthermore, high-resolution TERS imaging revealed differences in the ordering of DPPC molecules at the nanoscale level.
The outer leaflet of cell membranes has a surface pressure of ca. 30 mN m−1.45 At such surface pressures, formation of stable nanometer-sized phase-separated domains has been observed in supported lipid monolayers.46–48 However, phase separation behavior of these nanometer-sized domains is not well understood, primarily due to the lack of sensitivity and spatial resolution of the current analytical methods. Herein, TERS imaging was employed to study the phase separation behavior of the nanometer-sized domains in d62-1,2-dipalmitoyl-sn-glycero-3-phosphocholine: 1,2-dioleoyl-sn-glycero-3-phosphocholine (d62-DPPC:
DOPC) lipid monolayers supported on Au(111) transferred at a high surface pressure of 30 mN m−1 using Langmuir–Blodgett (LB) method. Hyperspectral TERS imaging successfully revealed the phase separation of d62-DPPC-rich and DOPC-rich domains. Interestingly, nanoscale deposits of d62-DPPC were observed in the DOPC-rich domains and vice versa. High-resolution TERS imaging also revealed the presence of a 50–100 nm wide interfacial region between the d62-DPPC-rich and DOPC-rich domains in supported d62-DPPC
:
DOPC monolayer signifying a smooth transition rather than a sharp boundary between them. This study demonstrates the potential of STM-TERS as a non-invasive, label-free analytical tool for imaging biomimetic phase-separated lipid systems at the nanoscale level.
![]() | ||
Fig. 1 Schematic diagram of the top-illumination STM-TERS setup used to investigate nanoscale phase separation in supported lipid monolayers in this work. |
The presence of the lipid monolayer on Au(111) surface was first verified using atomic force microscopy (AFM). Fig. S2† shows an AFM topography image of a d62-DPPC:
DOPC monolayer on a Au(111) surface, where numerous lipid patches are visible. The height of the transferred monolayer was found to be 2.5 ± 0.2 nm, which is in good agreement with previous reports.52,53
Simultaneous STM and hyperspectral TERS imaging of the supported d62-DPPC:
DOPC monolayer is shown in Fig. 2. Fig. 2a displays the STM topography image of the sample measured with the TERS image with a step size of 40 nm, where “high” and “low” refer to the relative height. It is well known that STM height can be influenced by the local electronic properties of the sample and the applied bias.54,55 Therefore, the absolute values may not represent the true height and only provide a qualitative description of the sample surface. The dark features visible in Fig. 2a represent a more disordered arrangement of lipid molecules (see discussion of Fig. 3d). Notably, even though the thickness of lipid monolayer is ca. 2.6 nm (Fig. S2†), electron tunneling between the STM tip and the sample surface still works effectively. The tunneling of electrons through lipids/alkanethiols has also been demonstrated in previous studies,54,56 although its precise mechanism is not entirely clear yet. Some studies indicate that the tunneling process could be facilitated by interfacial water57,58 and/or electron hopping through the lipid chains.8,45,54,56,59
![]() | ||
Fig. 3 (a) t-SNE dimension plot of the first five PCs obtained from the hyperspectral TERS image of d62-DPPC![]() ![]() |
The averaged spectrum of the hyperspectral TERS image showing the C–D and the C–H stretching regions is presented in Fig. 2b. The C–D stretching signal originates from the d62-DPPC, whereas the C–H stretching signal is dominated by the DOPC molecules as shown by the comparison of reference Raman spectra in Fig. S3.† A detailed assignment of the TERS bands is presented in Table S1.†Fig. 2c shows TERS image of the C–D signal where a heterogenous distribution of the d62-DPPC molecules is observed with a large d62-DPPC-rich domain located in the center surrounded by smaller domains. TERS image of the C–D and C–H signals ratio (IC–D/IC–H) presented in Fig. S4† confirms aggregation of the d62-DPPC molecules into distinct phase-separated domains.
Fig. 2d shows the TERS image of the C–H signal, which also shows a non-uniform distribution of DOPC molecules including a few regions of very low TERS signal. It should be noted that the TERS tip does not penetrate the lipid monolayer since several features observed in the TERS images in Fig. 2c and d correlate very well with the sample topography. This would not be the case if the TERS tip was penetrating the lipid monolayer. Furthermore, we can also rule out the possibility of the TERS tip being contaminated with lipid molecules and generating a SERS signal since a contaminated tip would give a very similar intensity of Raman signal throughout the TERS image. Notably, while the STM topography image provides no chemical information, the TERS image provides a detailed chemical characterization of the sample.
Further insight into the distribution of d62-DPPC and DOPC molecules is gained from the overlay of C–D and C–H signal TERS images in Fig. 2e. Notably, the separation of domains is quite apparent in the overlay image, where the blue areas represent a relatively higher concentration of DOPC and the red and pink areas represent a relatively higher concentration of d62-DPPC. Interestingly, it is found that the d62-DPPC-rich and DOPC-rich domains are not 100% “pure”. Instead, they exhibit a significant heterogeneity with the presence of several nanoscale d62-DPPC pockets inside the DOPC-rich domains and vice versa as highlighted by the arrows in Fig. 2e. For example, the TERS spectra measured at positions 1 and 2 marked in Fig. 2e are displayed in Fig. S5.† Position 1 shows a relatively higher C–H signal, whereas position 2 shows a relatively higher C–D signal indicating the predominance of DOPC and d62-DPPC at these locations, respectively. It should be noted that the previous AFM studies have reported submicron-size phase-separated lipid domains at 30 mN m−1 surface pressure based on the sample topography.46,60,61 However, no chemical information regarding the segregated lipid domains could be obtained in these studies.
Note that the spatial resolution of the TERS images in Fig. 2d and e is limited by the step size of hyperspectral imaging, which was set to 40 nm. In several areas of Fig. 2c, a strong C–D signal is observed in isolated single pixels or along a line of single pixels. If we define spatial resolution as the smallest feature distinguished in the sample, then the spatial resolution of the TERS imaging in Fig. 2 is estimated to be ≈40 nm.
To further analyze the phase separation in the d62-DPPC:
DOPC monolayer, principal component analysis (PCA), t-distributed stochastic neighbor embedding (t-SNE) and hierarchical clustering analysis (HCA) of the hyperspectral data measured in the TERS image was performed. t-SNE analysis was performed on the first five principal components (PCs) obtained from PCA and the reduced dimensions obtained from t-SNE were grouped using HCA, which showed that the TERS spectra clustered into three distinct groups (0, 1 and 2) as depicted in Fig. 3a. The number of clustering groups was selected after comparing the average spectra of several clusters and finding the minimum number of clusters which gave unique TERS spectra. Average TERS spectra corresponding to the three distinct clusters in Fig. 3a are shown in Fig. 3b, where cluster ‘0’ has a relatively higher C–D signal and therefore is assigned to the d62-DPPC-rich domains, cluster ‘2’ has a relatively high C–H signal and is assigned to the DOPC-rich domains, and cluster ‘1’ has similar C–D and C–H signals and is assigned to domains with an “intermediate” (neither high nor low) concentration of the d62-DPPC and DOPC molecules. An image created using the three cluster labels is presented in Fig. 3c, where the d62-DPPC-rich (purple) and DOPC-rich (dark green) domains are clearly distinguished from each other. The DOPC-rich, d62-DPPC-rich and the intermediate regions occupy 47%, 24% and 28% of the surface area, respectively. Interestingly, in most areas of the image, there is no sharp boundary between the d62-DPPC-rich and DOPC-rich domains. Instead, the transition from the d62-DPPC-rich to the DOPC-rich domains occurs via an “intermediate” region, which has a width of 40–120 nm. It should be noted that such subtle variations in the composition of segregated lipid domains cannot be visualized using conventional optical microscopy such as confocal Raman or infrared imaging due to their diffraction-limited spatial resolution and low sensitivity.
In supported lipid monolayers, the ordering of the acyl chains can be analyzed using the ratio of 2960 and 2850 cm−1 signals (I2960/I2850 ratio). The 2850 cm−1 signal is assigned to the CH2 symmetric while the 2960 cm−1 signal is assigned the Fermi resonance associated with the terminal CH3 symmetric stretch and C–H bending mode. The I2960/I2850 ratio represents intermolecular chain coupling in the DPPC monolayer, whereby a relatively higher I2960/I2850 ratio indicates a more disordered arrangement of the acyl chains.42,62 To visualize the ordering of DOPC molecules in the d62-DPPC:
DOPC monolayer, a TERS image of the I2960/I2850 ratio is presented in Fig. 3d. Notably, the DOPC-rich areas display a relatively low I2960/I2850 ratio (average I2960/I2850 = 0.9), whereas the average I2960/I2850 in the d62-DPPC-rich areas is found to be 9 times higher (average I2960/I2850 = 8.1). This signifies that in the d62-DPPC-rich areas, the DOPC molecules adopt a significantly more disordered arrangement compared to the DOPC-rich areas. We also performed TERS measurements in another area of the sample where a clear segregation of the lipid monolayer into the DPPC-rich, DOPC-rich, and intermediate regions was observed again (like Fig. 2 and 3), although the width of the Intermediate region was found to be ca. 2× larger.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4nr00816b |
This journal is © The Royal Society of Chemistry 2024 |