The orientation of graphene has been demonstrated previously as a crucial parameter for the graphene-cell interaction, the extraction of phospholipids and release of intracellular materials17,24. Considering this phenomenon, initially we chose to compare the lipid absorption capacity of two types of differently oriented graphene coatings: a graphene monolayer parallel to the coated surface (Fig. 1a) and multiple layers of graphene stacked perpendicularly to the coated surface (Fig. 1b).
Time lapse measurements of contact angle of lipid droplets diluted in 5% dimethyl sulfoxide on (a) surface coated with monolayer graphene deposited parallel to the coated surface and (b) surface coated with vertically oriented graphene, perpendicular to the coated surface. 5% of dimethyl sulfoxide was used as a negative control.
A droplet of lipid from biological membranes was deposited on both types of coated surfaces, and the contact angle between the surface and the droplet was measured over time. As shown in Fig. 1a, the contact angle of the lipid droplet deposited on a flat monolayer graphene coating remained constant throughout the experiment, indicating that a single layer of horizontal graphene cannot absorb significant amounts of lipid. By contrast, the contact angle of the lipid droplet deposited on vertically aligned graphene coating decreased progressively from 35º to less than 15º, within one hour (Fig. 1b). This result strongly suggested that a vertically oriented graphene coating, composed of many parallel layers of graphene, has excellent storage capacity for lipids derived from biological membranes.
Further, we speculated that this ability to absorb large amounts of lipids depends on intercalating lipid molecules between multiple layers of graphene in the coating. To examine this assumption, we next analyzed the interaction between multilayer graphene samples and biological lipids using X-ray diffraction (XRD) and Fourier-transform infrared (FT-IR) spectroscopy (Fig. 2). Figure 2a shows X-ray diffractograms obtained from multilayer graphene flakes alone and multilayer graphene flakes pre-soaked in either lipids or water. The main diffraction peak in the XRD pattern for untreated graphene was observed at 28.1°, which is characteristic of the (002) plane in graphene and graphite, as seen from comparison with the structural database entry JCPDS 75-1621. Bragg’s Law was applied to the (002) reflection pattern to calculate interplanar spacing distance, i.e., the distance between graphene layers, and was estimated at 0.32 nm for the untreated graphene. For the lipid-soaked sample, an additional peak was detected at 21.3°. The new peak corresponds to an interplanar spacing of 0.42 nm, a 31% increase from the layer distance in pure graphene.
(a) X-ray diffractograms of graphene flakes alone (Gr), graphene flakes soaked in Milli-Q water (Gr + water) and graphene flakes soaked in lipids (Gr + lipid). (b) FTIR spectra of graphene flakes alone (Gr) and graphene flakes soaked in lipids (Gr + lipid).
This finding suggests that some fraction of the lipids is absorbed by graphene and inserted between the graphene layers. The unchanged peak position of the water-soaked sample in Fig. 2a suggests that the peak shift observed in lipid-soaked samples is not caused merely by a change in packing characteristics of the graphene flakes due to wetting with lipids. The XRD results indicated extra spacing between graphene layers upon interaction with lipids but did not actually prove the lipid absorption. To provide more direct evidence for lipid binding to multilayer graphene flakes, the flakes in powder form were mixed with biological lipids, washed and characterized using FT-IR spectroscopy (Fig. 2b). The FT-IR spectra of graphene mixed with lipids showed typical peaks related to biological lipids, suggesting the efficient absorption of lipids. Specifically, the characteristic peaks related to triglycerides, which are a major component of biological lipids, were observed in the recorded spectra: at 2926 cm−1 (C–H stretching (asymmetry)), 2854 cm−1 (C–H stretching (symmetry)), 1746 cm-1 (C=O stretching), 1465 cm−1 (C–H bending (scissoring)), 1165 cm−1 (C–O stretching and C–H bending), and 722 cm−1 (C–H bending)30. Taking together the findings from contact angle measurements (Fig. 1), XRD and FT-IR analysis (Fig. 2), we concluded that coatings based on multilayer graphene can soak up significant amounts of biological lipids. Next, we attempted to reproduce such coatings on the surface of spherical metallic nanoparticles.
Iron oxide nanoparticles (average diameter of 25 nm) were axially coated with graphene, using the method of plasma-enhanced chemical vapor deposition (PE-CVD) (Fig. 3a). As a result, the surface of the iron oxide nanoparticles was completely covered with axially oriented graphene nano flakes, with height ranging from 60–100 nm (Fig. 3b). As shown in Fig. 3c, Raman spectroscopic examination of the coated nanoparticles revealed graphitic structure of the coatings, with relatively good quality. All Raman peaks specific for graphene, D, G and 2d, were detected31. The D peak (~ 1350 cm−1) being higher than the G peak (~ 1590 cm−1) is indicative of axial orientation and high density of flake boundaries. The low intensity of the 2D peak (~ 2650 cm−1) (intensity decreases with increasing number of graphene layers) indicated that the samples are a mixture of monolayer and multilayer graphene. Combining high magnification SEM imaging and Raman data, we concluded that the iron oxide nanoparticles were successfully axially coated with PE-CVD graphene, resulting in a complete and homogenous coating, with a large proportion of exposed sharp edges, containing on average < 10 layers of graphene.
Characterization of iron oxide nanoparticles axially coated with PE-CVD graphene. (a) Overview of the coating procedure and lipid extraction strategy (prepared using Microsoft PowerPoint). Iron oxide nanoparticles were axially coated with PE-CVD graphene. Graphene spikes, in interaction with yeast cell, are expected to provoke extraction of lipids. In this schematic overview all constituents are not drawn to scale. (b) SEM images showing orientation and density of axial graphene coatings. Red lines indicate different levels of magnification. (c) Raman spectra (laser wavelength of 638 nm) of axial graphene coatings on iron oxide nanoparticles. The three main peaks characteristic for Raman spectra of graphene are marked: D, G, and 2D.
Next, we asked whether these coated nanoparticles can effectively compromise the integrity of yeast cells to release intracellular lipids. Computational simulations of graphene interaction with cellular membranes describe the insertion of edges of free graphene nanosheets into the membrane, driven by strong van der Waals attractions and hydrophobic interactions with membrane lipids17,32. According to this mechanism, the graphene nanosheet cuts through the lipid membrane, with the edge of the nanosheet sinking into the lipid bilayer. Thereafter, the lipid extraction mode commences, where the nanosheet extracts the phospholipid molecules from the lipid bilayer onto its surface. This massive draining of lipids from the membrane onto the graphene sheet results in membrane deformation, and ultimately in loss of cell membrane integrity17,32,33,34. Following this extraction of phospholipids, strong hydrophobic interactions between the graphene and the lipids have been speculated to play an important role through nanoscale ‘dewetting’. This strong hydrophobic interaction can be compared to the hydrophobic collapse of many biomolecular self-assemblies, such as cell membrane formation and protein folding35,36. Hence, it has been suggested that nanoscale dewetting can provide the driving forces required for complete cellular collapse, leading to release of all intracellular content37. However, all available mechanistic simulations and experimental data showing this mechanism is based on the interaction of graphene with bacterial cell membranes. To the best of our knowledge, there are no reports showing the interaction of graphene nanosheets with yeast cells.
We therefore proceeded to test the ability of graphene-coated nanoparticles to extract lipids from yeast cells. In order to do so, fractions of Y. lipolytica culture grown for 5 days were treated with bare nanoparticles and graphene coated nanoparticles for 2 h with continuous agitation (1000 rpm). In order to assess if the release of lipid would be compatible with continuous growth in a fermentation process, no other treatment was applied. After 2 h of treatment with bare nanoparticles and graphene coated nanoparticles, yeast cells were examined by brightfield and fluorescent microscopy. The latter was combined with BODIPY staining (specific to lipid drops) that allows for visualization of lipids (including triglycerides), stained bright green (Fig. 4). All cells in the untreated control group were observed to contain large intracellular lipid droplets (Fig. 4a). By contrast, in all samples where cells were mixed with graphene-coated nanoparticles, extensive loss of intracellular lipids was observed. In some cases, the entire large lipid droplets moved out of the damaged cells (Fig. 4b). In other cases, lipids were more gradually drained from damaged cells, while interacting with graphene-coated nanoparticles.
Bright field (left) and fluorescence (right) microscopic images of (a) untreated Y. lipolytica cells and (b) Y. lipolytica cells treated with graphene-coated nanoparticles. Samples were stained with BODIPY, which provides green fluorescence upon binding to lipids. The orange circles in the graphene-treated group show yeast cells with a complete release of intracellular lipid drops; green arrows are showing released lipid drops in the extracellular space. (c) Dropwise release of lipid droplets from yeast cells. The green circle is showing dropwise release (green arrow) and extracellular accumulation of lipid droplets. Zoomed-in images (bottom panels) show the interaction of graphene-coated nanoparticles and lipid drops, and their stepwise release from yeast cells. The green arrows show released lipid droplets and orange arrows indicate graphene-coated nanoparticles. (d) Quantification of intracellular and extracellular lipids, stained with BODIPY, in control and graphene-treated samples. Cumulative data obtained by examination of 200 untreated and 200 treated cells.
A direct interaction between graphene-coated nanoparticles and smaller lipid droplets can be seen in Fig. 4c. All fluorescent microscopic images were further processed for quantification of total intracellular vs extracellular lipids (Fig. 4d). In the untreated group, only 5.9% of total lipid was found in the extracellular environment, due to the presence of some dead and disintegrated cells in the 5-day yeast culture. In samples treated with graphene-coated nanoparticles, about 50% of total lipid was detected outside the cells, suggesting a dramatic release of lipids from the Y. lipolytica strain. Finally, we asked whether the observed lipid release can be correlated exclusively to the presence of the axial graphene coating, or whether the iron oxide nanoparticles themselves have some capacity to damage yeast cells and release the intracellular lipids. To address this, cells treated with bare (non-coated) iron oxide nanoparticles were examined, and their lipid content was fully comparable to that of untreated cells, with only about 4.3% of total lipid present in the extracellular space. Thus, we concluded that the graphene coating is the key factor for release of lipids from cells. To more closely examine the morphology of Y. lipolytica cells exposed to graphene-coated nanoparticles, we further examined them by SEM (Fig. 5). The morphology of cells treated with bare iron oxide nanoparticles (Fig. 5b) was indistinguishable from cells that had received no treatment (Fig. 5a). This was consistent with fluorescent microscopy data (Fig. 4).
SEM images of Y. lipolytica cells (a) untreated, (b) treated with bare iron oxide nanoparticles, (c) treated with graphene-coated iron oxide nanoparticles. (d) Aggregation of Y. lipolytica cells on a glue-like matrix of extracted lipids.
By contrast, cells treated with graphene-coated nanoparticles exhibited various morphological defects (Fig. 5c), indicating extensive membrane damage and, in some cases, cell collapse. In the sample treated with graphene-coated nanoparticles, many Y. lipolytica cells were clustered, apparently held together by a lipid matrix presumably formed by lipids released from the damaged cells (Fig. 5d). Finally, the effect of graphene coated nanoparticles on growth pattern of Y. lipolytica was examined. As shown in Fig. 6, graphene-coated nanoparticles did not inhibit exponential growth of yeast. A weak concentration-dependent inhibition was observed at the entry of stationary phase, around OD600 = 3. However, the impact of nanoparticles on the final OD was minimal. This finding is very promising, since it indicates that lipid release from yeast cells can be achieved continuously without inhibiting growth, and therefore without compromising the metabolic yield of lipids. This technology could provide a significant improvement over existing methods to extract biofuels from microbial cell factories, thereby reducing energy consumption, extraction time and the use of hazardous chemicals. This encouraging finding is of course preliminary, and further optimization in scaled-up production setup should be performed in follow-up studies, to achieve an optimal balance between lipid production and lipid release.
Growth of Y. lipolytica with and without presence of graphene- coated nanoparticles. Cells were incubated at 30 °C and 200 rpm and growth was monitored for 96 h at the interval of 30 min. Each curve and its respective shadow represent the mean values and standard deviation of biological triplicate experiments. G-NP denotes the standard working concentration of graphene-coated nanoparticles. G-NP 1/2, G-NP 1/4 and G-NP 1/8 are respective dilutions of the standard working concentration (G-NP).
