Cryo-electron microscopy (cryo-EM) has emerged as a prominent technique for the structural determination of protein complexes, supplanting ‘traditional’ X-ray crystallography (Callaway, 2020; Kim et al., 2022). Vitreous ice embedding biological molecules at cryogenic temperatures is indispensable for preserving biological molecules during data collection in EM, which was developed in the 1980s (Dubochet and McDowall, 1981; Weissenberger et al., 2021). The aqueous sample solution should be loaded on a grid-mesh and rapidly frozen in liquid ethane or a mixture of ethane and propane by the plunge-freezing method to achieve vitreous ice in the EM specimen.

The preparation of a grid specimen with thin vitreous ice is a critical bottleneck for collecting high-resolution cryo-EM datasets (Hohle et al., 2022). To overcome this challenge, various strategies have been implemented to improve the grid design for sample preparation, including the use of copper mesh and various metals such as gold or nickel, to produce thinner and more uniform vitreous ice on the grids (Park et al., 2020). In addition, various post-processing methods have been explored, including applying graphene coatings or fabricating grids in the form of nanofluidic chips (Huber et al., 2022).

Classical plunge-freezing technology has long been accompanied by blotting of the excess solution with filter paper (Weissenberger et al., 2021). Semi-automatic equipment for plunge freezing, such as Vitrobot (Thermo Fisher Scientific, USA), EM GP (Leica, Germany), and Cryoplunge 3 (Gatan, USA), has been widely supplied to EM laboratories (Weissenberger et al., 2021). However, obtaining high-quality EM grids for high-resolution data collection remains challenging, relying heavily on trial-and-error practices (Park et al., 2020). One approach to overcome this involves a technique for determining the thickness of the solution on an EM grid by analyzing the diffraction of laser light passing through two-dimensional holes in the grid (Ahn et al., 2020). Automated-plunging machines have been developed to improve the reproducibility of EM grids by eliminating the need for blotting the protein solution (Koning et al., 2022). Notably, the Spotiton team’s Chameleon uses an inkjet-type protein loading device that enables loading only a few nanoliters of the protein sample. In contrast to the conventional blotting method, Chameleon employs self-wicking grids that comprise a grid mesh embedded with copper oxide nanowires and a holey carbon membrane. This design facilitates the absorption of excess solution without the need for external filter papers, owing to the high water-holding capacity of the dense copper nanowires on the grid surface (Dandey et al., 2020; Jain et al., 2012).

The preparation of self-wicking grids involves two sequential procedures: surface oxidation of the copper-rhodium grid and coating of the holy carbon film on the oxidized grid (Wei et al., 2018). During the first step, the oxidation should be restricted to the non-carbon-coated side of the grid to preserve the integrity of the holey carbon film on the opposite side. By using a copper-rhodium sandwich grid with a suitable oxidation solution, selective oxidation of the copper layer can be accomplished. However, fabricating the holey carbon film for the subsequent step is not feasible at the laboratory scale due to the expensive and intricate micropatterning procedures involved (Ermantraut et al., 1998). Moreover, access to purchasing self-wicking grids is limited to only registered users of the Chameleon manufacturer (https://www.quantifoil.com/products/quantifoil-active). Due to the limitations mentioned above, utilization of the self-wicking grid is currently limited despite its potential expandability. In this study, we present a simplified method that uses cellophane film to prepare EM grids with physical properties similar to those of self-wicking grids. Our method involves a simple chemical oxidation procedure to grow copper oxide spikes (COSs) on the square frame of commercially available holey carbon grids, enabling them to absorb excess water from the carbon film. This novel approach offers the potential for the expansion of classical blotting methods in cryo-EM applications.

Preparation of COS grids

In this experimental procedure, several materials are utilized including distilled water (ddH₂O), a 10 M solution of sodium hydroxide (NaOH), a 0.52 M solution of ammonium persulfate (APS), copper Quantifoil 1.2/1.3 grids of 400 mesh size sourced from Electron Microscopy Sciences (EMS, USA), cellophane film specifically the gel drying film from Promega (USA), and slide glass from Paul Marienfeld (Germany).

Grids were prepared according to the following procedure.

(1) Fresh copper-oxidizing solution (nanowire solution) was prepared by 2 parts ddH₂O + 1 part 0.52 M APS + 1 part 10 M NaOH (500 µl ddH₂O + 250 µl 0.52 M APS + 250 µl 10 M NaOH).

(2) The cellophane film was cut into a size range of 20 mm × 20 mm to 30 mm × 30 mm and soaked in water.

(3) A Quantifoil grid was placed with the carbon film side down on a slide glass and then the wiped-off cellophane film was put down as closely as possible. The cellophane film was gently pressed with fingers to attach it over the Quantifoil grid.

(4) Then, 20 µl of copper-oxidizing solution was dropped onto the Quantifoil grid and the reaction was allowed to proceed for 3 min. The reaction process was monitored using a visible-light microscope.

(5) The cellophane film was carefully removed, and a few drops of ddH₂O were added to the grid to facilitate the natural detachment of the Quantifoil grid from the slide glass.

(6) Finally, the grid was placed onto filter paper for drying. The procedure described above is depicted in Fig. 1.

Scanning electron microscopy (SEM) imaging

SEM micrographs of selected grids were acquired using a SIGMA field-emission scanning electron microscope (Carl Zeiss, Germany). The COS grids and standard Quantifoil Cu400 1.2/1.3 grids were affixed to the sample stub with the aid of carbon adhesive tape (3M, USA). Subsequently, the samples were imaged at magnifications of 150×, 4,000×, and 100,000× in the chamber.

Cryogenic transmission electron microscopy (cryo-TEM) imaging

A total of 100 µl of apoferritin from an equine spleen (A3660; Sigma-Aldrich, USA) was diluted with phosphate-buffered saline (PBS) to 500 µl and applied to a Superose 6 increase 10/300 column (GE Healthcare, USA) that had been pre-equilibrated with PBS. Human GroEL protein was expressed and purified as a previously reported procedure (Kim et al., 2021). Purification procedure of the archaeal chaperonin from Methanococcus maripaludis (MmCpn) followed a previously reported procedure (Zhao et al., 2021). An aliquot (2.5 µl) of the peak fraction of each protein (3.4 mg/ml of apoferritin, 4.5 mg/ml of GroEL, and 1.0 mg/ml of MmCpn) was applied to glow-discharged (15 mA, 60 s) Quantifoil 1.2/1.3 Cu 300 mesh carbon film grids or COS grids. Sample-applied grids were blotted for 2 s with blot force 0 at 15°C and 100% humidity and vitrified using Vitrobot Mark IV in CMCI at SNU (Center for Macromolecular and Cell Imaging at Seoul National University).

Cryo-TEM imaging was performed using 200 kV Talos Glacios (Thermo Fisher Scientific, CMCI at SNU) equipped with a Falcon 4 direct electron camera and EPU software (Thermo Fisher Scientific). The micrographs were obtained at magnifications of 115×, 510×, 6,700×, and 92,000× (Thermo Fisher Scientific).

Preparation of the COS grids from the commercially available EM grid

Wei et al. (2018) previously introduced a copper-oxidizing solution, consisting of sodium hydroxide and APS, to selectively oxidize copper surfaces on copper-rhodium sandwich grids. The solution selectively oxidizes copper, leaving the rhodium surface intact. However, due to the specialized equipment required for fabricating holey carbon membranes, this approach is not readily implementable in most individual laboratories.

We attempted to produce nanowires on the copper surface of Quantifoil grids coated with a carbon membrane using a copper-oxidizing solution. However, we encountered an issue where the carbon membrane dissolved before the copper oxidation process was completed. To overcome this issue, we placed a cellophane film between the copper-oxidizing solution and the copper side of the grid (Fig. 1A). The cellophane film contained small pores, which still allowed the oxidizing solution to permeate, although the rate of permeation was significantly reduced. This enabled easy and precise control of the oxidation rate by monitoring the reaction through a light microscope and detaching the cellophane film at an appropriate time point (Fig. 1B).

Initially, we hydrated the cellophane film with distilled water and removed the excess water with a paper towel to achieve the desired moisture level. We then applied the cellophane film onto the EM grid, ensuring that it adhered closely (Fig. 1C). Only the hydrated cellophane film was found to be easily spread on the grid compared to the dry film. After loading the copper-oxidizing solution onto the cellophane film, we observed air chambers or air bubbles only in the non-copper mesh regions of the grids under visible light microscopy, possibly due to the presence of air trapped between the layers of the cellophane film and the carbon coatings (Fig. 1D). However, these air chambers began to contract from the edges synchronously within 2-3 min, indicating the permeation of the oxidizing solution through the carbon membranes. After 5 min of incubation, almost all carbon coats on the grid surface were dissolved. To selectively oxidize the copper in the grid while minimizing damage to the carbon grid, we optimized the incubation time to 3 min, consistent with previous studies on self-wicking nanowire grids (Wei et al., 2018). Subsequently, we washed the grid with distilled water to remove the oxidation solution and carefully removed the cellophane film from the grids. To detach the grid from the slide glass, a few drops of water were dropped on the grid to detach the grid naturally, and we floated the grid on the water in a beaker. The grid that was detached was observed under an optical microscope and confirmed to possess a coarse copper mesh structure. Additionally, the carbon film appeared to be undamaged and intact (Fig. 1E). Finally, the grids were dried before use, and we performed plasma cleaning with the copper side facing up.

Formation of COSs on the surface of the COS grids

To examine the oxidized surface of the EM grids prepared by our method, we employed SEM (Fig. 2). We observed numerous spikes evenly distributed along the copper mesh area, while the carbon coat area displayed clear two-dimensional hole patterns. These observations indicated that the solution achieved close contact with the copper surface of Quantifoil grids and that the cellophane film selectively blocked the immediate contact of the oxidizing solution from the carbon coat (Fig. 2E).

Spikes were formed on the dense hairy lawn region in the COS grids (Fig. 2F), which is distinct from the smooth surface of the untreated grids (Fig. 2C). The contact of the oxidizing solution on the copper mesh surface resulted in the formation of nanosized COSs. Notably, the copper mesh of the Quantifoil grid had COSs formed not only in the flat area in direct contact with the cellophane film but also on the edge exposed to the holey carbon foil (Fig. 2G). The copper spikes had an elongated rod shape, with a length of approximately 500 nm and a width of 10 nm (Fig. 2H). The morphology of the copper spikes differed from the nanowires of the self-wicking grids. To differentiate between the grids produced in this study and the self-wicking grids, we term them the COS grid hereafter.

The COS grid exhibited sparse ~500 nm-long spikes on top of the short hairy lawn of the copper oxide on the copper mesh surface (Fig. 2F). However, the self-wicking grids had long and dense copper oxide wires covering the copper side surface of the copper/rhodium mesh. Thus, the density of the copper spikes on the COS grid appeared to be lower than that of the self-wicking grid in terms of the long spikes.

Application of the COS grids to the filter paper blotting method using Vitrobot

To assess the efficacy of the COS grid, we employed conventional filter paper blotting methods. A few microliters of 1 mg/ml apoferritin sample were loaded onto the copper-oxide surface of the EM grid using the semi-automatic blotting machine Vitrobot.

Notably, the COS grid exhibited a constant ice thickness throughout the entire grid surface, with the region adjacent to the entire grid being sufficiently thinned and cleaned to enable EM imaging at high resolution (Fig. 3A). From the other images in Fig. 3, it can be observed that the holes are not empty but filled with a thin vitrified water membrane.

Significantly, the COS grid displayed a larger imageable area with uniform vitrified ice formed throughout the entire grid surface in low-magnification cryo-TEM micrographs (Fig. 3A). In contrast, a conventional Quantifoil grid prepared by the same blotting method showed a reduced imageable area due to thickly frozen ice (Supplementary Fig. S1A). At higher magnifications, the overall ice thickness appeared thin both near the edge and the central area of the copper mesh square (Figs. 3B and 3C), and uniformly thin in the hole (Fig. 3D). In contrast, the conventional Quantifoil grid showed thicker ice near the edge (Supplementary Fig. S1B). For single particle imaging at high magnification (98,000×), the apoferritin particles were clearly visible with better contrast on the COS grid (Fig. 3E) than on the conventional Quantifoil grid (Supplementary Fig. S1C). Therefore, we conclude that COSs enhance the homogeneity of ice thickness during vitrification using the conventional blotting method.

To examine whether COS grids consistently generate good ice conditions, we performed cryo-EM imaging using two distinct model proteins: the archaeal chaperonin from M. maripaludis and human GroEL. As seen in Supplementary Fig. S2, the favorable ice conditions were obtained with COS grids for both proteins, which seemed suitable for further structural determinations.

The sample preparation step is a crucial bottleneck in obtaining high-resolution cryo-EM datasets. Despite several drawbacks associated with the filter-paper blotting method, including poor reproducibility, the blotting method with a semiautomatic plunger remains prevalent in the cryo-EM field. Although the new machines may provide more effective grid preparation, they are often unaffordable for individual laboratories, making the fabrication of preexisting grids a more practical option for obtaining better datasets.

In this study, we presented a simple method for preparing COS grids from commercially available Quantifoil grids. SEM images showed the formation of fine fiber lawns of long COSs throughout the copper surface of COS grids. The COS grids were applied to the filter-paper blotting method using the semi-automatic blotting machine in this study. We demonstrated that the COS grids produced better image quality with apoferritin, including ice thickness and data-collectible areas over the holes, with higher reproducibility. Furthermore, the COS grids consistently produced high-quality images when used with two different model proteins. While we cannot definitively claim that the image quality obtained with COS grids surpasses that of conventional Quantifoil grids, it is evident that COS grids offer an alternative option for challenging proteins.

The improved performance of the filter-paper blotting method using COS grids can be attributed to the following factors. In the conventional Quantifoil grid, the contact between the filter paper surface and the carbon coating over the holes is often uneven, leading to poor reproducibility (Armstrong et al., 2020). This is due to the large fluctuations in the filter paper surfaces, which are comparatively larger than the spacing between the holes in the carbon coating. As a result, only a portion of the carbon coating area comes into direct contact with the filter paper, leading to reduced absorption in other areas. In contrast, the copper-oxide spikes on the surface of the COS grid can promote absorption from the carbon grid. These nanoscale spikes on the oxidized copper mesh act as a bridge between the holey carbon membrane and filter paper, resulting in a stable and consistent ice thickness on the grids during vitrification process (Fig. 4). Although we did not use a robotic Spotiton (Chameleon), which requires a self-wicking grid, we expected that our COS grids would also be applicable in Spotition, similar to standardized self-wicking grids.

In conclusion, we have developed a novel method for fabricating COS grids using commercially available Quantifoil grids, which can be utilized in cryo-EM applications. Additionally, we have discovered that the copper oxide layers effectively facilitate the transfer of solutions from the carbon membrane to the filter paper in conventional blotting machines. We firmly believe that this innovative grid processing technique holds the potential to establish a distinctive and effective vitrification condition for protein solutions, while also offering a straightforward and easily implementable approach. Furthermore, the versatility of COS grids could be extended by integrating them with diverse techniques, such as microfluidic spraying devices, to prepare EM grids suitable for high-resolution data collection.

This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korean Government (2019M3E5D606387122, 2020R1A5A1018081, 2021M3A9I4021220). Purified GroEL and MmCpn proteins were kindly provided by Mingyu Jung and Sojeong Kim from the Lab of Molecular Imaging, SNU.

S.H.R. and N.C.H. designed the research. D.L., H.L., and J.L. conceived and performed experiments. S.H.R. and N.C.H. provided expertise and feedback. D.L., H.L., S.H.R., and N.C.H. wrote the manuscript. S.H.R. and N.C.H. secured funding.

The authors have no potential conflicts of interest to disclose.