Fast Grid Preparation for Time-Resolved Cryo-Electron Microscopy

Background
Recent developments in cryo-electron microscopy (cryo-EM) have enabled structural studies of increasingly complex systems at high resolution. With few exceptions, such studies have been limited to biological macromolecules at equilibrium¹ or relatively slow reactions². Many processes in vivo occur on a faster timescale (milliseconds) and there is increasing interest in time-resolved cryo-EM (TrEM) on these timescales³. However, conventional cryo-EM sample preparation by the blotting method is too slow for millisecond TrEM.

The blotting method has other limitations besides poor time resolution. Proteins and protein complexes can suffer from denaturation or preferred orientation on grids⁴. Reducing the exposure time to the air-water interface during sample preparation has been shown to mitigate preferred orientation and protein denaturation^5,6. Thus, fast grid preparation not only enables millisecond TrEM but can also improve grid quality.

Currently, there are three different approaches to automated grid preparation. The first approach uses a pin or capillary that holds a small amount of sample. After establishing contact between the liquid and the grid surface, the sample is 'written' onto the grid^7,8. The sample application process is relatively slow and takes a few seconds. An alternative approach uses controlled droplet generation by a piezo dispenser and self-wicking grids⁹. This allows faster dispense to freeze times, but is still limited by droplet and wicking speed (currently reaching 54 ms). The fastest approach so far is the direct spray approach, in which the sample is atomized in a spray nozzle and the small (~ 10 – 20 µm) and fast (> 5 m/s) droplets spread upon contact with the cryo-EM grid. The sample spray can be generated through different ways such as airblast atomizers, surface acoustic waves or ultrasonic humidifiers^10,11,12,13. In our experience, the ice thickness with the direct spraying approach is greater but direct spraying enables dispense to freeze times < 10 ms.

This protocol describes step-by-step how a time-resolved EM device (TED) equipped with a microfluidic spray nozzle can be used to prepare grids on a fast timescale^14,15. The device has been used to prepare grids with a minimum delay time of 6 ms between sample application and freezing and to rapidly mix and freeze two samples. The design of the TED is based on a previous version¹⁶ and is similar to other spray-based time-resolved cryo-EM devices¹⁷.

First, the four main parts of the TED setup are described. The core of the TED is the liquid handling unit, which is responsible for sample aspiration and dispensing. A pneumatic plunger moves the grid through the spray into the liquid ethane. Generation of the spray is achieved with microfluidic spray nozzles and freezing is done in a liquid ethane container, which are described briefly. Lastly, the additional features to control the grid environment, especially humidity, are highlighted. This is followed by detailed protocols for the operation of the device and for conducting TrEM experiments. Representative results are given for fast grid preparation and a simple TrEM experiment.

Experimental Setup

The liquid handling unit
The liquid handling system of the TED is formed by three syringe drive pumps ('pumps 1 – 3'), each equipped with a rotary valve (Figure 1). A power supply provides pumps 1 – 3 with 24 V DC. Communication with the control software (written in Visual Basic and C++) is via a RS232 interface to pump 1. Commands are distributed through the serial I/O expansion ports from pump 1 to pumps 2-3. Pumps 1-3 are equipped with glass syringes ('syringes 1-3', we use 250 µL/zero dead volume syringes here). Each valve has two positions, 'load' and 'dispense'. The 'load' position is used to aspirate sample into the syringe. A short piece (~ 3 – 4 cm) of 1/16" O.D., 0.01′′ I.D. FEP tubing is connected via ETFE/ETFE flangeless fittings to the 'load' position of valves 1-3. This short piece of tubing reaches into the sample reservoir (typically a 1.5 mL or 0.5 mL plastic tube). The 'dispense' position leads to the spray nozzle. Connection between the 'dispense' outlet and the spray nozzle is made by PE tubing (~ 20-30 cm length, 0.043" O.D., 0.015" I.D.), with a short piece of sleeve tubing (~ 0.5 cm) and ETFE/ETFE flangeless fittings.

The pneumatic plunger
The TED uses a pneumatic plunger to accelerate the grid and move it through the sample spray into the liquid ethane container. Negative pressure tweezers hold the grid, screwed into a home-built holder which is mounted to a dual rod pneumatic cylinder (Figure 2A).

Pressure is supplied from a large nitrogen gas cylinder (size W), equipped with a multistage regulator (0 – 10 bar, 'main pressure'). Flexible reinforced PVC tubing (12 mm O.D.) connects the regulator to a 12-port manifold where pressurized nitrogen is delivered to the nozzle and the pneumatic plunger. Gas flow through the nozzle is constant, regulated directly at the nitrogen cylinder (main pressure). The connection to the nozzle is made with PU tubing (4 mm O.D., 2.5 mm I.D.), a short piece of PE tubing (~ 8 cm length, 0.043" O.D., 0.015" I.D.) and appropriate connectors. Pressure on the pneumatic plunger is controlled through a solenoid valve. PU tubing (4 mm O.D., 2.5 mm I.D.) connects the solenoid valve with a regulator and the pneumatic plunger, to allow a reduced plunge pressure (≤ main pressure). The solenoid valve is computer controlled. A schematic overview of the setup is given in Figure 2B.

Note that with this setup the plunge pressure is always equal or smaller than the spray gas pressure (main pressure). However, the setup can easily be changed by incorporating a second regulator upstream of the spray nozzle to allow higher plunge speeds at low spray gas pressure. High pressures (>> 2 bar) can damage the PDMS spray nozzle.

CAUTION: This is a pressurized system and the 'main pressure' should always be < 7 bar.

Pressures between 0.5 and 2 bar are typically used for the pneumatic plunger and show an approximately linear relation between pressure and speed (at the vertical position of the spray). Plunge speeds are measured with an oscilloscope, connected in line with a slide potentiometer (10 kΩ) and in parallel with a 2 kΩ resistor (Figure 2C). A power supply provides the potentiometer with 9 V DC. While the approximate plunge speed is set prior to the experiment by setting the plunge pressure, the potentiometer gives a precise readout of the speed after the experiment.

Spray nozzles and liquid ethane container
The fabrication and operation of gas-dynamic virtual nozzles for spray-based sample delivery has been described elsewhere in detail¹⁵. As described above, the 'dispense' outlets of valves 1-3 are connected to the liquid inlets of the nozzle (Figure 3A). The pressurized spray gas is connected to the gas inlet of the nozzle. The inlets in the PDMS spray nozzles are such that 0.043" O.D. PE tubing can be used directly without the need for fittings. Our nozzle design contains a 'jet-in-jet' geometry for mixing of two samples, similar to the device described in ref.¹⁸. A schematic of the design is shown in Figure 3B, a microscopic image of a nozzle is shown in Figure 3C. The layout of the microfluidic device requires the use of three syringes to mix two samples. The spray nozzle is typically positioned at 1-1.5 cm distance from the grid (during sample application).

We use liquid ethane as a cryogen, in a liquid ethane/nitrogen container as used for the standard blotting method. Vertical positioning of the liquid ethane cup is achieved with a laboratory lifting platform.

Control of the spray and grid environment
The plunger and spray nozzle are contained within a custom built PMMA (acrylic glass) box with a double door (Figure 4A). High relative humidity inside the box is achieved by an air-humidification system at the back of the TED (Figure 4B). Air is supplied by a pump and fed into a first 10" canister (typically used for under sink water purification). The canister is filled with a low (~ 5-10 cm) level of water and also houses a humidifier unit. Mains power to the humidifier is controlled by a digital humidity/temperature controller and a humidity/temperature sensor located inside the acrylic glass box. The controller is set to turn off the pump when the relative humidity reaches ≥ 90 %. Humidified air from the first canister is pumped through a diffuser, immersed in water in a second 10" canister and then enters the acrylic glass box.

CAUTION: Because the sample is aerosolized in the spray nozzle, hazardous biological or chemical specimen are not suitable as samples.

The run sequence
The Run Script button in the control software initiates the run sequence. This sequence of commands can be pre-defined in a script file and altered through the software. The most important variables are explained here:

Spray speed: The spray speed determines the liquid flowrate used by the syringe pump. The flowrate can be calculated as follows: The syringe pump motors used here have a fixed step size. The full range of the pump is divided into 48,000 steps. The second important factor is the syringe volume. We typically use 250 µL syringes. The spray speed in the control software is set as number of steps/second. A spray speed of 1000 steps/second corresponds to:

Spray volume: The spray volume determines the total volume to be sprayed. Thus, it also determines the duration of the spray. The spray volume in the control software is set as a number of steps. A spray volume of 2000 steps, at a spray speed of 1000 steps/second, leads to a spray duration of 2 s and a total volume of 10.4 µL.

Pre-spray time: This variable defines the time between initiation of the spray and plunge. It is important to choose the delay time such that the spray has sufficient time to stabilize before plunging the grid. Usually, the spray is given 1.5 – 4 s to stabilize before the grid is plunged. The spray is maintained until the grid has moved through. Usually, the liquid flow (and therefore the spray) is stopped 0.5 to 1 s after the grid has been plunged. Using a spray speed of 1000 steps/s and a spray volume of 2000 steps, a typical pre-spray time is 1.5 s, for example.

An exemplary sequence of commands is shown in Figure 5A, the grid position over time is illustrated in Figure 5B.