High-resolution single-particle imaging at 100-200 keV with the Gatan Alpine direct electron detector

Abstract

Developments in direct electron detector technology have played a pivotal role in enabling high-resolution structural studies by cryo-EM at 200 and 300 keV. Yet, theory and recent experiments indicate advantages to imaging at 100 keV, energies for which the current detectors have not been optimized. In this study, we evaluated the Gatan Alpine detector, designed for operation at 100 and 200 keV. Compared to the Gatan K3, Alpine demonstrated a significant DQE improvement at these energies, specifically a ∼ 4-fold improvement at Nyquist at 100 keV. In single-particle cryo-EM experiments, Alpine datasets yielded better than 2 Å resolution reconstructions of apoferritin at 120 and 200 keV on a ThermoFisher Scientific (TFS) Glacios microscope fitted with a non-standard SP-Twin lens. We also achieved a ∼ 3.2 Å resolution reconstruction of a 115 kDa asymmetric protein complex, proving Alpine's effectiveness with complex biological samples. In-depth analysis revealed that Alpine reconstructions are comparable to K3 reconstructions at 200 keV, and remarkably, reconstruction from Alpine at 120 keV on a TFS Glacios surpassed all but the 300 keV data from a TFS Titan Krios with GIF/K3. Additionally, we show Alpine's capability for high-resolution data acquisition and screening on lower-end systems by obtaining ∼ 3 Å resolution reconstructions of apoferritin and aldolase at 100 keV and detailed 2D averages of a 55 kDa sample using a side-entry cryo holder. Overall, we show that Gatan Alpine performs well with the standard 200 keV imaging systems and may potentially capture the benefits of lower accelerating voltages, bringing smaller sized particles within the scope of cryo-EM.

Keywords: 100 keV imaging; Advances in microscope hardware; Direct detectors; Gatan Alpine; LKB1 complex; Single-particle cryo-EM.

Fig. 1.. Detective Quantum Efficiency (DQE) of the Alpine and K3 camera.

DQEs at 100 and 200 keV for the Gatan Alpine direct detector and the Gatan K3 direct detector across frequencies. DQEs were measured in super-resolution counting CDS mode at 7.5 electrons pixel⁻¹ s⁻¹.

Fig. 2.. High-resolution reconstructions of apoferritin from the Alpine detector on a TFS Glacios microscope at 120 and 200 keV.

3D reconstructions of apoferritin colored by the resulting local resolution from (A) 120 keV collection and (B) 200 keV collection. The average Q-score and the resolution associated with that Q-score is reported below each reconstruction. Close-up view of residue Trp93 fit into each density map shows high resolution features.

Fig. 3.. Reconstructions of the LKB1 complex across multiple microscope/detector configurations at a range of accelerating voltages.

3D reconstructions of LKB1 heterotrimeric complex are colored by local resolution. To the right of each structure is a close-up view for a representative α-helix fit into each density map (LKB1 subunit residues 151–169). Overall Q-scores and Q-score derived effective resolution are shown below each structure and the representative helix. In (A) and (B) are reconstructions from TFS Glacios microscope with the Alpine detector at 120 keV and 200 keV, respectively. In (C) and (D) are reconstructions from the K3 detector collections on TFS Glacios at 200 keV and TFS Titan Krios at 300 keV, respectively.

Fig. 4.. Comparison of data collections using per-particle SSNR.

A plot of ppSSNR as a function of spatial frequency for (A) apoferritin and (B) LKB1 collected using the Alpine detector at 120 keV (blue), 200 keV (red), and the K3 detector at 200 keV (green) and 300 keV (yellow). Each curve is truncated to where the FSC from the smallest particle stack reaches 0.143.

Fig. 5.. Reconstructions of apoferritin and aldolase, and 2D class averages of transthyretin using the Alpine detector on a TFS Talos F200C microscope with a side-entry cryo-holder.

3D reconstructions of (A) apoferritin and (B) aldolase are colored by local resolution. To the right of each structure is a close-up view for a representative α-helix fit into each density map (apoferritin residues 13–42, aldolase residues 154–180). Overall Q-scores and Q-score derived effective resolution are shown below each structure and the representative helix. (C) Representative 2D class averages of transthyretin (55 kDa). Secondary structure is clearly visible in the 2D averages.