2008 Workshop on Electron Microscopy

Introduction
The purpose of this day-and-a-half Workshop was to (1) evaluate the extent to which results in cryo-electron microscopy (cryo-EM) of biological specimens still fall quite short of what physics would allow, and (2) identify directions in which research needs to be pursued in order to bring the capabilities of cryo-EM closer to what physics would allow.

The Workshop was organized under the sponsorship of the Howard Hughes Medical Institute (HHMI), and the list of invited participants is given in Appendix A. In addition to HHMI Investigators whose research programs rely heavily on cryo-EM, the participants included PIs (or their representative) of NIH National Centers for Research Resources (NCRR) Facilities that specialize in cryo-EM. Other participants with expertise relevant to the theme of the Workshop, several of whom currently do not work in the field of biological cryo-EM, were also recruited to participate in the Workshop.

Workshop Program
The detailed Program of the Workshop is given in Appendix B. The format consisted of a series of 90-minute sessions devoted to six specific "Propositions" (Topics). An Opening Presenter was selected to provide background in "defense" of each Proposition, with frequent interruptions encouraged in order to ensure that points being reviewed were clearly understood by all. Upon completion of the ~45-minute Opening Presentation, the floor was opened for further discussion, including the introduction of additional information relevant to the Proposition. Copies of slides shown during both the Opening Presentations and the subsequent discussion can be accessed from links provided in the detailed Program.

The final session of the Workshop was then devoted to establishing the major points on which a clear and enthusiastic consensus emerged during the preceding discussions. This final session was actually the most important one, since the ultimate goals of the Workshop were (1) to establish whether a consensus on a number of major points did exist amongst the participants, and (2) to encourage all in the EM community (not just those at the Workshop) to address the very challenging problems on which there is, indeed, consensus that progress is urgently needed. The Workshop was thus intended to stimulate both individual and collaborative research projects that have potential to overcome - at least to some extent - the current barriers that prevent cryo-EM from realizing its full, physically allowed potential.

Workshop Conclusions and Recommendations
Essentially unanimous consensus became apparent on several points. As is summarized in the outline for the final session of the Workshop, these points include:

1. In-focus phase contrast. The first experimental results (from Nagayama's lab in Okazaki) with a Zernike-type (thin carbon film) phase plate show truly impressive contrast for protein molecules as small as 80 kDa. While this unprecedented contrast is achieved under very desirable "in focus" conditions, work still remains to be done on a number of key points. Among the questions that now need to be addressed are:

   1. Is it practical to achieve a resolution with phase plates (as judged by the FSC curve) that is similar to what has been achieved previously with large particles, especially those that have high point-group symmetry?

   2. Does the increased contrast achieved with a phase plate now make it possible to also obtain high-resolution reconstructions for smaller particles than before? In addition, does the increased contrast improve our ability to sort images of particles into separate conformational subsets? In other words, work now needs to be done to establish whether phase plates truly do add value beyond what is provided by defocus-based phase contrast.

   3. Work also is needed to improve the technology of the phase-contrast apertures themselves.

      1. The lifetime of thin (carbon) film phase plates needs to be increased.

      2. A variety of additional design concepts for electrostatic or magnetic phase plates has been described in just the past few years, such as the anamorphotic phase plate and the "Hilbert-contrast" phase plate that is based on the Aharanov-Bohm effect. The further development of these and other new concepts is certainly to be encouraged. At the same time, it now needs to be demonstrated that one or more of these designs can deliver benefits that might be more difficult to achieve with thin-film phase plates.

2. Beam-induced movement. Although there is consensus that beam-induced movement is a major factor that causes experimental data to fall well short of what physics would otherwise allow, it still remains unknown what causes the observed movement. In particular, current data do not allow one to cleanly decide whether the movement is electron optical, due to specimen charging, or simply a physical movement of the specimen itself (again, potentially due to specimen charging, or quite possibly due to mechanical stresses that build up due to radiation damage.) Even so, there is strong support for three directions of research:

   1. The development of new types of specimen substrates (EM grids) that are specifically designed to provide better mechanical support against stress-driven movement, better electrical conductivity, or both.

   2. The extension of spot-scan illumination to spot sizes much smaller than a hundred nanometers. This will require the use of Scherzer-defocus conditions in order to avoid the delocalization of high-resolution information outside the footprint of the illumination spot. If smaller illumination spot sizes prove to be effective, the approach could be further enhanced by the use of a phase plate.

   3. Exploratory experiments to determine whether "ultrafast" electron exposures are able to "outrun" at least some of the beam-induced movement.

      1. Although it is not possible to make a strong case that microsecond exposures should be fast enough to outrun a significant fraction of the beam-induced movement, answering the question experimentally is clearly a high-priority objective.

      2. If the outcome of experiments using microsecond pulses is negative, work should nevertheless continue with even shorter pulses as soon as the technology can be developed to produce shorter pulses with the required number of electrons and with no significant loss in beam quality.

3. Advances in detector performance. Although CCD cameras have had a tremendously beneficial impact on the field of cryo-EM, and EM tomography in particular, the performance of existing detectors still falls well short of being physically perfect. This fact strongly impacts EM data quality. The ideal detector should be noise-free, and in particular the pulse-height spectrum for single-electron counts should approach that of a digital counter. In addition, the modulation transfer function (MTF) should be close to the mathematical limit for a pixilated detector. Finally, the number of pixels should be similar to what is achievable when digitizing photographic film, i.e. in the range of 8kx8k or higher, and the detector itself must be radiation-hardened for use with 300 keV electrons.

   1. There was high enthusiasm for current efforts with CMOS technology, for example the development that is part of the TEAM program in the Materials Science area of electron microscopy. Even this soon-to-be-available technology will still fall well short, however, of producing an ideal, noise-free detector.

   2. Pixel detectors and other design concepts have promise for doing substantially better, and thus there was very high enthusiasm for supporting the substantial investment that is needed to bring one or more of these technologies online.

   3. Large pixel-number detectors that perform close to the noise-free limit, and whose MTF is also close to the mathematical limit, would have truly dramatic impact on everything that the cryo-EM community does (both single-particle EM and EM tomography). Everyone wants such a detector ASAP. The community is thus eager to know how they can support an accelerated development of a next generation of such detectors.

4. Cc and Cs correctors. As was shown in the presentation by Max Haider (CEOS), simultaneous correction of both Cs and Cc has now been achieved, and instruments with this capability are now commercially available.

   1. The use of a Cs corrector with thin (weak-phase) objects only makes sense, of course, if it is used together with a phase-contrast aperture. Furthermore, there is no strong case at this time for the use of a Cc corrector with specimens whose thickness is much less than the mean free path for inelastic scattering.

   2. On the other hand, there is clearly exciting potential for using a Cc corrector for samples in which a substantial fraction of the incident electrons are inelastically scattered.

   3. The next issue, however, is to better understand how to interpret the contrast in images of thick specimens, especially when both inelastically scattered electrons and elastically scattered electrons are perfectly focused by the electron optics.

   4. As a result, it is clear that further theoretical understanding of image interpretation is needed when a Cc corrector is used to image thick specimens, and it is especially important that theory be accompanied by experimental validation.