Broadening access to cryoEM through centralized facilities

Abstract

Cryogenic electron microscopy (cryoEM) uses images of frozen hydrated biological specimens to produce macromolecular structures, opening up previously inaccessible levels of biological organization to high-resolution structural analysis. CryoEM has the potential for broad impact in biomedical research, including basic cell, molecular, and structural biology, and increasingly in drug discovery and vaccine development. Recent advances have led to the expansion of molecular and cellular structure determination at an exponential rate. National and regional centers have emerged to support this growth by increasing the accessibility of cryoEM throughout the biomedical research community. Through cooperation and synergy, these centers form a network of resources that accelerate the adoption of best practices for access and training and establish sustainable workflows to build future research capacity.

Technology development improves cryoEM accessibility

Developments over the past decade have established cryogenic electron microscopy (cryoEM) as a reliable method for determining high-resolution structures of biological molecules [1–3]. CryoEM is uniquely able to provide macromolecular structures in a near-native context by imaging samples in a frozen hydrated environment. Major advances in detector technology, microscope hardware, and computational software, combined with significant incorporation of automation, have resulted in the ability to obtain near-atomic resolution reconstructions within hours to days depending on sample quality [4,5]. Additionally, standardization of cryoEM workflows and data formats have promoted wider adoption of this technology in biomedical research. With the ability to access a resolution regime that can inform structure-based drug discovery and vaccine development, cryoEM has become an important technique for structural studies of medically relevant targets [6–8].

To promote and maintain productive growth in the field, a feedback loop between scientists and technology developers helps to identify and resolve bottlenecks that limit scalability for use by the wider biomedical research community. Biologists adopting cryoEM have experienced bottlenecks in multiple parts of the cryoEM workflow, and solutions to these bottlenecks include innovative improvements in detector and microscope technology that allow for data of higher quality to be collected in less time [9–11]. In parallel, issues with consistency in sample preparation have driven innovations blot free vitrification techniques [12,13]. A constant push to improve image quality has driven development of new sample support films and grids [14,15]. Further, with the ability to acquire information at near-atomic resolution in some cases, questions of non-uniformity and dynamics have driven innovation in data processing software [16–20], and the interoperability of analyzing data using modeling software from other structural biology fields [21,22].

Biological questions can drive innovations in technology but access to the newest technology is necessary to remain at the forefront of utility and applicability for researchers. Herein, we argue that a collaborative approach between researchers and technology developers is best facilitated by the establishment of cryoEM centers, as these promote rapid feedback between the various groups. Vital to this feedback loop are the physical spaces where the work is done. Throughout, we use the example of the Simons Electron Microscopy Center (SEMC)ⁱ located at the New York Structural Biology Center (NYSBC)ⁱⁱ as it provides an ecosystem we find valuable not only for providing access to the best available instrumentation, but also for inspiring research and development in multiple areas of cryoEM. SEMC brings together microscopists, engineers, computational scientists, and biologists, and provides an environment for productive training opportunities.

SEMC – an example of a centralized cryoEM facility

SEMC is a “center of centers” supporting five individual centers, each of which has its own focus while benefiting from sharing the overall physical space, equipment, and a large staff with a wide variety of expertise (Figure 1). The five areas of activity include: (i) NYSBC member institution support providing regular access to and training on dedicated instrumentation for local users; (ii) the National Resource for Automated Molecular Microscopy (NRAMM)ⁱⁱⁱ focusing on technology development for automation and conducting collaborative research to drive and transfer this technology; (iii) the National Center for CryoEM Access and Training (NCCAT)^iv providing instrumentation access and cross-training at no-cost after peer-review of submitted applications to users across the United States; (iv) the National Center for In-situ Tomographic Ultramicroscopy (NCITU)^v dedicated to developing and providing access and cross-training in standard and advanced methods for in-situ cyro electron tomography (cryoET) specimen preparation; and (v) the Simons Machine Learning Center (SMLC)^vi, an initiative to develop machine learning algorithms and software for accelerating all aspects of the cryoEM pipeline.

Figure 1: A centralized control room provides an environment for efficient collaboration.

As an example, the Simons Electron Microscopy Center (SEMC) has two control rooms that provide remote microscope operation. Pictured here is the control room where operators and users collect data on four dedicated National Center for CryoEM Access and Training (NCCAT) Krios microscopes and a Glacios microscope, and the dedicated National Center for In-situ Tomographic Ultramicroscopy (NCITU) Aquilos2 focused ion beam-scanning electron microscope (FIB-SEM). Each microscope station comprises microscope, camera, and data collection computers. There is also open desk space for an embedded engineer and open seating for researchers, staff, or trainees to comfortably sit together for discussions. This arrangement keeps experienced help close to every data collector. Adjacent to the control room, behind blue doors, are microscope rooms and a dehumidified room for sample preparation and autoloader grid handling.

Users and trainees accessing any of these centers become part of the SEMC culture, bringing with them driving biological projects, requests for novel technology, and adding to the overall energy of the enterprise. SEMC also supports a small cohort of internal and visiting postdoctoral fellows and graduate students who conduct their own research projects, providing a regular flow of unique scientific problems and perspectives. Permanent staff scientists and engineers with cryoEM expertise are essential to maintain instrument uptime and provide training to non-expert users. The staff, at the same time, gain a broad perspective of how cryoEM is being utilized by research groups across the nation which can help for planning future allocation of resources.

Creating a network of resources

CryoEM, similar to other structural biology methods, including nuclear magnetic resonance (NMR), X-ray crystallography, and solution X-ray scattering, employs sophisticated instrumentation that requires a high level of expertise for operation and maintenance. The infrastructure requirements to site and support science that makes use of high-end instrumentation is not trivial and pooling resources leads to an economy of scale [23]. The centralization of resources is well established in structural biology with access to regional NMR centers and biologically focused X-ray beamlines at synchrotrons. Before the increased interest in using cryoEM that exploded over the last decade, individual labs and smaller local cores provided access to electron microscopes, but centralized resources of the scale of NMR and X-ray facilities did not exist. This changed dramatically over the last 5 years with significant investment globally to establish large cryoEM centers that offer a mixture of service, collaboration, software, and training to support the biomedical research community (Figure 2) [24,25].

Figure 2: CryoEM resources provide access to instrumentation.

We identified 49 centers, after soliciting information from the cryoEM community, marked with blue dots, that provide general access to cryoEM instrumentation. We acknowledge that this is not a comprehensive list, but it includes many of the world’s currently largest centers: the 4 National Institutes of Health (NIH) centers in the United States, 11 iNEXT consortium^xxv sites in Europe and the United Kingdom, as well as Electron Bio-Imaging Centre (eBIC) and Southern University of Science and Technology (SUSTech) in China. National facilities have also been established in Australia and India, and South America has one cryoEM capable microscope located in Brazil. The geographical distribution of resources is uneven, but large centers can provide access if researcher can ship samples for analysis. A list of cryoEM instrumentation centers with links to websites is available^xxvi.

Trends emerge when examining three of the largest cryoEM facilities in the world: (i) SEMC at NYSBC in the United States is part of a multi-modal research center, also housing an NMR facility and a protein production facility at the same site, and supporting the NYX beamline at the National Synchrotron Light Source II (NSLS-II); (ii) the Electron Bio-Imaging Centre (eBIC)^vii and co-located eBIC for industry facility [26], is located at Diamond Light Source, the United Kingdom’s national synchrotron, and also the site of the new Rosalind Franklin Institute^viii; and (iii) the Southern University of Science and Technology (SUSTech)^ix Cryo-EM Innovation and Research Center combines biomedical research, a strong material science program, and industry outreach in Shenzhen, China. Even though electron microscopes are more easily distributed geographically than, for example, high energy X-ray beamlines that require physical proximity to synchrotron sources that are concentrated in only a handful of locations around the world, these large cryoEM facilities are all co-located with other core utilized services, creating a high degree of strategic flexibility and support through incorporation of different specialties.

In the United States, the national centers established by the National Institutes of Health (NIH) Transformative High-Resolution Cryo Electron Microscopy Program^x similarly maintain this strategic co-location with helpful resources: the NCCAT is housed at SEMC at the NYSBC; the Stanford-SLAC Cryo-EM Center (S²C²)^xi is housed at the Department of Energy (DOE) SLAC National Accelerator Laboratory at Stanford University and co-located with the Stanford Synchrotron Radiation Lightsource; the Pacific Northwest Center for Cryo-EM (PNCC)^xii is a partnership with the DOE Pacific Northwest National Laboratory (PNNL) and Oregon Health and Science University (OHSU); and the National Cryo-Electron Microscopy Facility (NCEF)^xiii is located at the Frederick National Laboratory for Cancer Research and is part of the NIH National Cancer Institute. This network of national centers, along with providing routine access to instrumentation, can contribute to advancement in the field by combining several components for success: state-of-the-art equipment, technical support, and cross-training (see below). Importantly, resources dedicated to general access integrated with resources that allow technology development driven by compelling biological problems add to the sustainability of these centers. This strategy was shown to be particularly productive over the past year, as evidenced by the number of contributions to SARS-CoV-2 research (Box 1). Furthering accessibility to cryoEM for routine use in the biomedical community, these large NIH established centers join a distributed community of smaller centers that provide access to instrumentation and training in cryoEM (Figure 2).

Box 1: CryoEM centers provided rapid structural insights into SARS-CoV-2.

During the COVID-19 global pandemic, the value of access to cryoEM resources to the biomedical research community was clear. As research institutions were ramping down operations or closing, the United States national service centers (National Center for CryoEM Access and Training (NCCAT), Pacific Northwest Center for Cryo-EM (PNCC), and Stanford SLAC CryoEM Center (S2C2)) rapidly refocused efforts towards research into vaccine and therapeutic development (Figure I). More specifically, at SEMC, we made use of our ability to draw on resources from the multiple groups within the center to provide assistance and training at all steps of the cryoEM pipeline including grid preparation, data collection, and processing [44,45]. It was through this collaborative system that structural biologists with minimal or no expertise, along with established cryoEM groups, were enabled to make important contributions to the fight against the pandemic on multiple fronts by accessing these service centers. Many of the research questions required the collection of multiple large data sets, making time for data acquisition the largest bottleneck [46,47]. However, access to Krios microscopes at NCCAT equipped with K3 cameras with fast data collection speeds allowed some groups to obtain high-resolution reconstructions on the same day as their data collection. Rapid access to instrumentation allowed for data to be collected at a speed that made it relevant to addressing the unfolding public health crisis [48].

Box 1, Figure I:

Many structures of SARS-CoV-2 proteins were determined using data collected at National Center for CryoEM Access and Training (NCCAT) in 2020. These include A) the S-protein bound to antibody fragments (EMDB: 22156), B) and C) the S-protein under different experimental conditions (EMDB: 22001, 21999), D) S-protein with a D614G mutation (EMDB: 22826) E) the S-protein bound to the ACE2 receptor (EMDB: 22941), and F) the viral replication-transcription complex (EMDB: 22160).

Whereas the large United States cryoEM centers increase the capacity of the nation to conduct cryoEM research, accessibility of the technique is still highly dependent on access to local resources. Specimen reproducibility remains a major bottleneck for cryoEM projects and fast feedback, both to guide sample preparation and decide if further biochemistry could provide a better sample, is needed to drive projects on a reasonable timeline. Additionally, many samples are not stable enough to travel long distances, and travel costs cannot be ignored. Preliminary and intermediate results produced close to the lab bench are difficult to scale at the level of a national resource. Because the housing requirements for cryoEM instrumentation allow for wide geographic distribution, the growing number of regional resources for data collection and training (Figure 2) are vital to the sustained use of cryoEM in biomedicine, closing a full loop for a cryoEM project, and further increasing the overall capacity for high-resolution data collection.

Centralized resource centers drive the advancement of relevant technology

Large facilities have a greater return from the resources invested because they can increase efficiency and quality through establishment of common standards and protocols [23]. A primary example of how standardization has assisted the adoption of structural biology are X-ray beamlines at synchrotron light sources. These centralized resources have fostered the development and implementation of common tools, and increased capacity of existing resources with high throughput automated remote data collection services [27,28]. A parallel trend is taking place in cryoEM. We highlight some specific examples of such user-driven technology development, below.

Centers can improve efficiency in the form of software tools that increase the capacity of existing hardware. As part of the push to automate data collection, we developed Leginon^xiv, a system designed for automated collection of images for transmission electron microscopes (TEMs), to progress from manually collecting hundreds of images to automatically collecting thousands of images [29]. This and other academically driven efforts, such as Serial EM^xv [30], have encouraged commercial offerings, such as EPU [Thermo Fisher Scientific]^xvi, JADAS [JEOL]^xvii, and Latitude [Gatan]^xviii to further increase the accessibility to automated data collection workflows. An increasingly important companion to data collection is quality control feedback. Addressing this need led to the development of Appion^xix, a pipeline for straightforward processing and analysis of EM images [31]. Appion is integrated with Leginon data acquisition and images and metadata are forwarded to processing packages without user intervention; users can control the processing pipeline from a web-based user interface, as well as access all outputs from the data collection in real-time through user friendly web-based tools. These software developments have helped push forward interoperability of data collected from low-dose multiscale imaging and processing for single-particle experiments, as well as integration of tomography and fiducial-less alignment of tomographs [32]. We have continued to develop this platform to include microED data collection [33], smart data collection based on ice thickness [34], and integration with aberration free hardware correction with beam image-shift to increase data collection throughput [33]. A feedback loop between users actively engaged in the data collection process and software developers ensures advances are focused on the development of tools that will have immediate impact on scientific research.

Adoption of cryoEM by non-expert research groups has been assisted by this high level of automation. Vendor application programming interfaces (APIs), scripting capability, and open-source platforms will continue to allow the community to develop and implement applications, making data collection even more user-friendly and efficient. Already, the ability to collect thousands of images a day, compared to hundreds just a few years ago, effectively increases the availably of microscope time on existing instrumentation by an order of magnitude, with the result that other steps of the cryoEM workflow have become the new bottlenecks in structure determination.

Automation to reduce these new bottlenecks, particularly in sample preparation and data collection feedback, are active areas of development. With the availability of graphics processing units (GPUs), the cost of computing platforms has decreased along with processing times. The shrinking of processing timeframes opens up the ability for on-the-fly processing to give real time feedback for data collection [18,20,31,35]. Depending on sample quality, these approaches allow initial structure determination to occur concurrently with data collection, giving researchers the tools to modify experimental conditions during data collection and thus optimize research outcomes and efficient use of resources. To support these types of developments, centers can facilitate communication between programmers, engineers, and practitioners, and also provide a platform to workshop improvements.

Finally, even the physical tools used for sample handling can improve in a centralized environment. As one specific example from SEMC, a trained X-ray crystallographer interested in making use of cryoEM workflows spent time embedded as a visiting scientist and this led to the creation of cryoEM pucks, a storage system designed to simplify labeling, tracking, and retrieval of grid boxes [36]. The EM puck system is based on crystal pucks widely used by the X-ray crystallographic community for storage and shipping of crystals. As new practitioners enter the field, several of their existing tools may be reused in the cryoEM space, and vice-versa, and this can help lower the barrier to entry.

Maintaining space for innovation while establishing standards

In contrast to offering access to established services, novel technology development involves some risk because successful workflows have yet to be defined and expertise from several fields needs to be combined. Large centers can still play a role in these cases. An emerging application of cryoEM is the ability to image macromolecular assemblies in vitrified cells or tissues by cryoET. Traditionally, structure determination has been largely confined to the analysis of isolated molecules, but cryoET uniquely allows the visualization of macromolecular organization in situ [37]. Although still in its early days of development, cryoET has the potential to have a transformational impact on fundamental biomedical research. Proof of concept methods have shown that cryogenic correlative light and electron microscopy (cryoCLEM), focused ion beam-scanning electron microscopy (FIB-SEM), and cryoET workflows to make targeted “windows” into the cell can open cellular biology to high-resolution cryoEM studies [38–41].

As with other cryoEM modalities, the cost of microscopes and supporting instrumentation for specimen preparation have been barriers to the development and utilization of cryoET. Just a few years behind already established cryoEM centers, specialized collaborative cryoET centers have been created to make inroads to establishing useful workflows. In Europe, eBIC offers user access to FIB-SEM to enable precision milling of frozen-hydrated samples to generate thin lamellae of cellular samples for cryoET [42]. In the United States, the NIH has recently funded a network of cryoET focused service centers analogous to the national cryoEM service centers. This National Network for CryoET^xx is comprised of four centers with the Midwest Center for Cryo-Electron Tomography (MCCET) ^xxi serving as a hub providing administration and high-resolution data collection, and three spoke centers providing optimal sample preparation and screening. The spokes are located at Colorado University Boulder Center for Cryo-ET (CCET)^xxii, NCITU at SEMC/NYSBC, and Stanford-SLAC Cryo-ET Specimen Preparation Service Center (SCSC)^xxiii. A critical part of these centers are their internal development activities to accelerate adoption by researchers by turning new technical advances into robust and reproducible workflows in an accessible way.

Lowering the barriers of adoption through cross-training

As the user base of cryoEM methods grows, the training goals of aspiring practitioners becomes more diverse. In our group, we use the term “cross-training” to distinguish training for researchers who do not aim to establish cryoEM-centered research groups themselves but want to effectively use the technique as part of their existing research program at some level of independence. This is not the only type of trainee we work with at SEMC, but we are particularly invested in this type of training because we are of the opinion that making technology more user friendly and establishing defined and standardized workflows helps researchers at all levels to participate in the field.

Cross-training requires differentiated curricula for different types of practitioners. Further, to create a sustaining training program, we aim to be mindful of the local resources a trainee can access so they are not dependent on center access for all aspects of a cryoEM project. With access to large centers, however, even a user with minimal local instrumentation can gain sufficient experience to qualify as a cryoEM practitioner. Towards the establishment of best practices that can guide new practitioners, the NIH sponsored cryoEM centers have begun an initiative to train to common standards across labs using a “merit badge” program, further described in Box 2. The merit badge framework takes advantage of many already existing training materials and organizes them into modules that support hands-on training, but can be tailored to trainees of many types.

Box 2: Cross-training at national centers: Establishing merit badges.

Training is a core component of any large multiuser facility. Large centers have the ability to support training at many scales including both large and small workshops, one-on-one immersive embedded training, and facility manager training (“training the trainer”). Training should reach diverse audiences; some users aim to be independent at sample preparation or instrumentation operations, while others use data collected by collaborators and require a theoretical rather than hands on competency in the technique. Customized and modular training plans allow us to provide cross-training for users with diverse training goals. Establishing relationships between centers can also help the field to establish best practices.

To expand standards and training efforts, the National Institutes of Health (NIH) sponsored cryoEM centers and Curriculum Development Program sites that are developing open-access instructional material on cryoEM for those with or without a structural biology background have created cryoEM merit badges^x. CryoEM merit badges are proficiency badges awarded to users of any of the centers in three main skill areas: (i) sample preparation, (ii) microscope operations, and (iii) data processing (Figure I). A merit badge certifies a researcher as independent on a particular instrument or in a skill area. After trainees complete a specified set of requirements (Figure I), they may apply for a merit badge. Merit badges are cross-honored at other United States national service centers, minimizing duplication of training effort and ensuring a minimal level of core competency is attained by cryoEM practitioners. Badges are modular, making the design of a training plan simple by combining a subset of badges that can satisfy particular training goals.

Box 2, Figure I:

Merit badges are modular training units organized in three focus areas (top), and each badge is a standalone qualification. Here, we highlight one example of a sample preparation merit badge for use of a Vitrobot plunge freezing instrument (middle). The requirements completed by trainees include testing of base knowledge with a quiz (links to relevant training resources are provided), followed by hands-on training that includes supervised practice and demonstration of competent independent use. Location specific details are separated from a standardized standard operating procedure that can be used at any center.

To foster a community of scientists that can utilize cryoEM, it is insufficient to only provide access or develop novel ideas, we need to act as a bridge to educate a growing number of researchers whose science may be enabled by these methods. For example, over the years we have trained many laboratories to capitalize on the cryoEM workflows that we have developed at SEMC. We find training most effective when trainees spend time immersed in our group while working on their own projects, not just following standard tutorials with test specimens. This requires some additional effort tailored to each trainee but has real payoff for knowledge retention and continued use of the technique in their own lab. A second important component of our training is network building. Many trainees who have spent time at SEMC have subsequently continued in the field, interacted with other facilities around the world, and reported back on their best practices. This feedback has helped our facility refine our standard operating procedures and create a mature learning program that we utilize to train our own operational staff.

An example of how cross-training activities at our center also drives technology adoption can be found in the challenge that many users lack high-end computational resources needed for data management and processing at their home institutions. To address this, we have investigated cloud computing solutions [43] as a possible solution. Based on our experience using on-demand cloud computing resources during large workshops, we have developed STION ^xxiv, a web application to easily deploy cloud infrastructure, manage resources, and process cryoEM data through a web browser from a local computer or laptop. STION can reduce the infrastructure overhead for researchers to process and analyze datasets before making large computational infrastructure purchases, and also provides a platform for cross-training workshops.

Concluding remarks

We support the position that a fundamental motivation for establishing cryoEM centers is to have a catalytic effect in the field by not only scaling services, but also enabling researchers to perform their own research rather than doing it for them. As an example, we have focused our discussion here on SEMC, a center that allows use of state-of-the-art equipment for the collection and analysis of high-resolution data, while also providing technical support and cross-training to establish a community of independent users of cryoEM across the United States. These services are offered at no cost to non-profit institutions, thus eliminating the high-cost barrier usually associated with cryoEM and structural biology in general. To operate an effective facility with multiple instruments that serves a user base with varying skill levels, centers need to be at the forefront of this technology to keep up with software and hardware developments, broaden the applicability, and lower the barriers to access. This is achieved when resources for technology development can be co-located with such user access centers.

Establishing large cryoEM centers near other centers, such as X-ray synchrotron sources, NMR centers, and national laboratories, encourages cryoEM to be integrated with other structural biology modalities and biomedical research fields. Such integration requires some normalization of data and vocabularies that is facilitated by a close physical proximity of groups focused on different techniques. CryoEM data still lacks standardization of data file formats, validation metrics, and strict rules for deposition of data into publicly accessible databases. Consensus on these issues will be facilitated by centralization and advance broader adoption of the technique. Further collaboration with experts from fields outside of traditional structural biology and microscopy is also beneficial. We believe advances in machine learning will be of great importance for multiple aspects of cryoEM from data collection and structure determination to model building and annotation (see Outstanding questions). Towards this end, our SMLC brings together machine learning experts with cryoEM practitioners with the goal of driving these advancements.

Outstanding Questions.

What will be the optimal balance of smaller local or regional versus larger national centers to satisfy the demand for cryoEM instrumentation access?

How can the United States national centers interact with the global community, particularly regions with fewer cryoEM resources to advance the field?

Will distinct resources for sample generation and screening, separate from high-resolution data collection, allow for wider adoption of the technology?

What will the future personnel and staffing needs be to support cryoEM research?

Will machine learning simplify microscope operations enough to further increase accessibility of cryoEM?

Will future breakthrough developments come from hardware and software, or biological applications?

How will commercialization affect the development of the field?

What is the long-term outlook for funding cryoEM centers?

How will we handle the large data storage and transfer bandwidth requirements for cryoEM data?

What data collection and processing formats will the community utilize and standardize?

What requirements should scientific journals demand for cryoEM papers? How should they update their guidelines?

What are the (minimum) requirements for data deposition to ensure Findable, Accessible, Interoperable, and Reusable (FAIR) use of data for other structural biology modalities and developers?

Are current platforms sufficient for deposition and annotation of cryoEM data?

Can the experience with contributing to SARS-CoV-2 response serve to inform the response to future public health emergencies?

In summary, emerging fields like cryoEM benefit from a network of collaborative centers by establishing a synergistic environment that catalytically supports the growth and maturation of the technology. These developments facilitate access, sustain innovation, and ultimately broaden the reach of cryoEM in structural biology, biomedical research, education, and translational research.

Highlights.

• Cryogenic electron microscopy EM (cryoEM) and cryogenic electron tomography (cryoET) are revealing structures of medically relevant biomolecules at an ever-increasing pace, though the technique is still undergoing rapid development, increasing the barrier to entry.

• Large collaborative centers drive the standardization of workflows, lowering the barrier to entry as the co-localization of technologists and biologists drives the development of useful technology to simplify and speed workflows in a way that increases accessibility of the techniques.

• Significant contributions in the structural characterization of Sars-CoV-2 proteins highlight the success of collaborative cryoEM centers.

• Cross-training at cryoEM centers allows practitioners to enter the field at varied levels, increasing the impact of the technique in diverse research areas; such centers also promote a sustaining community by building bridges and providing resources in a collaborative environment.