In science, hyper-specialization is often the path to a successful career. Dedicated researchers can devote decades of their lives to answering a specific question or studying a single protein or organism, chipping away at sharply defined biological questions.

However, this can mean scientists at adjacent institutions, or even just down the hall from one another, may never find time or reason to collaborate, even if they share complementary research interests. In turn, this may lead to a sort of “Blind Men and the Elephant” situation, with experts working on closely related questions publishing in different journals and going to different conferences, and rarely joining forces to answer big-picture questions.

In 2018, CZ Biohub San Francisco launched the Intercampus Research Awards, in an effort to encourage scientists at Bay Area institutions to venture out of their comfort zones and across research silos to build collaborations that would take a more holistic approach toward tackling some of the biggest problems in the life sciences. Despite the geographical proximity of the Biohub’s three partner universities – UC San Francisco, Stanford, and UC Berkeley – researchers at these institutions have sometimes found it difficult to forge seamless scientific collaborations; by requiring that teams include at least one person from each of the three universities, the Intercampus Awards were designed to provide incentives for cross-institutional initiatives.

One team, led by Wah Chiu, John Boothroyd, and William Moerner of Stanford, Carolyn Larabell of UCSF, and Berkeley’s James Sethian, set themselves the task of “imaging complex biological machines in action.” The group – a mix of microbiologists, bioengineers, cell biologists, biophysicists, chemists, and mathematicians – aimed to effectively combine their expertise to take snapshots of the parasite Toxoplasma gondii using some of the most powerful methods of imaging. They aimed to then piece the data from these images together to identify the largely unknown molecular machinery T. gondii uses to invade cells. In a technical tour de force (see video), the team achieved this goal, producing some of the first near-atomic-level resolution images of T. gondii and related parasites.

“That was the first time anyone had seen the parasites in such detail,” says Chiu. “which to me was very rewarding.”

Members of the team credit their success to a “cross-training” approach: an eagerness and willingness to teach and learn from one another meant that everyone on the team ultimately understood each element of the research, no matter how distant from their previous interests.

“The fact that everyone cared about one another’s work, and that we wanted to learn and understand what tools are out there really allowed us to think outside of the box,” says Li-av Segev-Zarko, who was a postdoc in Boothroyd’s lab at the time and is coauthor on several of the group’s studies. “I think that was really the strength of this collaboration.”

In addition, the team credits Chiu for his effective leadership and unwavering focus. “Wah’s leadership was pretty fearless, and very goal-oriented,” Boothroyd says.

From overlapping interests to shared goals

Toxoplasma “is a fascinating parasite,” Segev-Zarko says, “because it has the ability to invade and infect any nucleated cell in any warm-blooded animal—that’s impressive.”

Infection results in toxoplasmosis, a disease that, in humans, may cause mild symptoms in most but can be fatal for some, especially developing fetuses or immunocompromised patients (e.g., those with HIV/AIDS, transplant recipients, etc.). Due to its prevalence worldwide, Toxoplasma presents a global health problem. Understanding how this simple single-cell organism adapts to infect such a wide variety of cells is an avenue toward developing a cure for infection. For her postdoctoral research, Segev-Zarko had been trying to elucidate this unique host–pathogen interaction.

“At the time, we didn’t know much about this really cool invasion machinery that the parasites have that allows them to infect such a broad range of cells and species,” Segev-Zarko says. “I came across these very old electron microscope images of Toxoplasma, and they really struck me conceptually as a good way to tackle questions about this very complex biological machinery. The problem, though, was that existing methods did not have the resolution we needed to gain really new insight. This is why an approach that combines different scales of imaging, including cryogenic electron tomography, with the very latest methods of analysis, was needed.”

Boothroyd suggested they reach out to Chiu, an expert in microscopic imaging techniques, to ask for tips with imaging parasites. To their delight, Chiu and others in his group, including research scientist Stella Sun, were interested in studying the same questions. So when Biohub SF announced the Intercampus Research Awards, Chiu, Boothroyd and their colleagues jumped at the opportunity.

“We’ve been studying this parasite for almost 40 years, and doing a lot of work on its biology, on its population biology, and how it causes disease and how it infects cells, but never with the kind of tools that Wah was proposing we bring to the problem,” Boothroyd says. “So I was thrilled. I felt like a kid in a candy shop.”

The researchers didn’t want to simply use existing techniques and apply them to a new problem, Boothroyd explained. “We also wanted to take a really important problem and use it to build techniques and take them to the next level. The things we wanted to do had never been done before in any system.”

In most biological research, scientists will focus on a single, particular aspect of their organism of interest: how it moves, how it responds to certain stimuli, what it secretes, and so on. But the CZ Biohub intercampus researchers simply wanted to see what they could see. “At the onset, our hypothesis was open, because anything we discovered would be new,” says Chiu. “And that was exciting, because from the technology point of view, we thought we should be able to do it all – it was just a matter of time.”

Revealing the unseen

Cryogenic electron microscopy involves freezing samples in liquid ethane before placing them under an electron microscope and capturing 2D images at different angles that are then computationally combined to render a 3D model. The method is ideal for imaging single molecules, such as proteins and RNAs. But larger structures, such as protein machines and cellular organelles, are a bit trickier because of their complexity and the need to image them in their native cellular environment. Therefore, to create the most accurate and detailed image possible, the team combined different methods of imaging to visualize the region of Toxoplasma that contains the machinery it uses to invade cells.

First, they employed “cryogenic soft X-ray tomography,” a technique that creates moderate-detail imaging of cellular structures by piecing together 2D images of several slices of the sample, similar to how a CAT scan works. For this element, the team enlisted the help of X-ray tomography expert Larabell.

This level of imaging doesn’t provide fine detail, Segev-Zarko says, but it “gave us information at the cellular level that we could use to learn about the motility of the parasite and what the parasite looks like when it’s induced towards invasion versus in a relaxed or dormant stage.”

Taking it a step deeper, the team then imaged the parasite via cryogenic electron tomography, or cryo-ET. At this level, the researchers could see how cytoskeletal elements and secretory organelles that form the parasite’s invasion machinery are built – the proteins that form them, and how they interact with one another.

Since the cytoplasm is rich in proteins and other molecules, the team also took advantage of the Moerner lab’s methods for integrating cryogenic electron tomography and fluorescence microscopy to obtain the spatial and structural organization of the molecular components within the crowded cellular environment in the parasite.

To be confident of what they were seeing, Sethian, a mathematician, took the lead in developing a computer model to identify the structures and organelles showing up in the images. Using artificial intelligence to identify those components ensured the researchers couldn’t accidentally introduce bias or jump to conclusions about what they were seeing.

Additionally, the application of artificial intelligence greatly increased the throughput of analyzing images without bias. Traditionally, annotations had to be performed by hand, but through an interactive process between the expert biologists and computer scientists on the team, refinement and training of the AI program allowed primary annotation to be done automatically, with tools designed to enable subsequent proofreading and correction by the expert biologists.

Through this detailed visualization and subsequent computer modeling of the parasite’s organelles, the researchers were able to reveal a new layer of detail regarding the assembled, complex machinery, and provide new insights and clues about how the machinery, or components of it, move and contribute to contribute to its work and function.

“We were able to use all these tools that existed and push them one step further, in order to really study a biological sample in native conditions from cellular to molecular level. That was pretty amazing,” Segev-Zarko says.

Indeed, that approach and the novel analytical tools they developed can be applied to learning how other complex machines in cells work to further our understanding of biology and mechanisms of human diseases. These collaborations have paved the way to allow these investigators to seek federal funding to extend the in situ imaging technologies at higher resolution, as well as the AI-based image analysis techniques, to answer biomedical questions and guide therapeutic development beyond what is currently possible.

Secret to success

The team’s work is captured in five published papers. Among them are a colorful and detailed 3D diagram of the machinery T. gondii uses to invade cells, which appeared in PNAS in early 2022, and another describing how the parasite deploys this invasion machinery, published in September 2022 in PNAS Nexus.

The collaboration was a special one for a few reasons, the scientists say. First, the timing was right. “When people study what makes big advances in science happen, they often talk about a time when all the pieces come together, and I think we’re at one such moment, where the mix of things like cryo-tomography is coming together with the AI side, which was an important dimension of the work that we were doing,” Boothroyd says.

Segev-Zarko and Boothroyd both note that Chiu’s excellent leadership and ability to keep the team motivated and on task also deserves credit. “Having a leader who’s going to say, ‘Let’s keep going when we hit a stumbling block; let’s figure out a way to work around it,’” Boothroyd says, “is crucial.”

But perhaps most importantly, Boothroyd adds, this was a collaboration that was truly collaborative. “Rather than operating like a relay team and handing off the baton to someone else, the postdocs running the experiments wanted to run beside each other, and learn from each other.”

“From the moment we started talking, the interaction between everyone was so good, and everyone was so on board,” Segev-Zarko says. “It was so fun, and I think it’s reflected in both our work and the development of my own career.”