Caroline Shamu, PhD, lives and breathes team science.

She cut her teeth helping set up one of the first academic centers to apply automation methods used in drug development to discovery science, a pioneering effort in collaborative research.

Today the center, the ICCB-Longwood Screening Facility, is one of a network of research cores that Shamu oversees as associate dean for research cores and technologies at Harvard Medical School (HMS).

She describes research cores as centralized sources of sophisticated technology and expertise with the capacity to conduct experiments that would otherwise be prohibitively expensive for any single lab or department to run. Cores enable individual scientists across the Longwood-HMS community to access highly specialized services, equipment, and staff at some 35 schoolwide centers located throughout HMS, Harvard TH Chan School of Public Health, and Harvard School of Dental Medicine.

In addition to serving as faculty director of the ICCB-Longwood Screening Facility, Shamu holds assistant professorships at HMS and Massachusetts General Hospital. We caught up with her to learn how investigators can benefit from the available–and often invisible–research cores.

What do investigators need to know about HMS research cores?

Well, many may not know all the areas of expertise that research cores cover, or that almost all of them are open to members of the community. The HMS core facilities website lists all the cores and who to contact.  The LMA cores website is another good resource, and the Harvard Catalyst site has information on research cores across all Harvard-affiliated research centers. These are all great places to start.

“Many may not know all the areas of expertise that research cores cover, or that almost all of them are open to members of the community.”

Sometimes it also helps to talk to someone on campus directly, particularly if you’re looking for a new instrument or technique. My peers, the administrators of research cores at other Harvard Schools and Harvard-affiliated academic healthcare centers, are accessible and happy to answer questions. They are helpers like me. They can help find what you need to support your research.

The directors of our research cores–and this is true across the LMA–are all excellent scientists. Often, they’re leaders in their fields, with national and international reputations for the methods or techniques carried out in their core. They are valuable sources of advice when writing a grant, and can provide letters of support to strengthen grant applications by demonstrating that the PI has thought carefully about the methods proposed and will have access to necessary equipment and expertise.

The ICCB-Longwood Screening Facility you’ve directed for more than 20 years pioneered a collaborative approach to harnessing new technology with vast potential for science. What’s its origin story?

The Institute of Chemistry and Cell Biology (ICCB) was established at HMS by Stuart Schreiber and Timothy Mitchison to support the field of chemical genetics. It was one of the first programs to bring technologies and methods used by pharma and biotech for drug discovery into academe for discovery biology.

At the time, pharma was generally focused on very targeted screens. They would identify a likely drug target, purify the relevant protein(s), and run assays to find inhibitors or activators of the function. In the 1990s, the technology to make chemical libraries and automate “high-throughput” screening assays advanced to the point where pharma was screening hundreds of thousands of, or even a million, compounds in their assays. Yet they remained focused on the relatively limited set of mostly enzyme targets for which they had assays. But there were so many more possible targets for small molecules that would perturb biology and possibly lead to the discovery of whole new categories of drugs.

Academic chemists like Stu Schreiber, Matt Shair, and others at Harvard saw this and were quite interested in making new small molecule libraries in their own labs, to push both drug development and chemical synthesis techniques further. Creative experimental biologists like Tim Mitchison and Randy King wanted to develop new high-throughput assays to screen for activities in the libraries.

It was very cool to apply these techniques to discovery biology. Using new assays, you could study pathways that were previously untested to determine whether or not they were druggable. If they were, you could screen broadly for small molecules that perturb those pathways to advance your understanding of the biology.

That was the theory Stuart and Tim pioneered. They developed the ICCB to make it happen.

How has that contributed to advances in academic research?

I think it has changed things in both academe and industry, bringing them closer together in new areas. Industry was using high-throughput screening for very controlled biochemical assays. Introducing that technology to academe enabled collection of more and different types of data, in a different context with different aims.

The approach of many academic screeners was less target-directed and more biology-directed. They want to monitor not one protein at a time, but a whole pathway, to find out what in the pathway might be druggable, and then identify a small molecule that acts in a way suggestive of clinical benefit. We don’t need to know the exact target of the drug in advance; we can identify it retrospectively if warranted.

Pharma didn’t think that way previously. Over time, they’ve become more receptive to conducting these kinds of “black-box” screens, then working backward to identify targets.

That’s been quite interesting. It’s pushed forward the way that drug discovery happens, even in pharma, and it’s allowed laboratory automation to come into academe. Many scientists are taking advantage of laboratory automation to multiplex assays, to do many, many variations of an experiment. Say you want to test 15 different cancer cell lines in 100 different conditions. Automation makes it possible to obtain all that data in a reasonable amount of time.

You’ve been a big advocate of data sharing, and have helped establish guidelines for maintaining the integrity and accessibility of research data in repositories. What are the big challenges ahead in this area?

There’s no point in redoing a screen that someone else has done very well, right? The only way you know what has been done is if the data are shared. Maybe someone else found so many hits in their screen they can’t follow up on them all, and their data could be informative for you.

Sharing screening results also helps characterize the libraries of compounds that we use to screen, if you can compare the results of many different assays used to screen the same libraries. The more information you share, the smarter your analysis is, because you’re learning from the results of others about the functionality of the compounds. You can understand your data and draw conclusions more quickly. There’s so much to be gained. That became clear early on. It’s why data repositories such as PubChem and ChEMBL were established.

These repositories are used by pharma, biotech, and academia alike. They get better as more data is added. But the data is useful only if it’s well-described. To compare your results with someone else’s, you need to know exactly what they did. The screening protocol, assays, reagents, and cell lines used, experimental data points and all variables must be clear and understandable to anyone who wants to use that data.

“The team science is what I’ve enjoyed the most, being part of the community as we figure out how to use new technologies”.

All of that takes time, both to track the information and to compile it for deposition in a data repository, which is now required of most NIH-funded studies. This is a real challenge for individual laboratories.

To facilitate data sharing, what academia is missing is information-management systems in everyday labs to track all those datasets and the key information about them. Many research core facilities have these systems–Laboratory Information Management Systems (LIMS). ICCB started developing its own, called Screensaver, in the mid-2000s. We and other cores also deposit screening data into PubChem for investigators willing to share their data.

So I think what is really needed now is to develop more robust data-management workflows, to ensure our data repositories are the best they can be.

What’s the coolest thing about your job?

The team science is what I’ve enjoyed the most, being part of the community as we figure out how to use new technologies. I started working in the field when we were just beginning to figure out how to implement automation in academic core facilities so we could run high-throughput small-molecule screens. We made progress because we had a community working together to figure it out.

We had investigators coming in to run screens, including postdocs and graduate students who were trying it for the first time. They would collaborate and learn from each other as they developed assays and analyzed their results. We interacted with colleagues in pharma and attended international meetings to learn what people were doing elsewhere and adapt it to our own investigations.

Around 2006 or 2007 we started offering functional genomic screening using RNAi reagents, and a bit later, CRISPR screening. These were both brand new technologies requiring new methods, timelines, techniques, and expertise. They were super fun challenges and incredibly rewarding to figure out. They fed my love for team science and sharing information.

That’s the neat thing about core facilities. You’re always trying to stay at the top, to bring state-of-the-art technology and techniques into your institution. But it’s not you against the world. You’re working with your colleagues in the core, with investigators in the local community, and with scientists internationally to efficiently and effectively adopt new methods and advance research. You’re by definition serving the community.