Every Protein Is Different: Difficult Targets, GPCRs and Drug Discovery

Ann Nguyen:
Hello. I'm Ann Nguyen, Senior Associate Conference Producer with Cambridge Healthtech Institute. Today we have a podcast interview for the Applying Expression Platforms conference, part of 2016's PepTalk: The Protein Science Week, running this January 18-22, in San Diego, California.

We're chatting with Dr. Ian Hunt, head of the Protein Science Group at NIBR, or Novartis Institutes for BioMedical Research, in Cambridge, Massachusetts.

Hi, Ian! Nice to have you joining us today.

Ian Hunt:
Thank you very much.

Ann Nguyen:
Can you describe your path toward providing Protein Sciences support for Novartis? What got you interested in the field, and what's it like working in it now?

Ian Hunt:
Novartis has changed dramatically over the last 10 or 15 years. Initially I worked in a Novartis Research institute in the U.K. for 5 years, in the respiratory group. I was responsible for making proteins for screening, and for biologics purposes. Then in 2005, Novartis completely reorganized how they did research. For the last 10 years, I've been over in Cambridge, in the headquarters. In that 10 years, it's changed dramatically, in terms of the types of targets and how we execute on drug discovery projects.

The types of targets is different, and therefore, the types of proteins we're having to make now has changed dramatically. As a consequence of that, there's been a real need to come up with new technologies and new methods of making some of these proteins. When you ask, "Has it changed?", yes it has changed dramatically. The types of targets are very different. They're actually a lot harder to make. Lots of the low hanging fruits have been done, and we're not interested in. It's all about protein complexes, membrane proteins, and proteins which are a lot more biologically relevant as well.

It's become very hard and that's what's really been driving a lot of the technology development internally by our group here in Cambridge, but by other groups. You can see this also, mirrored in a lot of other pharma and biotech companies, in terms of developing high-throughput protein technologies and different strategies to make some of these difficult proteins.

Ann Nguyen:
Your group at Novartis develops new and enabling technologies in protein science, especially in high-throughput and multi-parallel protein expression strategies. What particular problems are you trying to solve? What factors do you have to consider in your development process?

Ian Hunt:
I touched on this in the previous question. I think the targets are a lot more complex now. When we first started this group 10 years ago, we were very much focused on a lot of kinases, and to support structural biology, X-ray crystallography. We set out a lot of very small-scale, high-throughput profile in, to try and identify their best, most stable construct. This is really taking some of the ideas from some of the structure genomic initiatives, which were being funded by the NIH at that time. Very small-scale, and very high-throughput strategies to critique a number of different constructs for a particular protein.

We spent a lot of time developing these tools and these strategies. Ironically, I think, we found that we weren't using them. It's partly because we actually got very, very good at knowing how to make these proteins. Rather than needing 96, or 200 constructs, we often could be successful in generating the lead construct in about 12, 24 different truncations. That doesn't need very high throughput. You can do that on a bench without any automation, but we did develop a number of different technologies. A lot of other groups have developed similar technologies to enable, to expedite that process.

That approach, typically, is very successful in about 70, 80% of most of the targets. There's a remaining 20% that are really intrinsically very, very difficult targets. You do need very high-throughput technologies to really crack that difficult nut. We developed, in collaboration with some of our colleagues, GNF in San Diego, this high-throughput, very heavily automated system. One where you can profile many, many constructs very quickly. This has been used once or twice by ourselves, but also by other colleagues within the Novartis organization.

You don't need it every week, but it's great to have access to it within an organization the size of Novartis. When on those 2 or 3 projects you're working on in any year, which you're struggling with, you gain access to that technology and really try to use it quite effectively.

In addition to high-throughput technologies, to enable us to critique constructs and truncations, the last 2 or 3 years has been kind of a dramatic explosion in technologies to solve X-ray crystallography of GPCRs.

A recent and exciting development is the work really pioneered by Brian Kobilka at Stanford Medical School, and also Ray Stevens at the Scripps Institute down in La Jolla. Previously, generating crystal structures or G coupled protein receptors has been incredibly difficult, partly driven by the fact they’ve got 7 transmembrane-spanning domains, and very unstable once you have solubilized them away from the cellular environment.

Work by Brian and Ray and others associated with their labs has been really groundbreaking. Now, using some of their protocols, we can now express purified and stabilized GPCRs, and apply those to biophysics and biological therapeutic applications. In addition, X-ray crystallography. It's been an amazing sea change from what was very active in this field 15, 20 years ago when I was postdoc. It was a pipe dream that people would ever get crystal structures of these incredibly important proteins, and now we can!

This is a really – if you look in Nature or Science, at GPCR crystal structures coming out, they're front and center. One of the challenges we have in Novartis is trying to take some of those technologies and seeing if you can apply them into drug discovery context. A lot of this work, you can generate GPCR crystal structures, but it takes a long time. It takes 6 months, 9 months, a year. Our challenge is, can you take some of those technologies and get crystal structures and make proteins that timeline which is commensurate with lead discovery of small molecule and biological lead discovery timelines.

This is a real exciting area, exciting field in protein sciences. A lot of those high-throughput, multi-parallel technologies I discussed earlier, they are clearly applicable to this work. Then you're also taking on board some of the work from Ray, and Brian, and blending that together. It's a really exciting time. It's not just in small molecule, but as I earlier I referred, in biologics, too. If you can stabilize some of these GPCRs in a particular conformation, in a biologically relevant conformation, you can then use those as tools in a far more explicit way than perhaps we've ever done before.

The GPCR targets, they're very well-known drug targets. But one thing we've never been able to do is generate recombinant, stabilized forms locked into particular transitional states. Now, we're beginning to learn how to do that. That really opens up new avenues in which we can really go after allosteric binding pockets, to try and dial in selectivity between different and/or closely related GPCRs. That's something which has always been a big issue, is selectivity. Now with the structures, and now with these protein science tools, we can really start looking at being able to do that. GPCRs and membrane proteins, particularly in protein sciences, is a really exciting field.

Another exciting field -- we're certainly seeing this in Novartis, but other companies are, too -- is we're generating a very large protein complexes. How you can make protein complexes in an in vitro setting and use those in drug discovery. In concert with the work in co-expression and generation of large protein complexes, is some incredible work being done by a number of different labs in Cryo-EM, and looking and now being able to see really great resolutions from these protein complexes.

If somebody asked me, "What are the two exciting areas, in terms of protein sciences?", I'd say membrane proteins and protein complexes. Now we're developing tools in both of these spaces to really attract previously intractable questions. That's been very, very exciting.

Ann Nguyen:
You'll discuss key strategies developed to meet challenges in protein science in drug discovery during your keynote presentation on January 21. What's the main theme you'd like your audience to absorb?

Ian Hunt:
The key theme is every protein is different. Going back to an earlier comment I made about some of the structural genomic consortiums, they really went after whole protein families and then cherry-picked their constructs of proteins, which seem to express well and really went after those proteins to get structural information on them.

Now in pharma and biotech, you're not gifted with the opportunity. You have to work on specific targets that lead to these relevant protein targets. You can't select which ones to work on. You have to work on protein X or protein Y. That's when you can leverage all the high-throughput technologies and some of the newer technologies which have been done in protein complexes in membrane proteins. But each protein's different. You can use all these tools, you can leverage all of these ideas, but every protein behaves differently.

For example, a big topic a few years ago was around gene optimization. Optimizing genes for elevated protein expression. A number of the companies were selling this as a potential benefit. In our experience, and we've done this a lot to gene optimizing constructs ... it didn't always work. In some cases, it really elevates and enhances protein expression. In others, actually, it has a detrimental effect. In other cases, the positioning of different affinity tags, either on the N terminus or the C terminus of the protein, can affect the expression levels and activity profoundly. You can never predict what's going to work. Whether it's going to be an N terminus, a C terminus, an extended tag, a small tag. Different expression systems. Some proteins work beautifully while they're in insect cells. Most of the GPCRs do. Others don't. Different cell lines. Proteins behave differently in different cell lines. You can never predict what's going to work.

Another key message is, I hope people come away with, is you've got to be pragmatic. You need a cohort of different tools and protocols and strategies in place. A toolbox, basically. You need to set up a toolbox, but everyone's going to be different. You just got to be aware that, or develop the skill to recognize when you need particular tools.

That's the key message, is every protein's different. Be pragmatic. Also, really try and leverage some of the great technologies being developed out there, too. There's some wonderful work being done around the world in the generation of protein complexes and membrane proteins. It's an exciting area.

I think another exciting area which – and I'll touch on this in the presentation -- is the cell engineering. Technology like CRISPR, which is so prevalent in many, many different fields and space. Technologies like that can really be used to develop new cell lines. There's a number of different groups actively working on this around the world, trying to manipulate cell lines to become better biofactories for making, not just tool proteins, but also for therapeutics, for biologicals.

We're really on the cusp, I think, now in protein sciences and cell engineering. It's to really go in and really manipulate cell lines become better protein factories. Not just using CRISPR, but also in synthetic genomes. Manufacturing whole new synthetic genomes. Taking all the bad pieces out and adding lots of goodies into the cells. We're really on the cusp of some incredible breakthroughs, I think. They're not there yet, but I think there's going to be some really exciting developments in the next 2 or 3 years in this space. It'll be interesting looking back, in terms of how difficult it was to make proteins.

I think in a few years’ time, we're going to have tools and strategies which are going to be able to enable us to make much more biologically relevant, disease-relevant proteins. I think that's going to be really exciting. That's one of the big issues we've had, I think, in the past is when we're making recombinant proteins in in vitro setting, either in insect cells or in E. coli, the question always which gets leveled at you is "That's great. You've made this protein, but you've made it in E. coli and it is a truncation [non-natural form]. Is it the disease-relevant protein?" We've never known that. What you can potentially begin to look at developing, is making these proteins in a disease-relevant cell line, which is going to be quite interesting.

Ann Nguyen:
Definitely. Thank you so much for your time, Ian. We're looking forward to hearing more insights from you later at the conference.

That was Dr. Ian Hunt of Novartis. He'll be speaking during the Applying Expression Platforms conference at PepTalk in San Diego, taking place January 18-22. To learn more from Dr. Hunt, visit www.chi-peptalk.com for registration info and enter the keycode "Podcast".

This is Ann Nguyen. Thanks for listening.