Ten protein engineering papers that changed the world

By Paul Ko Ferrigno

For a scientist, describing what you do at parties is fraught with danger… Of all the newsworthy science, though – Frankenstein crops, nano-tech’s grey goo, God particle physics – we protein engineers have actually got some pretty interesting things to say. Here are 10 papers that changed the world, something that we can brag about (even if we haven’t co-authored these particular papers).

10. Antibody Active Sites and Immunoglobulin Molecules, by S. J. Singer and Russell F. Doolittle. Science 153 13-25 (1966) (link is external). Written in the days when people still thought there might be as many as 100,000 genes encoding immunoglobulins, and when journals like Science would publish papers with more than 5 pages, this paper established that the “active sites” of antibodies (note the analogy with enzymes) appeared to be spread across one heavy chain and one light chain. There was no precedent for this from enzymology, where active sites of enzymes were all found to be formed within a single polypeptide chain. The authors also found that the heavy and light chains of antibodies appeared to share considerable chemical similarity, anticipating the finding that both immunoglobulin chains are made up of highly conceived immunoglobulin domains that have probably arisen through divergent evolution from a single precursor. Singer and Doolittle also commented on the finding that tyrosine residues appeared to be present in the vast majority of antibody active sites – foreshadowing two of the other papers discussed below. And all of these biological insights came through the application of chemistry…  

9.  Humanization of an anti-p185HER2 antibody for human cancer therapy, by P Carter, L Presta, CM Gorman, JBB Ridgway, D Henner, WLT Wong, AM Rowland, C Kotts, ME Carver, & HM Shepard. PNAS 89 4285-4289 (1992) (link is external). Others had already shown that it was possible to humanise a mouse antibody – a key step in enabling long term treatment without the patient mounting an immune response to the therapy. Carter et al. were the first to describe a fast and efficient strategy though, using molecular modelling to inform the synthesis of mutagenic cassettes targeting multiple nucleotides in one step, leading to 24 simultaneous amino acid changes in the light chain and 32 to the heavy chain. This might seem trivial in the gene synthesis era, but this paper laid the groundwork for the treatment of more than 1.2 million breast cancer patients worldwide by October 2012 (Roche press release (link is external)). World-changing.  

8 and 7. Not one but two papers here, sorry. The first describes a library, and the second the success. Iterative Optimization of High-Affinity Protease Inhibitors Using Phage Display. 1. Plasmin, by William Markland, Arthur Charles Ley, Stanley W. Lee, and Robert Charles Ladner; Biochemistry 35, 8045-8057 8045 (1996) (link is external); and Iterative Optimization of High-Affinity Protease Inhibitors Using Phage Display. 2. Plasma Kallikrein and Thrombin, by William Markland, Arthur Charles Ley, and Robert Charles Ladner. Biochemistry (link is external), 35, 8058-8067 (1996).  Patients undergoing excessive blood loss during or after surgery had been treated with bovine trypsin inhibitor since at least the 1960s, and there was a clear need for a more human-like protease inhibitor. The goal was to use human homologues of the tried and tested Kunitz domain family of protease inhibitors as a starting point, and to drive their specificity towards a target protease in the clotting cascade. Again, others had tried, but the importance here was that in making their library, the Ladner group used structural information to guide their library design, and an iterative library strategy to guide their screens. The end result was a plasmin inhibitor with 87 pM Ki – not bad at all! In the second paper, the same starting library and similar iterative approaches were used to identify a highly specific kallikrein inhibitor, EPI-K503, with a Ki of just 40 pM. The affinity for kallikrein was at least 100 times better than the therapeutically useful bovine pancreatic trypsin inhibitor (BPTI, aprotinin). EPI-K503 became DX-88, which became ecallantide and finally hit the market in 2008 as Kalbitor. At less than 4000 doses a year, not a blockbuster, but a development program that led the way in phage-display driven protein drug discovery.

6. Constrained peptides as binding entities. Robert C Ladner. Trends in Biotechnology 13 426-430 (1995) (link is external). A rich source of inspiration for myself and others over the years…  Some of the concepts are now questioned (is Kd really driven by entropic cost of binding?) but the concise exposition of ideas gives me food for thought whenever I take another look.  

5 and 4. Two papers again. Molecular Recognition by a Binary Code. Frederic A. Fellouse, Bing Li, Deanne M. Compaan, Andrew A. Peden Sarah G. Hymowitz and Sachdev S. Sidhu. J. Mol. Biol. 348, 1153–1162 (2005) (link is external); and: Tyrosine Plays a Dominant Functional Role in the Paratope of a Synthetic Antibody Derived from a Four Amino Acid Code, by Frederic A. Fellouse, Pierre A. Barthelemy, Robert F. Kelley and Sachdev S. Sidhu. J. Mol. Biol. 357, 100–114, (2006) (link is external). The authors set out to ask whether there was a functional basis for the apparent over-representation of tyrosine and serine residues in antibody binding sites. They had previously addressed this question by selecting VEGF binders from a Fab library whose CDRs were composed of just 4 amino acids: Ser, Tyr, Ala and Asp. The selected binders were dominated by Tyr, with little or no contribution from Asp. In the 2005 paper, Fellouse et al. showed that high affinity (10 nM) binders could be selected from a library where the CDRs were composed of only 2 residues: serine and tyrosine. Although the binding affinities were no better than 34 nM, specific binders were identified against all of the 6 targets screened. Again, binding seemed to be dominated by tyrosine residues, with the smaller serine residue appearing to play a predominantly structural role, in allowing specifically spaced tyrosine residues to orient their side chains in the right way that contacts could be made with residues in the target. Interestingly, target residues contacted by tyrosine residues fall into all classes, not simply those with aromatic side chains, which in fact account for just 14% of the interactions in one of the co-crystal structures. So, mixed chains of just two amino acids are sufficient for the derivation of highly specific and high affinity binding interactions. Who knew? One random thought on re-reading these papers: specificity was defined against a panel of just 7 other proteins, maybe the abundance of tyrosine in naturally occurring antibodies is responsible for the cross-reactivity seen in so many commercially sourced antibodies?  

3 and 2. SH2 Domains Recognize Specific Phosphopeptide Sequences, by Zhou Songyang, Steven E. Shoelson, Manas Chaudhuri, Gerald Gish, Tony Pawson, Wayne G. Haser, Fred King, Tom Roberts, Sheldon Ratnofsky, Robert J. Lechleider, Benjamin G. Neel, Raymond B. Birge, J. Eduardo Fajardo, Margaret M. Chou, Hidesaburo Hanafusa, Brian Schaffhausen, and Lewis C. Cantley, Cell 72 (link is external)

1. K-Ras(G12C) inhibitors allosterically control GTP affinity and effector interactions, by Jonathan M. Ostrem, Ulf Peters, Martin L. Sos, James A. Wells & Kevan M. Shokat, Nature 503, 548-551 (2013) (link is external). This represents the first time that anyone has come anywhere close to developing a small molecule inhibitor – a drug candidate – against the protein encoded by the first oncogene ever to have been identified, by a gene that is known to be mutated in more than a quarter of human cancers, to be a driver of cancer and associated with drug resistance in patients: K-ras. This paper uses structure-guided design of an inhibitor targeting the mutant residue and then uses the inhibitor to discover a new conformation of the target. The best lead shows activity against a panel of cell lines, despite the fact that it competes only poorly with GTP for Ras protein binding. So, a paper that hasn’t had time to change the world yet, but one that might, one day, with a lot of hard graft, lead to another revolution in cancer treatment.