Christopher King

Recombinant immunotoxin for cancer treatment with low immunogenicity by identification and silencing of human T-cell epitopes.

Significance Recombinant immunotoxins have produced complete remissions in leukemia patients where many doses can be given but are less active in patients with solid tumors because their immune system makes antidrug antibodies, which inactivate the immunotoxin. To suppress the immune response, we have identified and largely silenced the T-cell epitopes responsible for the immune response. A redesigned immunotoxin with T-cell epitope mutations is highly cytotoxic to cell lines and to cells isolated from cancer patients and produces complete remissions in mice with human cancer xenografts. The approach described can be applied to deimmunize other therapeutically useful foreign proteins. Nonhuman proteins have valuable therapeutic properties, but their efficacy is limited by neutralizing antibodies. Recombinant immunotoxins (RITs) are potent anticancer agents that have produced many complete remissions in leukemia, but immunogenicity limits the number of doses that can be given to patients with normal immune systems. Using human cells, we identified eight helper T-cell epitopes in PE38, a portion of the bacterial protein Pseudomonas exotoxin A which consists of the toxin moiety of the RIT, and used this information to make LMB-T18 in which three epitopes were deleted and five others diminished by point mutations in key residues. LMB-T18 has high cytotoxic and antitumor activity and is very resistant to thermal denaturation. The new immunotoxin has a 93% decrease in T-cell epitopes and should have improved efficacy in patients because more treatment cycles can be given. Furthermore, the deimmunized toxin can be used to make RITs targeting other antigens, and the approach we describe can be used to deimmunize other therapeutically useful nonhuman proteins.

Removing T-cell epitopes with computational protein design.

Significance Proteins represent the fastest-growing class of pharmaceuticals for a diverse range of clinical applications. Computational protein design has the potential to create a novel class of therapeutics with tunable biophysical properties. However, the immune system reacts to T-cell epitope sequences in non-human proteins, leading to neutralization and elimination by the immune system. Here, we combine machine learning with structure-based protein design to identify and redesign T-cell epitopes without disrupting function of the target protein. We test the method experimentally, removing T-cell epitopes from GFP and Pseudomonas exotoxin A while maintaining function. Immune responses can make protein therapeutics ineffective or even dangerous. We describe a general computational protein design method for reducing immunogenicity by eliminating known and predicted T-cell epitopes and maximizing the content of human peptide sequences without disrupting protein structure and function. We show that the method recapitulates previous experimental results on immunogenicity reduction, and we use it to disrupt T-cell epitopes in GFP and Pseudomonas exotoxin A without disrupting function.

Structure‐based prediction of protein–peptide specificity in rosetta

Protein–peptide interactions mediate many of the connections in intracellular signaling networks. A generalized computational framework for atomically precise modeling of protein–peptide specificity may allow for predicting molecular interactions, anticipating the effects of drugs and genetic mutations, and redesigning molecules for new interactions. We have developed an extensible, general algorithm for structure‐based prediction of protein–peptide specificity as part of the Rosetta molecular modeling package. The algorithm is not restricted to any one peptide‐binding domain family and, at minimum, does not require an experimentally characterized structure of the target protein nor any information about sequence specificity; although known structural data can be incorporated when available to improve performance. We demonstrate substantial success in specificity prediction across a diverse set of peptide‐binding proteins, and show how performance is affected when incorporating varying degrees of input structural data. We also illustrate how structure‐based approaches can provide atomic‐level insight into mechanisms of peptide recognition and can predict the effects of point mutations on peptide specificity. Shortcomings and artifacts of our benchmark predictions are explained and limits on the generality of the method are explored. This work provides a promising foundation upon which further development of completely generalized, de novo prediction of peptide specificity may progress. Proteins 2010.

Automating human intuition for protein design.

In the design of new enzymes and binding proteins, human intuition is often used to modify computationally designed amino acid sequences prior to experimental characterization. The manual sequence changes involve both reversions of amino acid mutations back to the identity present in the parent scaffold and the introduction of residues making additional interactions with the binding partner or backing up first shell interactions. Automation of this manual sequence refinement process would allow more systematic evaluation and considerably reduce the amount of human designer effort involved. Here we introduce a benchmark for evaluating the ability of automated methods to recapitulate the sequence changes made to computer‐generated models by human designers, and use it to assess alternative computational methods. We find the best performance for a greedy one‐position‐at‐a‐time optimization protocol that utilizes metrics (such as shape complementarity) and local refinement methods too computationally expensive for global Monte Carlo (MC) sequence optimization. This protocol should be broadly useful for improving the stability and function of designed binding proteins. Proteins 2014; 82:858–866.

Conformational preference of ChaK1 binding peptides: a molecular dynamics study

TRPM7/ChaK1 is a recently discovered atypical protein kinase that has been suggested to selectively phosphorylate the substrate residues located in α-helices. However, the actual structure of kinase-substrate complex has not been determined experimentally and the recognition mechanism remains unknown. In this work we explored possible kinase-substrate binding modes and the likelihood of an α-helix docking interaction, within a kinase active site, using molecular modeling. Specifically kinase ChaK1 and its two peptide substrates were examined; one was an 11-residue segment from the N-terminal domain of annexin-1, a putative endogenous substrate for ChaK1, and the other was an engineered 16-mer peptide substrate determined via peptide library screening. Simulated annealing (SA), replica-exchange molecular dynamics (REMD) and steered molecular dynamics (SMD) simulations were performed on the two peptide substrates and the ChaK1-substrate complex in solution. The simulations indicate that the two substrate peptides are unlikely to bind and react with the ChaK1 kinase in a stable α-helical conformation overall. The key structural elements, sequence motifs, and amino acid residues in the ChaK1 and their possible functions involved in the substrate recognition are discussed. PACS Codes: 87.15.A-

Abstract 5573: A high-affinity Optide (optimized peptide) inhibitor of the Hippo pathway’s YAP-TEAD interaction

The HIPPO pathway plays a critical role in contact inhibition, a pathway that is commonly dysregulated in many human cancers (including liver, colon, ovarian, and lung). The signaling pathway culminates in the intranuclear interaction of the transcriptional co-activator YAP and the transcription factor TEAD. This is representative of a number of cancer driving pathways that have proven nearly impossible to drug, as they are mediated by intracellular protein-protein interactions. High throughput screening campaigns with small molecule libraries have failed to provide specific, high affinity binders capable of disrupting larger protein-protein interfaces (such as YAP-TEAD), while at the same time, antibodies cannot penetrate the cell membrane to access cytosolic and nuclear targets. Optides are small disulfide-knotted peptides (knottins) that are large enough to interfere with protein-protein interactions, but small enough to access compartments beyond the reach of antibodies. Examples include the calcines, activators of sarcoplasmic reticulum ryanodine receptors, and BLZ-100, a knottin-fluorophore conjugate that is capable of accumulating in a wide range of tumor types. Using the computational design software Rosetta, we created a library of Optides designed to interact with TEAD in locations that overlap YAP binding. Mammalian surface display screening against soluble TEAD yielded a candidate (Hit1) that binds TEAD with nanomolar affinity and inhibits YAP binding. Affinity maturation, using site saturation mutagenesis, produced an improved sub-nanomolar variant (IV1) with potent YAP inhibition. This variant was also found to be highly resistant to reduction and proteolysis, crucial for a disulfide-knotted peptide with a cytosolic target in the proteinase-rich tumor milieu. With this highly potent YAP inhibitor, efforts are now focused on cell penetration and biodistribution with the long-term goal of advancing a clinical development candidate. Citation Format: Zachary R. Crook, Philip Bradley, Gregory Sevilla, Della Friend, Chris King, Andrew Mhyre, Roland Strong, David Baker, James M. Olson. A high-affinity Optide (optimized peptide) inhibitor of the Hippo pathway’s YAP-TEAD interaction [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 5573. doi:10.1158/1538-7445.AM2017-5573

Abstract 2971: Optides (optimized knottin peptides) computationally designed to target the oncogenic HIPPO pathway

The HIPPO pathway plays a critical role in contact inhibition, a pathway that is commonly dysregulated in many human cancers (including liver, colon, ovarian, and lung) and which relies on the intranuclear interaction of the transcriptional coactivator YAP and the transcription factor TEAD(1-4). This pathway also plays a crucial role in recovery from injury; for example, its regulated repression allows hepatocytes to divide and replace tissue lost to a partial hepatectomy, after which its activation suppresses cell growth and prevents tissue overgrowth. While small molecule enzyme inhibitors have proven to be a revelation in cancer therapy, cell growth signaling via protein-protein interactions has proven much more difficult to drug. While Antibodies can be effective in targeting extracellular or cell surface epitopes, intracellular targets, such as the interaction between YAP and TEAD, are not amenable to antibody-based therapeutics. Optides are small disulfide-knotted peptides (knottins) and serve to bridge these capabilities; they are large enough to interfere with protein-protein interactions, but small enough to penetrate into the cytosol. Examples include imperatoxin, an activator of mitochondrial ryanodine receptors, and Tumor Paint, which contains an optimized variant of chlorotoxin conjugated to a fluorescent probe and is capable of accumulating in a wide range of tumor types. To test whether Optides can abrogate oncogenic signaling mediated by protein-protein interactions, we created libraries of computationally designed candidates to target the TEAD/YAP interface. The library is expressed on the surface of mammalian cells, chosen for the improved fidelity of disulfide bridge connectivity observed in the mammalian secretory pathway as compared to that found in yeast. By repetitive screening against soluble TEAD protein, we are optimizing the pool of candidates for targeting TEAD. The lead Optides will be characterized for their ability to reduce YAP-TEAD interaction, and to impair YAP-mediated cell growth. Owing to the wide variety of knottin scaffolds, both natural and in silico designed, this flexible technology could be applied to other targets in order to impair oncogenic protein-protein interactions. Citation Format: Zachary R. Crook, Philip Bradley, Chris King, Andrew J. Mhyre, David Baker, James M. Olson. Optides (optimized knottin peptides) computationally designed to target the oncogenic HIPPO pathway. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 2971.