RosettaAntibodyDesign (RAbD): A general framework for computational antibody design

A structural-bioinformatics-based computational methodology and framework have been developed for the design of antibodies to targets of interest. RosettaAntibodyDesign (RAbD) samples the diverse sequence, structure, and binding space of an antibody to an antigen in highly customizable protocols for the design of antibodies in a broad range of applications. The program samples antibody sequences and structures by grafting structures from a widely accepted set of the canonical clusters of CDRs (North et al., J. Mol. Biol., 406:228-256, 2011). It then performs sequence design according to amino acid sequence profiles of each cluster, and samples CDR backbones using a flexible-backbone design protocol incorporating cluster-based CDR constraints. Starting from an existing experimental or computationally modeled antigen-antibody structure, RAbD can be used to redesign a single CDR or multiple CDRs with loops of different length, conformation, and sequence. We rigorously benchmarked RAbD on a set of 60 diverse antibody-antigen complexes, using two design strategies-optimizing total Rosetta energy and optimizing interface energy alone. We utilized two novel metrics for measuring success in computational protein design. The design risk ratio (DRR) is equal to the frequency of recovery of native CDR lengths and clusters divided by the frequency of sampling of those features during the Monte Carlo design procedure. Ratios greater than 1.0 indicate that the design process is picking out the native more frequently than expected from their sampled rate. We achieved DRRs for the non-H3 CDRs of between 2.4 and 4.0. The antigen risk ratio (ARR) is the ratio of frequencies of the native amino acid types, CDR lengths, and clusters in the output decoys for simulations performed in the presence and absence of the antigen. For CDRs, we achieved cluster ARRs as high as 2.5 for L1 and 1.5 for H2. For sequence design simulations without CDR grafting, the overall recovery for the native amino acid types for residues that contact the antigen in the native structures was 72% in simulations performed in the presence of the antigen and 48% in simulations performed without the antigen, for an ARR of 1.5. For the non-contacting residues, the ARR was 1.08. This shows that the sequence profiles are able to maintain the amino acid types of these conserved, buried sites, while recovery of the exposed, contacting residues requires the presence of the antigen-antibody interface. We tested RAbD experimentally on both a lambda and kappa antibody-antigen complex, successfully improving their affinities 10 to 50 fold by replacing individual CDRs of the native antibody with new CDR lengths and clusters.

BRCA2, EGFR, and NTRK mutations in mismatch repair-deficient colorectal cancers with MSH2 or MLH1 mutations

Deficient mismatch repair (MMR) and microsatellite instability (MSI) contribute to ~15% of colorectal cancer (CRCs). We hypothesized MSI leads to mutations in DNA repair proteins including BRCA2 and cancer drivers including EGFR. We analyzed mutations among a discovery cohort of 26 MSI-High (MSI-H) and 558 non-MSI-H CRCs profiled at Caris Life Sciences. Caris-profiled MSI-H CRCs had high mutation rates (50% vs 14% in non-MSI-H, P < 0.0001) in BRCA2. Of 1104 profiled CRCs from a second cohort (COSMIC), MSH2/MLH1-mutant CRCs showed higher mutation rates in BRCA2 compared to non-MSH2/MLH1-mutant tumors (38% vs 6%, P < 0.0000001). BRCA2 mutations in MSH2/MLH1-mutant CRCs included 75 unique mutations not known to occur in breast or pancreatic cancer per COSMIC v73. Only 5 deleterious BRCA2 mutations in CRC were previously reported in the BIC database as germ-line mutations in breast cancer. Some BRCA2 mutations were predicted to disrupt interactions with partner proteins DSS1 and RAD51. Some CRCs harbored multiple BRCA2 mutations. EGFR was mutated in 45.5% of MSH2/MLH1-mutant and 6.5% of non-MSH2/MLH1-mutant tumors (P < 0.0000001). Approximately 15% of EGFR mutations found may be actionable through TKI therapy, including N700D, G719D, T725M, T790M, and E884K. NTRK gene mutations were identified in MSH2/MLH1-mutant CRC including NTRK1 I699V, NTRK2 P716S, and NTRK3 R745L. Our findings have clinical relevance regarding therapeutic targeting of BRCA2 vulnerabilities, EGFR mutations or other identified oncogenic drivers such as NTRK in MSH2/MLH1-mutant CRCs or other tumors with mismatch repair deficiency.

The Rosetta All-Atom Energy Function for Macromolecular Modeling and Design

Over the past decade, the Rosetta biomolecular modeling suite has informed diverse biological questions and engineering challenges ranging from interpretation of low-resolution structural data to design of nanomaterials, protein therapeutics, and vaccines. Central to Rosetta's success is the energy function: a model parametrized from small-molecule and X-ray crystal structure data used to approximate the energy associated with each biomolecule conformation. This paper describes the mathematical models and physical concepts that underlie the latest Rosetta energy function, called the Rosetta Energy Function 2015 (REF15). Applying these concepts, we explain how to use Rosetta energies to identify and analyze the features of biomolecular models. Finally, we discuss the latest advances in the energy function that extend its capabilities from soluble proteins to also include membrane proteins, peptides containing noncanonical amino acids, small molecules, carbohydrates, nucleic acids, and other macromolecules.

Functional analysis of rare variants in mismatch repair proteins augments results from computation-based predictive methods

The cancer-predisposing Lynch Syndrome (LS) arises from germline mutations in DNA mismatch repair (MMR) genes, predominantly MLH1, MSH2, MSH6, and PMS2. A major challenge for clinical diagnosis of LS is the frequent identification of variants of uncertain significance (VUS) in these genes, as it is often difficult to determine variant pathogenicity, particularly for missense variants. Generic programs such as SIFT and PolyPhen-2, and MMR gene-specific programs such as PON-MMR and MAPP-MMR, are often used to predict deleterious or neutral effects of VUS in MMR genes. We evaluated the performance of multiple predictive programs in the context of functional biologic data for 15 VUS in MLH1, MSH2, and PMS2. Using cell line models, we characterized VUS predicted to range from neutral to pathogenic on mRNA and protein expression, basal cellular viability, viability following treatment with a panel of DNA-damaging agents, and functionality in DNA damage response (DDR) signaling, benchmarking to wild-type MMR proteins. Our results suggest that the MMR gene-specific classifiers do not always align with the experimental phenotypes related to DDR. Our study highlights the importance of complementary experimental and computational assessment to develop future predictors for the assessment of VUS.

Benchmarking predictions of allostery in liver pyruvate kinase in CAGI4

The Critical Assessment of Genome Interpretation (CAGI) is a global community experiment to objectively assess computational methods for predicting phenotypic impacts of genomic variation. One of the 2015-2016 competitions focused on predicting the influence of mutations on the allosteric regulation of human liver pyruvate kinase. More than 30 different researchers accessed the challenge data. However, only four groups accepted the challenge. Features used for predictions ranged from evolutionary constraints, mutant site locations relative to active and effector binding sites, and computational docking outputs. Despite the range of expertise and strategies used by predictors, the best predictions were marginally greater than random for modified allostery resulting from mutations. In contrast, several groups successfully predicted which mutations severely reduced enzymatic activity. Nonetheless, poor predictions of allostery stands in stark contrast to the impression left by more than 700 PubMed entries identified using the identifiers "computational + allosteric." This contrast highlights a specialized need for new computational tools and utilization of benchmarks that focus on allosteric regulation.

Modeling and docking of antibody structures with Rosetta

We describe Rosetta-based computational protocols for predicting the 3D structure of an antibody from sequence (RosettaAntibody) and then docking the antibody to protein antigens (SnugDock). Antibody modeling leverages canonical loop conformations to graft large segments from experimentally determined structures, as well as offering (i) energetic calculations to minimize loops, (ii) docking methodology to refine the V_L-V_H relative orientation and (iii) de novo prediction of the elusive complementarity determining region (CDR) H3 loop. To alleviate model uncertainty, antibody-antigen docking resamples CDR loop conformations and can use multiple models to represent an ensemble of conformations for the antibody, the antigen or both. These protocols can be run fully automated via the ROSIE web server (http://rosie.rosettacommons.org/) or manually on a computer with user control of individual steps. For best results, the protocol requires roughly 1,000 CPU-hours for antibody modeling and 250 CPU-hours for antibody-antigen docking. Tasks can be completed in under a day by using public supercomputers.

Abstract 5286: A pipeline for developing a novel, predictive tool to classify variants of uncertain significance (VUS) in Lynch Syndrome

Lynch syndrome (LS) highly predisposes individuals and their families to an increased risk of colorectal cancers and to cancers of the endometrium, ovary, gastrointestinal tract, pancreas, upper urinary tract, and other tissues. The occurrence of LS has been attributed to germline, heterozygous mutations in four genes in the DNA mismatch repair (MMR) pathway: MLH1, MSH2, MSH6 and PMS2. Mutations in the MLH1 and MSH2 genes account for the majority of detectable mutations in LS, with more infrequent mutations in MSH6 and PMS2. Clinical diagnosis of LS helps direct the management of the disease and risk assessment for future cancers in the family.

A major challenge is the determination of variant pathogenicity in MMR genes, particularly those that code for missense mutations. According the InSiGHT database, there are &gt;5000 unclassified variants of unknown significance (VUS), and this number continues to grow every day due to the increased prevalence of genomic testing. Such VUS present an enormous challenge to effective genetic counseling of LS families. There are numerous predictive methods, such as SIFT and Polyphen-2, for predicting the effects of missense mutations in any gene including LS genes. Further, there is a great deal of promise for gene-specific predictors, such as PON-MMR and CoDP, that are trained on the physical and evolutionary features of known deleterious and benign mutations in the target genes of interest.

Since the number of available mutations for the LS genes with experimental phenotypes is limited, we have developed and applied a pipeline to perform experimental validation of VUS in MMR genes. MMR genes with introduced VUS were assessed in vitro in comparison to wild type genes for the effect of the VUS on steady state protein expression, localization, and functionality in inducing DNA damage checkpoints and influencing cell survival following treatment with a panel of DNA-damaging agents. We then assessed a panel of VUS in MSH2, MLH1 and PMS2 in HEK293 kidney cells and colorectal cancer (CRC) cell lines bearing deletions in the MMR genes being tested.

This mid-throughput pipeline allowed effective validation of the functional consequences of each VUS, generating additional data to enhance the efficacy of gene-specific predictors for the LS genes that we are developing. Such tools may greatly benefit the assessment of MMR gene variants identified through genetic testing and for genetic counseling of LS families.

Citation Format: Sanjeevani Arora, Peter J. Huwe, Rahmat Sikder, Manali Shah, Sanat Deshpande, Michael J. Hall, Roland L. Dunbrack, Erica A. Golemis. A pipeline for developing a novel, predictive tool to classify variants of uncertain significance (VUS) in Lynch Syndrome. [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 5286.

Assessment of refinement of template-based models in CASP11

CASP11 (the 11th Meeting on the Critical Assessment of Protein Structure Prediction) ran a blind experiment in the refinement of protein structure predictions, the fourth such experiment since CASP8. As with the previous experiments, the predictors were provided with one starting structure from the server models of each of a selected set of template-based modeling targets and asked to refine the coordinates of the starting structure toward native. We assessed the refined structures with the Z-scores of the standard CASP measures, which compare the model-target similarities of the models from all the predictors. Furthermore, we assessed the refined structures with "relative measures," which compare the improvement in accuracy of each model with respect to the starting structure. The latter provides an assessment of the extent to which each predictor group is able to improve the starting structures toward native. We utilized heat maps to display improvements in the Calpha-Calpha distance matrix for each model. The heat maps labeled with each element of secondary structure helped us to identify regions of refinement toward native in each model. Most positively scoring models show modest improvements in multiple regions of the structure, while in some models we were able to identify significant repositioning of N/C-terminal segments and internal elements of secondary structure. The best groups were able to improve more than 70% of the targets from the starting models, and by an average of 3-5% in the standard CASP measures. Proteins 2016; 84(Suppl 1):260-281. © 2016 Wiley Periodicals, Inc.

Biological function derived from predicted structures in CASP11

In CASP11, the organizers sought to bring the biological inferences from predicted structures to the fore. To accomplish this, we assessed the models for their ability to perform quantifiable tasks related to biological function. First, for 10 targets that were probable homodimers, we measured the accuracy of docking the models into homodimers as a function of GDT-TS of the monomers, which produced characteristic L-shaped plots. At low GDT-TS, none of the models could be docked correctly as homodimers. Above GDT-TS of ∼60%, some models formed correct homodimers in one of the largest docked clusters, while many other models at the same values of GDT-TS did not. Docking was more successful when many of the templates shared the same homodimer. Second, we docked a ligand from an experimental structure into each of the models of one of the targets. Docking to the models with two different programs produced poor ligand RMSDs with the experimental structure. Measures that evaluated similarity of contacts were reasonable for some of the models, although there was not a significant correlation with model accuracy. Finally, we assessed whether models would be useful in predicting the phenotypes of missense mutations in three human targets by comparing features calculated from the models with those calculated from the experimental structures. The models were successful in reproducing accessible surface areas but there was little correlation of model accuracy with calculation of FoldX evaluation of the change in free energy between the wild-type and the mutant. Proteins 2016; 84(Suppl 1):370-391. © 2016 Wiley Periodicals, Inc.

Assessment of template-based modeling of protein structure in CASP11

We present the assessment of predictions submitted in the template-based modeling (TBM) category of CASP11 (Critical Assessment of Protein Structure Prediction). Model quality was judged on the basis of global and local measures of accuracy on all atoms including side chains. The top groups on 39 human-server targets based on model 1 predictions were LEER, Zhang, LEE, MULTICOM, and Zhang-Server. The top groups on 81 targets by server groups based on model 1 predictions were Zhang-Server, nns, BAKER-ROSETTASERVER, QUARK, and myprotein-me. In CASP11, the best models for most targets were equal to or better than the best template available in the Protein Data Bank, even for targets with poor templates. The overall performance in CASP11 is similar to the performance of predictors in CASP10 with slightly better performance on the hardest targets. For most targets, assessment measures exhibited bimodal probability density distributions. Multi-dimensional scaling of an RMSD matrix for each target typically revealed a single cluster with models similar to the target structure, with a mode in the GDT-TS density between 40 and 90, and a wide distribution of models highly divergent from each other and from the experimental structure, with density mode at a GDT-TS value of ∼20. The models in this peak in the density were either compact models with entirely the wrong fold, or highly non-compact models. The results argue for a density-driven approach in future CASP TBM assessments that accounts for the bimodal nature of these distributions instead of Z scores, which assume a unimodal, Gaussian distribution. Proteins 2016; 84(Suppl 1):200-220. © 2016 Wiley Periodicals, Inc.

Cancer Signature Investigation: ERBB2 (HER2)-Activating Mutation and Amplification-Positive Breast Carcinoma Mimicking Lung Primary

Next-generation sequencing of primary and metachronous metastatic cancer lesions may impact patient care. We present a case of relapsed metastatic breast cancer with a dominant pulmonary lesion originally identified as lung adenocarcinoma. A 72-year-old, never-smoker woman with a protracted cough was found to have a large lung mass and regional lymphadenopathy on a chest CT. Lung mass biopsy showed adenocarcinoma with focal TTF-1 (thyroid transcription factor 1) positivity, favoring a lung primary. In addition to stereotactic brain radiation for cerebral metastases, she was started on carboplatin/pemetrexed. As part of the workup, the tumor was analyzed by a 50-gene targeted mutation panel, which detected 3 somatic mutations: ERBB2 (HER2) D769H activating missense mutation, TP53 Y126 inactivating truncating mutation, and SMARCB1 R374Q missense mutation. Of note, the patient had a history of stage IIA triple-negative grade 3 invasive ductal carcinoma of the left breast 1.5 years ago and received neoadjuvant chemotherapy and adjuvant radiation, and underwent a lumpectomy. Further analysis of her primary breast tumor showed a mutational profile identical to that of the lung tumor. Fluorescence in situ hybridization revealed HER2 amplification in the lung tumor, with a HER2/CEP17 ratio of 3.9. The patient was diagnosed with recurrent HER2-positive metastatic breast carcinoma with a coexisting ERBB2 (HER2) activating mutation. Chemotherapy was adjusted to include dual HER2-targeted therapy containing trastuzumab and pertuzumab, resulting in an ongoing partial response. This case demonstrates that a unique genetic mutational profile can clarify whether a tumor represents a metastatic lesion or new malignancy when conventional morphological and immunohistochemical methods are indeterminate, and can directly impact treatment decisions.

CASP 11 target classification

Protein target structures for the Critical Assessment of Structure Prediction round 11 (CASP11) and CASP ROLL were split into domains and classified into categories suitable for assessment of template-based modeling (TBM) and free modeling (FM) based on their evolutionary relatedness to existing structures classified by the Evolutionary Classification of Protein Domains (ECOD) database. First, target structures were divided into domain-based evaluation units. Target splits were based on the domain organization of available templates as well as the performance of servers on whole targets compared to split target domains. Second, evaluation units were classified into TBM and FM categories using a combination of measures that evaluate prediction quality and template detectability. Generally, target domains with sequence-related templates and good server prediction performance were classified as TBM, whereas targets without sequence-identifiable templates and low server performance were classified as FM. As in previous CASP experiments, the boundaries for classification were blurred due to the presence of significant insertions and deteriorations in the targets with respect to homologous templates, as well as the presence of templates with partial coverage of new folds. The FM category included 45 target domains, which represents an unprecedented number of difficult CASP targets provided for modeling. Proteins 2016; 84(Suppl 1):20-33. © 2016 Wiley Periodicals, Inc.

Identifying three-dimensional structures of autophosphorylation complexes in crystals of protein kinases

Protein kinase autophosphorylation is a common regulatory mechanism in cell signaling pathways. Crystal structures of several homomeric protein kinase complexes have a serine, threonine, or tyrosine autophosphorylation site of one kinase monomer located in the active site of another monomer, a structural complex that we call an "autophosphorylation complex." We developed and applied a structural bioinformatics method to identify all such autophosphorylation complexes in x-ray crystallographic structures in the Protein Data Bank (PDB). We identified 15 autophosphorylation complexes in the PDB, of which five complexes had not previously been described in the publications describing the crystal structures. These five complexes consist of tyrosine residues in the N-terminal juxtamembrane regions of colony-stimulating factor 1 receptor (CSF1R, Tyr(561)) and ephrin receptor A2 (EPHA2, Tyr(594)), tyrosine residues in the activation loops of the SRC kinase family member LCK (Tyr(394)) and insulin-like growth factor 1 receptor (IGF1R, Tyr(1166)), and a serine in a nuclear localization signal region of CDC-like kinase 2 (CLK2, Ser(142)). Mutations in the complex interface may alter autophosphorylation activity and contribute to disease; therefore, we mutated residues in the autophosphorylation complex interface of LCK and found that two mutations impaired autophosphorylation (T445V and N446A) and mutation of Pro(447) to Ala, Gly, or Leu increased autophosphorylation. The identified autophosphorylation sites are conserved in many kinases, suggesting that, by homology, these complexes may provide insight into autophosphorylation complex interfaces of kinases that are relevant drug targets.

Systematic evaluation of underlying defects in DNA repair as an approach to case-only assessment of familial prostate cancer

Risk assessment for prostate cancer is challenging due to its genetic heterogeneity. In this study, our goal was to develop an operational framework to select and evaluate gene variants that may contribute to familial prostate cancer risk. Drawing on orthogonal sources, we developed a candidate list of genes relevant to prostate cancer, then analyzed germline exomes from 12 case-only prostate cancer patients from high-risk families to identify patterns of protein-damaging gene variants. We described an average of 5 potentially disruptive variants in each individual and annotated them in the context of public databases representing human variation. Novel damaging variants were found in several genes of relevance to prostate cancer. Almost all patients had variants associated with defects in DNA damage response. Many also had variants linked to androgen signaling. Treatment of primary T-lymphocytes from these prostate cancer patients versus controls with DNA damaging agents showed elevated levels of the DNA double strand break (DSB) marker γH2AX (p < 0.05), supporting the idea of an underlying defect in DNA repair. This work suggests the value of focusing on underlying defects in DNA damage in familial prostate cancer risk assessment and demonstrates an operational framework for exome sequencing in case-only prostate cancer genetic evaluation.

Stability engineering of anti-EGFR scFv antibodies by rational design of a lambda-to-kappa swap of the VL framework using a structure-guided approach

Phage-display technology facilitates rapid selection of antigen-specific single-chain variable fragment (scFv) antibodies from large recombinant libraries. ScFv antibodies, composed of a VH and VL domain, are readily engineered into multimeric formats for the development of diagnostics and targeted therapies. However, the recombinant nature of the selection strategy can result in VH and VL domains with sub-optimal biophysical properties, such as reduced thermodynamic stability and enhanced aggregation propensity, which lead to poor production and limited application. We found that the C10 anti-epidermal growth factor receptor (EGFR) scFv, and its affinity mutant, P2224, exhibit weak production from E. coli. Interestingly, these scFv contain a fusion of lambda3 and lambda1 V-region (LV3 and LV1) genes, most likely the result of a PCR aberration during library construction. To enhance the biophysical properties of these scFvs, we utilized a structure-based approach to replace and redesign the pre-existing framework of the VL domain to one that best pairs with the existing VH. We describe a method to exchange lambda sequences with a more stable kappa3 framework (KV3) within the VL domain that incorporates the original lambda DE-loop. The resulting scFvs, C10KV3_LV1DE and P2224KV3_LV1DE, are more thermodynamically stable and easier to produce from bacterial culture. Additionally, C10KV3_LV1DE and P2224KV3_LV1DE retain binding affinity to EGFR, suggesting that such a dramatic framework swap does not significantly affect scFv binding. We provide here a novel strategy for redesigning the light chain of problematic scFvs to enhance their stability and therapeutic applicability.

Abstract 537: Information-driven approaches to predicting familial risk for prostate cancer

Given the known genetic heterogeneity of prostate cancer, risk assessment for this disease is a field that is likely to be positively impacted by improved interpretation of genetic variants identified through exome sequencing. At present, there is insufficient evidence to base decisions regarding screening, diagnosis and management of prostate cancer on genetic tests. While a few genes (such as BRCA2 and HOXB13) are definitively linked to prostate cancer risk, they are not found in sizeable patient populations. However, a growing number of candidates for prostate cancer risk have been proposed. As information resources including the TCGA and other consortium databases expand, clinicians and patients will face issues with how to interpret genetic data variant regarding cancer risk. The goal of the present work is to develop a paradigm that addresses the plausible scenario of a clinician seeking to provide prostate cancer genetic risk assessment for patients with personal and/or a family history of one or more types of cancer, but for whom access to relative specimens for sequencing analysis is highly limited. In particular, we focus on the development of an operational framework through which to select and evaluate germ-line variants which may contribute to familial cancer risk.

Our study evaluates the current potential of integrated informatics resources to help identify clinically informative variants following exome sequencing of germ-line DNA from 12 prostate cancer patients with elevated risk of prostate or other cancers based on family history. We explored variants in genes meeting specific criteria: 1) genes thought to be relevant to hereditary prostate cancer, or experimentally validated as regulating the growth of prostate tumors in cell or animal models; 2) genes affected by somatic changes in cancer; 3) genes involved in DNA damage repair (DDR) including recently described androgen-regulated DDR genes; and 4) genes involved in disorders of glycosylation. We first filtered exome data based on the observed frequency of gene variants in the general population. We then utilized five independent in silico predictors and expert analysis of protein structure and interactions to identify function-disrupting rare variants. We described an average of 4 potentially disruptive variants in each individual and annotated them in the context of the human variation data accumulated over the last few years and represented in various public databases. Novel variants were found in PALB2, RAD54L2, HSD3B1, NRIP1, SCN11A, CYPBP1, SULT1E1 and UBE2D3. Molecular modeling of a p.(S221N) mutation in the steroid binding site of the aldo-keto reductase AKR1C1 showed it would be predicted to disrupt the active site, influencing androgen metabolism. Our study highlights the need and potential for well-curated databases of variants with clinical relevance that will ultimately facilitate germ-line exome testing in the clinical setting.

Citation Format: Emmanuelle Nicolas, Yan Zhou, Ilya G. Serebriiskii, Mark D. Andrake, Elizabeth A. Handorf, Roland L. Dunbrack, Veda N. Giri, Eric A. Ross, Erica A. Golemis, Michael J. Hall, Mary B. Daly. Information-driven approaches to predicting familial risk for prostate cancer. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 537. doi:10.1158/1538-7445.AM2015-537

Abstract LB-034: Identification of novel autophosphorylation structures in crystals of protein kinases

Cancer therapy depends heavily on the ability to effectively control the activity of oncogenic kinases. Autophosphorylation is a common regulatory mechanism of kinases in signaling pathways, and commonly elevated in cancer. Several autophosphorylation complexes have been identified from within crystals of protein kinases, with a known autophosphorylation site of one kinase monomer sitting in the active site of another monomer of the same protein in the crystal. We have utilized a structural bioinformatics method to identify all such autophosphorylation complexes in X-ray crystal structures in the Protein Data Bank by generating all unique kinase/kinase interfaces within and between asymmetric units of each crystal and measuring the distance between the hydroxyl oxygen of the autophosphorylation sites and the oxygen atoms of the active site aspartic acid residue side chain. With this approach, we have identified 15 autophosphorylation complexes in the PDB, of which 5 complexes have not previously been described.

Of greatest interest are five structures of activation loop autophosphorylation - PAK1 (T423), IRAK4 (T345), IGF1R (Y1165 and Y1166), and LCK (Y394), two of which we have identified for the first time (IGF1R-Y1166 and LCK). We show that 269 human kinases have potential S/T or Y phosphorylation sites at positions analogous to the PAK1, IGF1R-Y1165/LCK, and IGF1R-Y1166 structures, and that there are 182 such positions that are annotated as phosphorylation sites in Uniprot. To assess the functional importance of the LCK dimer, we performed mutational analysis of residues in the autophosphorylation complex interface of LCK and found that mutations disrupting the interface either severely impaired autophosphorylation (T445D and N446D) or increased it (P447L,A,G). The P447L mutation has been previously found in a T-cell leukemia cell line and associated with activation of LCK.

Three structures of receptor tyrosine kinases contain autophosphorylation complexes of the juxtamembrane segment just N-terminal to the kinase domain, two of which are identified for the first time. One of these, CSF1R (Y561) is a homologous site to a known c-KIT (Y568) autophosphorylation structure. The other is in EPHA2 (Y594), which is homologous to Y570 in c-KIT, which is also an autophosphorylation site. Twenty receptor tyrosine kinases contain autophosphorylation sites at one or both of these positions. Phosphorylation at these sites is associated with interaction with SRC family kinases and other downstream effectors of receptor tyrosine kinase signaling.

These structures provide critical information on domain-domain interactions and substrate specificity in autophosphorylation, as well as opportunities for understanding the role of certain cancer driver mutations and the development of non-ATP-competitive inhibitors that block dimerization shown in these structures.

Note: This abstract was not presented at the meeting.

Citation Format: Qifang Xu, Kimberly Malecka, Jeffrey Peterson, Roland L. Dunbrack. Identification of novel autophosphorylation structures in crystals of protein kinases. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr LB-034. doi:10.1158/1538-7445.AM2015-LB-034

Defects in DNA Repair Genes Predict Response to Neoadjuvant Cisplatin-based Chemotherapy in Muscle-invasive Bladder Cancer

BACKGROUND

Cisplatin-based neoadjuvant chemotherapy (NAC) before cystectomy is the standard of care for muscle-invasive bladder cancer (MIBC), with 25-50% of patients expected to achieve a pathologic response. Validated biomarkers predictive of response are currently lacking.

OBJECTIVE

To discover and validate biomarkers predictive of response to NAC for MIBC.

DESIGN, SETTING, AND PARTICIPANTS

Pretreatment MIBC samples prospectively collected from patients treated in two separate clinical trials of cisplatin-based NAC provided the discovery and validation sets. DNA from pretreatment tumor tissue was sequenced for all coding exons of 287 cancer-related genes and was analyzed for base substitutions, indels, copy number alterations, and selected rearrangements in a Clinical Laboratory Improvements Amendments-certified laboratory.

OUTCOME MEASUREMENTS AND STATISTICAL ANALYSIS

The mean number of variants and variant status for each gene were correlated with response. Variant data from the discovery cohort were used to create a classification tree to discriminate responders from nonresponders. The resulting decision rule was then tested in the independent validation set.

RESULTS AND LIMITATIONS

Patients with a pathologic complete response had more alterations than those with residual tumor in both the discovery (p=0.024) and validation (p=0.018) sets. In the discovery set, alteration in one or more of the three DNA repair genes ATM, RB1, and FANCC predicted pathologic response (p<0.001; 87% sensitivity, 100% specificity) and better overall survival (p=0.007). This test remained predictive for pathologic response in the validation set (p=0.033), with a trend towards better overall survival (p=0.055). These results require further validation in additional sample sets.

CONCLUSIONS

Genomic alterations in the DNA repair-associated genes ATM, RB1, and FANCC predict response and clinical benefit after cisplatin-based chemotherapy for MIBC. The results suggest that defective DNA repair renders tumors sensitive to cisplatin.

PATIENT SUMMARY

Chemotherapy given before bladder removal (cystectomy) improves the chance of cure for some but not all patients with muscle-invasive bladder cancer. We found a set of genetic mutations that when present in tumor tissue predict benefit from neoadjuvant chemotherapy, suggesting that testing before chemotherapy may help in selecting patients for whom this approach is recommended.

Abstract 4298: Defects in DNA repair genes and sensitivity to cisplatin based neoadjuvant chemotherapy (NAC) for bladder cancer

Background: Cisplatin based NAC prior to cystectomy is standard of care for MIBC, with 40-50% expected to respond with ≤pT1N0M0. Biomarkers predictive of response are lacking.

Methods: MIBC pts who received 3 cycles of cisplatin based NAC on 1 of 2 prospective multicenter clinical trials were included. Pts treated with accelerated methotrexate, vinblastine, doxorubicin + cisplatin (AMVAC) provided the discovery set [n = 34, 15/34 (44%) ≤pT1N0M0]. Pts treated with dose dense gemcitabine + cisplatin (DDGC) provided the validation set [n = 24, 11/24 (46%) ≤pT1N0M0]. DNA from pre-treatment tumor tissue underwent sequencing for all coding exons of 287 cancer related genes and was analyzed for presence of base substitutions, indels, copy number alterations, and selected re-arrangements. The mean number of variants and variant status for each gene were correlated with response using two-sample t-test and Fisher's exact tests. Variant data were used to create a classification tree to discriminate responders vs. non-responders in the AMVAC discovery cohort. The resulting decision rule was then tested in the independent DDGC validation set. Overall survival analysis was performed using Kaplan-Meier.

Results: Pts with pT0 had significantly more alterations than those with residual tumor in both the AMVAC discovery (p = .024) and DDGC validation (p = 0.018) set. In the AMVAC discovery set, alteration in ≥1 of the three DNA repair genes ATM, RB1 or FANCC predicted for ≤pT1N0M0 (p&lt;0.001, 87% sensitivity, 100% specificity) and improved overall survival (OS) (p = 0.007). This test remained predictive for ≤pT1N0M0 in the DDGC validation set (p = 0.033), with a trend towards improved OS (p = 0.07) at short median follow up of 14.3 mo.

Conclusions: Alterations in ≥1 of ATM, RB1 and FANCC predict response to cisplatin based chemotherapy defined as ≤pT1N0M0 in both our AMVAC discovery and DDGC validation sets. We hypothesize that defects in these genes, which are important for maintenance of chromatin structure and DNA repair, confer sensitivity to DNA damaging chemotherapy and explain the accumulation of alterations seen among pts with pT0. External validation in collaboration with the cooperative groups is planned.

Citation Format: Elizabeth R. Plimack, Roland L. Dunbrack, Timothy A. Brennan, Mark D. Andrake, Yan Zhou, Ilya Serebriiskii, Essel Dulaimi Al-Saleem, Jean Hoffman-Censits, Marijo Bilusic, Yu-Ning Wong, Alexander Kutikov, Rosalia Viterbo, Richard Greenberg, David Chen, Costas D. Lallas, Edouard J. Trabulsi, Roman Yelensky, Vincent A. Miller, Erica Golemis, Eric Ross. Defects in DNA repair genes and sensitivity to cisplatin based neoadjuvant chemotherapy (NAC) for bladder cancer. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 4298. doi:10.1158/1538-7445.AM2015-4298

Characteristics of clinical trials that require participants to be fluent in English

BACKGROUND/AIMS

Diverse samples in clinical trials can make findings more generalizable. We sought to characterize the prevalence of clinical trials in the United States that required English fluency for participants to enroll in the trial.

METHODS

We randomly chose over 10,000 clinical trial protocols registered with ClinicalTrials.gov and examined the inclusion and exclusion criteria of the trials. We compared the relationship of clinical trial characteristics with English fluency inclusion requirements. We merged the ClinicalTrials.gov data with US Census and American Community Survey data to investigate the association of English-language restrictions with ZIP-code-level demographic characteristics of participating institutions. We used Chi-squared tests, t-tests, and logistic regression models for analyses.

RESULTS

English fluency requirements have been increasing over time, from 1.7% of trials having such requirements before 2000 to 9.0% after 2010 (p < 0.001 from Chi-squared test). Industry-sponsored trials had low rates of English fluency requirements (1.8%), while behavioral trials had high rates (28.4%). Trials opening in the Northeast of the United States had the highest regional English requirement rates (10.7%), while trials opening in more than one region had the lowest (3.3%, p<0.001). Since 1995, trials opening in ZIP codes with larger Hispanic populations were less likely to have English fluency requirements (odds ratio=0.92 for each 10% increase in proportion of Hispanics, 95% confidence interval=0.86-0.98, p=0.013). Trials opening in ZIP codes with more residents self-identifying as Black/African American (odds ratio=1.87, 95% confidence interval=1.36-2.58, p<0.001 for restricted cubic spline term) or Asian (odds ratio=1.16 for linear term, 95% confidence interval=1.07-1.25, p<0.001) were more likely to have English fluency requirements. ZIP codes with higher poverty rates had trials with more English-language restrictions (odds ratio=1.06 for a 10% poverty rate increase, 95% confidence interval=1.001-1.11, p=0.045). There was a statistically significant interaction between year and intervention type, such that the increase in English fluency requirements was more common for some interventions than for others.

CONCLUSION

The proportion of clinical trials registered with ClinicalTrials.gov that have English fluency requirements for study inclusion has been increasing over time. English-language restrictions are associated with a number of characteristics, including the demographic characteristics of communities in which the sponsoring institutions are located.

Conformational analysis of the DFG-out kinase motif and biochemical profiling of structurally validated type II inhibitors

Structural coverage of the human kinome has been steadily increasing over time. The structures provide valuable insights into the molecular basis of kinase function and also provide a foundation for understanding the mechanisms of kinase inhibitors. There are a large number of kinase structures in the PDB for which the Asp and Phe of the DFG motif on the activation loop swap positions, resulting in the formation of a new allosteric pocket. We refer to these structures as “classical DFG-out” conformations in order to distinguish them from conformations that have also been referred to as DFG-out in the literature but that do not have a fully formed allosteric pocket. We have completed a structural analysis of almost 200 small molecule inhibitors bound to classical DFG-out conformations; we find that they are recognized by both type I and type II inhibitors. In contrast, we find that nonclassical DFG-out conformations strongly select against type II inhibitors because these structures have not formed a large enough allosteric pocket to accommodate this type of binding mode. In the course of this study we discovered that the number of structurally validated type II inhibitors that can be found in the PDB and that are also represented in publicly available biochemical profiling studies of kinase inhibitors is very small. We have obtained new profiling results for several additional structurally validated type II inhibitors identified through our conformational analysis. Although the available profiling data for type II inhibitors is still much smaller than for type I inhibitors, a comparison of the two data sets supports the conclusion that type II inhibitors are more selective than type I. We comment on the possible contribution of the DFG-in to DFG-out conformational reorganization to the selectivity.

PyIgClassify: a database of antibody CDR structural classifications

Classification of the structures of the complementarity determining regions (CDRs) of antibodies is critically important for antibody structure prediction and computational design. We have previously performed a clustering of antibody CDR conformations and defined a systematic nomenclature consisting of the CDR, length and an integer starting from the largest to the smallest cluster in the data set (e.g. L1-11-1). We present PyIgClassify (for Python-based immunoglobulin classification; available at http://dunbrack2.fccc.edu/pyigclassify/), a database and web server that provides access to assignments of all CDR structures in the PDB to our classification system. The database includes assignments to the IMGT germline V regions for heavy and light chains for several species. For humanized antibodies, the assignment of the frameworks is to human germlines and the CDRs to the germlines of mice or other species sources. The database can be searched by PDB entry, cluster identifier and IMGT germline group (e.g. human IGHV1). The entire database is downloadable so that users may filter the data as needed for antibody structure analysis, prediction and design.

Issues in interpreting the in vivo activity of Aurora-A

INTRODUCTION

Based on its role as a mitotic regulatory kinase, overexpressed and associated with aneuploidy in cancer, small-molecule inhibitors have been developed for Aurora-A (AURKA) kinase. In preclinical and clinical assessments, these agents have shown efficacy in inducing stable disease or therapeutic response. In optimizing the use of Aurora-A inhibitors, it is critical to have robust capacity to measure the kinase activity of Aurora-A in tumors.

AREAS COVERED

We provide an overview of molecular mechanisms of mitotic and non-mitotic activation of Aurora-A kinase, and interaction of Aurora-A with its regulatory partners. Typically, Aurora-A activity is measured by use of phospho-antibodies targeting an autophosphorylated T288 epitope. However, recent studies have identified alternative means of Aurora-A activation control, including allosteric regulation by partners, phosphorylation on alternative activating residues (S51, S98), dephosphorylation on inhibitory sites (S342) and T288 phosphorylation by alternative kinases such as Pak enzymes. Additional work has shown that the relative abundance of Aurora-A partners can affect the activity of Aurora-A inhibitors, and that Aurora-A activation also occurs in interphase cells.

EXPERT OPINION

Taken together, this work suggests the need for comprehensive analysis of Aurora-A activity and expression of Aurora-A partners in order to stratify patients for likely therapeutic response.