Search

The Assay Guidance Manual (AGM) is an eBook of best practices for the design, development, and implementation of robust assays for early drug discovery. Initiated by pharmaceutical company scientists, the manual provides guidance for designing a "testing funnel" of assays to identify genuine hits using high-throughput screening (HTS) and advancing them through preclinical development. Combined with a workshop/tutorial component, the overall goal of the AGM is to provide a valuable resource for training translational scientists.

A major focus of our pediatric cancer research is the discovery of chemical probes to further our understanding of the biology of leukemia harboring fusion proteins arising from chromosomal rearrangements, and to develop novel specifically targeted therapies. The NUP98-NSD1 fusion protein occurs in a highly aggressive subtype of acute myeloid leukemia after rearrangement of the genes NUP98 and NSD1. The methyltransferase activity of NSD1 is retained in the fusion, and it gives rise to abnormally high levels of methylation at lysine 36 on histone 3, enforcing oncogene activation. Therefore, inhibition of the methyltransferase activity of NUP98-NSD1 may be considered a viable therapeutic strategy. Here, we report the development and validation of a highly sensitive and robust luminescence-based assay for NSD1 and other methyltransferases that use S-adenosylmethionine (SAM) as a methyl donor. The assay quantifies S-adenosylhomocysteine (SAH), which is produced during methyl transfer from SAM. SAH is converted enzymatically to adenosine monophosphate (AMP); in the process, adenosine triphosphate (ATP) is consumed and the amount of ATP remaining is measured using a luminescent assay kit. The assay was validated by pilot high-throughput screening (HTS), dose-response confirmation of hits, and elimination of artifacts through counterscreening against SAH detection in the absence of NSD1. The known methyltransferase inhibitor suramin was identified, and profiled for selectivity against the histone methyltransferases EZH2, SETD7, and PRMT1. HTS using the luminescent NSD1 assay described here has the potential to deliver selective NSD1 inhibitors that may serve as leads in the development of targeted therapies for NUP98-NSD1-driven leukemias.

Rearrangements of the mixed-lineage leukemia (MLL) gene occur predominately in pediatric leukemia cases and are generally predictors of a poor prognosis. These chromosomal rearrangements result in fusion of the protein MLL to one of more than 60 protein partners. MLL fusions are potent inducers of leukemia through activation of oncogene expression; therefore, targeting this transcriptional activation function may arrest MLL-rearranged (MLL-R) leukemia. Leukemic cell lines harboring the most common fusion protein, MLL-AF4, require the direct interaction of AF4 with the transcription factor AF9 to survive and self-renew; disrupting this interaction with a cell-penetrating AF4-derived peptide results in cell death, suggesting that the AF4-AF9 interaction could be a viable target for a novel MLL-R leukemia therapy. Here we describe the use of AlphaScreen technology to develop a high-throughput screening (HTS) assay to detect nonpeptidic inhibitors of AF4-AF9 binding. The assay is economical, requiring only low nanomolar concentrations of biotinylated AF4-derived peptide and FLAG-tagged AF9 in low-volume 384-well plates. A Z'-factor of 0.71 and a signal-to-background ratio of 21.3 showed the assay to be robust, and sensitivity to inhibition was demonstrated with competing AF4-derived peptides. Two pilot screens comprising 5,680 compounds served as validation for HTS at Nemours and the Broad Institute. Assay artifacts were excluded using a counterscreen comprising a biotinylated FLAG peptide. This is the first reported HTS-compatible assay to identify compounds that inhibit a key binding interaction of an MLL fusion partner, and the results presented here demonstrate suitability for screening large chemical libraries in high-density, low-volume plate formats.

RAD51 is a key protein of homologous recombination that plays a critical role in the repair of DNA double-strand breaks (DSB) and interstrand cross links (ICL). To better understand the cellular function(s) of human RAD51, we propose to develop specific RAD51 inhibitors. RAD51 inhibitors may also help to increase the potency of anticancer drugs that act by inducing DSBs or ICLs, e.g., cisplatin or ionizing radiation. In vitro, RAD51 promotes DNA strand exchange between homologous ss- and dsDNA. Here, we developed a DNA strand exchange assay based on fluorescence resonance energy transfer and used this assay to identify RAD51 inhibitors by high throughput screening of the NIH Small Molecule Repository (>200,000 compounds). Seventeen RAD51 inhibitors were identified and analyzed for selectivity using additional non-fluorescent DNA-based assays. As a result, we identified a compound (B02) that specifically inhibited human RAD51 (IC50 = 27.4 μM), but not its E. coli homologue RecA (IC50 > 250 μM). Two other compounds (A03 and A10) were identified that inhibited both RAD51 and RecA, but not the structurally unrelated RAD54 protein. The structure-activity relationship (SAR) analysis allowed us to identify the structural components of B02 that are critical for RAD51 inhibition. The described approach can be used for identification of specific inhibitors of other human proteins that play an important role in DNA repair, e.g., RAD54 or Bloom’s syndrome helicase.

Microbial adhesion and biofilms have important implications for human health and disease. Candida albicans is an opportunistic pathogen which forms drug-resistant biofilms that contribute to the recalcitrance of disease. We have developed a high-throughput screen for potentiators of clotrimazole, a common therapy for Candida infections, including vaginitis and thrush. The screen was performed against C. albicans biofilms grown in microtitre plates in order to target the most resilient forms of the pathogen.Biofilm growth, in individual wells of 384-well plates, was measured using the metabolic indicator alamarBlue® and found to be very consistent and reproducible. This assay was used to test the effect of more than 120 000 small molecule compounds from the NIH Molecular Libraries Small Molecule Repository, and compounds that enhanced the activity of clotrimazole or acted on the biofilms alone were identified as hits.Nineteen compounds (0.016% hit rate) were identified and found to cause more than 30% metabolic inhibition of biofilms compared with clotrimazole alone, which had a modest effect on biofilm viability at the concentration tested. Hits were confirmed for activity against biofilms with dose-response measurements. Several compounds had increased activity in combination with clotrimazole, including a 1,3-benzothiazole scaffold that exhibited a100-fold improvement against biofilms of three separate C. albicans isolates. Cytotoxicity experiments using human fibroblasts confirmed the presence of lead molecules with favourable antifungal activity relative to cytotoxicity.We have validated a novel approach to identify antifungal potentiators and completed a high-throughput screen to identify small molecules with activity against C. albicans biofilms. These small molecules may specifically target the biofilm and make currently available antifungals more effective.

Parallel high-throughput automated assays are described for the measurement of cell growth and β-galactosidase reporter gene expression from a single culture of the yeast S. cerevisiae. The dual assay measures the effect of test compounds on expression of a specific gene of interest linked to the β-galactosidase reporter gene, and simultaneously tests for compound toxicity and other effects on cell growth. Examples of assay development and validation results are used to illustrate how this protocol may be used to screen two yeast cell lines in parallel. Yeast cells are grown overnight in V-bottom polypropylene 384-well plates, after which portions of the cell suspension are transferred to clear and to white flat-bottom 384-well plates for measurement of cell growth and reporter gene expression, respectively. Cell growth is determined by measurement of absorbance at 595 nm, and β-galactosidase expression is quantified by Beta-Glo, a commercially available luminescent β-galactosidase substrate. Curr. Protoc. Chem. Biol. 3:1-14 © 2011 by John Wiley & Sons, Inc.

Recently, we identified a thiocarbazate that exhibits potent inhibitory activity against human cathepsin L. Since this structure represents a novel chemotype with potential for activity against the entire cysteine protease family, we designed, synthesized, and assayed a series of analogs to probe the mechanism of action, as well as the structural requirements for cathepsin L activity. Molecular docking studies using coordinates of a papain-inhibitor complex as a model for cathepsin L provided useful insights.

Orthogonal pooling was evaluated as a strategy for the rapid screening of multiple cysteine and serine proteases against large compound libraries. To validate the method the human cysteine protease cathepsin B was screened against a library of 64,000 individual compounds and also against the same library mixed 10 compounds per well. The orthogonal pooling method used resulted in each compound being present in two wells, mixed with a different set of nine other compounds in each location. Thus hits were identified based on activity in both locations, avoiding the need for retesting of each component of active mixtures. Hits were tested in dose–response both in the dithiothreitol (DTT)-containing buffer used in the primary HTS and in buffer containing cysteine in place of DTT to rule out artifacts due to oxidative inactivation of the enzyme. Comparison of the confirmed actives from single-compound and mixture screening showed that mixture screening identified all of the actives from single-compound HTS. Based on these results the orthogonal pooling strategy has been used successfully to rapidly screen several cysteine and serine proteases.

A novel small molecule thiocarbazate (PubChem SID 26681509), a potent inhibitor of human cathepsin L (EC 3.4.22.15) with an IC(50) of 56 nM, was developed after a 57,821-compound screen of the National Institutes of Health Molecular Libraries Small Molecule Repository. After a 4-h preincubation with cathepsin L, this compound became even more potent, demonstrating an IC(50) of 1.0 nM. The thiocarbazate was determined to be a slow-binding and slowly reversible competitive inhibitor. Through a transient kinetic analysis for single-step reversibility, inhibition rate constants were k(on) = 24,000 M(-1)s(-1) and k(off) = 2.2 x 10(-5) s(-1) (K(i) = 0.89 nM). Molecular docking studies were undertaken using the experimentally derived X-ray crystal structure of papain/CLIK-148 (1cvz. pdb). These studies revealed critical hydrogen bonding patterns of the thiocarbazate with key active site residues in papain. The thiocarbazate displayed 7- to 151-fold greater selectivity toward cathepsin L than papain and cathepsins B, K, V, and S with no activity against cathepsin G. The inhibitor demonstrated a lack of toxicity in human aortic endothelial cells and zebrafish. In addition, the thiocarbazate inhibited in vitro propagation of malaria parasite Plasmodium falciparum with an IC(50) of 15.4 microM and inhibited Leishmania major with an IC(50) of 12.5 microM.

Substituted pyrazole esters were identified as hits in a high throughput screen (HTS) of the NIH Molecular Libraries Small Molecule Repository (MLSMR) to identify inhibitors of the enzyme cathepsin B. Members of this class, along with functional group analogs, were synthesized in an effort to define the structural requirements for activity. Analog characterization was hampered by the need to include a reducing agent such as dithiothreitol (DTT) or cysteine in the assay, highlighting the caution required in interpreting biological data gathered in the presence of such nucleophiles. Despite the confounding effects of DTT and cysteine, our studies demonstrate that the pyrazole 1 acts as alternate substrate for cathepsin B, rather than as an inhibitor.

A conformational analysis of a stereochemically complete set of peptide analogues based on a cis-enediol unit is presented. The cis-enediol unit, which can replace a two or a three amino acid segment of a peptide, contains two "side chains", four asymmetrical carbon atoms, and six free dihedral angles. To determine the accessible conformational space, the molecules are divided into three fragments, each containing two free dihedral angles. The energy surfaces are computed for all dihedral angle values, and the possible conformations of the cis-enediol unit analogues are built using all combinations of the surface minima. Such a "build-up" procedure, which is very fast, is able to reproduce 75% of the minima obtained from a full dihedral angle exploration of the conformational space. The cis-enediol unit minima are compared with the corresponding di- and tripeptide minima; all peptide minima can be closely matched by a cis-enediol unit minimum of low energy (less than 2.2 kcal/mol above the lowest energy conformer). However, there are low energy minima of the cis-enediol unit that have no corresponding minima in peptides. The results are shown to depend strongly on the chirality of the analogues. The ability of each of the stereoisomers to mimic natural peptides, evaluated by the present approach, is correlated with its experimental activity in a renin inhibition assay.

A miniaturized fluorescent assay is described that monitors the conversion of NADPH to NADP+. The same assay format may also be used to measure NADH to NAD+ conversion. Examples of assay development and validation results are presented to illustrate the use of this protocol to screen an enzyme that consumes NADPH as a cofactor during conversion of substrate to a reduced product. Enzymatic assays are carried out low volume 384-well plates, in which the turnover of NADPH is monitored by the decrease in fluorescent emission at 460 nm between an initial fluorescence measurement and a second reading after 90 minutes. A follow-up assay is used to rule out false positive artifacts arising from compounds that fluoresce at 460 nm.

Has the impact of irreproducibility on the discovery and development of drugs, as with global warming, metaphorically speaking, crept up on us as we slept? Or is the problem more an issue of heightened awareness? We currently find ourselves in a time when the impact of irreproducibility can easily be amplified by the combinatorial effect of our increasing reliance on advanced technologies and unrealistic expectations of how scientific truths unfold. How and why we got here is a topic that has been written on extensively1–3 and is probably as complex as any other problem, given the dependence of science today on so many external forces. Through a series of questions, we asked members of our editorial board their opinions on scientific irreproducibility. They chose to answer the same questions from different levels, indicating the depth of the problem and perhaps where they each believe change for the better needs to begin. My thanks to the participants. —Jim Inglese, PhD Editor-in-Chief Assay and Drug Deve...

Malaria is a global health problem that causes significant mortality and morbidity, with more than 1 million deaths per year caused by Plasmodium falciparum . Most antimalarial drugs face decreased efficacy due to the emergence of resistant parasites, which necessitates the discovery of new drugs. To identify new antimalarials, we developed an automated 384-well plate screening assay using P. falciparum parasites that stably express cytoplasmic firefly luciferase. After initial optimization, we tested two different types of compound libraries: known bioactive collections (Library of Pharmacologically Active Compounds [LOPAC] and the library from the National Institute of Neurological Disorders and Stroke [NINDS]) and a library of uncharacterized compounds (ChemBridge). A total of 12,320 compounds were screened at 5.5 μM. Selecting only compounds that reduced parasite growth by 85% resulted in 33 hits from the combined bioactive collection and 130 hits from the ChemBridge library. Fifteen novel drug-like compounds from the bioactive collection were found to be active against P. falciparum . Twelve new chemical scaffolds were found from the ChemBridge hits, the most potent of which was a series based on the 1,4-naphthoquinone scaffold, which is structurally similar to the FDA-approved antimalarial atovaquone. However, in contrast to atovaquone, which acts to inhibit the bc 1 complex and block the electron transport chain in parasite mitochondria, we have determined that our new 1,4-napthoquinones act in a novel, non- bc 1 -dependent mechanism and remain potent against atovaquone- and chloroquine-resistant parasites. Ultimately, this study may provide new probes to understand the molecular details of the malaria life cycle and to identify new antimalarials.

Pharmacological inhibition of dietary lipid absorption induces favorable changes in serum lipoprotein levels in patients that are at risk for cardiovascular disease and is considered an adjuvant or alternative treatment with HMG-CoA reductase inhibitors (statins). Here we demonstrate the feasibility of identifying novel inhibitors of intestinal lipid absorption using the zebrafish system. A pilot screen of an unbiased chemical library identified novel compounds that inhibited processing of fluorescent lipid analogues in live zebrafish larvae. Secondary assays identified those compounds suitable for testing in mammals and provided insight into mechanism of action, which for several compounds could be distinguished from ezetimibe, a drug used to inhibit cholesterol absorption in humans that broadly inhibited lipid absorption in zebrafish larvae. These findings support the utility of zebrafish screening assays to identify novel compounds that target complex physiological processes.

High-throughput screening of 201,368 compounds revealed that 1-(3-(5-ethyl-5H-[1,2,4]triazino[5,6-b]indol-3-ylthio)propyl)-1H-benzo[d]imidazol-2(3H)-one (SID 7975595) inhibited RmlC a TB cell wall biosynthetic enzyme. SID 7975595 acts as a competitive inhibitor of the enzyme's substrate and inhibits RmlC as a fast-on rate, fully reversible inhibitor. An analog of SID 7975595 had a K(i) of 62nM. Computer modeling showed that the binding of the tethered two-ringed system into the active site occurred at the thymidine binding region for one ring system and the sugar region for the other ring system.

A pre-steady-state multiple-turnover kinetic burst is observed during hydrolysis of phenyl acetate by the catalytic antibody, 20G9. The burst is caused by partial product inhibition by phenol (Ki,app = 2.5 microM), which lowers both kcat and KM by almost an order of magnitude without affecting kcat/KM. The acid limb of the steady-state kcat pH profile of native 20G9 has a pKa of 9.6, suggesting a catalytic role for tyrosine. Additional evidence for an essential tyrosyl residue is that mild treatment of 20G9 with tetranitromethane nitrates a single tyrosine per equivalent of antigen binding sites and the mononitrated derivative has less than 5% of the native activity. Near-UV absorbance spectroscopy suggests that the alternative substrates N-carbobenzoxyglycine O-phenyl ester (ZG-OPh) and N-acetylglycine O-phenyl ester (AcG-OPh) acylate multiple tyrosines on the antibody. Neither ZG-OPh nor AcG-OPh are measurably catalyzed once appreciable acylation has taken place. Antibody acylated by ZG-OPh is inactive toward phenyl acetate hydrolysis, but can be reactivated by hydroxylamine. The data and derived kinetic rate equations are consistent with an acyl mechanism for phenyl acetate hydrolysis in which phenol inhibits by binding to a covalent O-acetyltyrosyl intermediate, slowing deacylation. Although the data are consistent with such a mechanism, they do not rule out other plausible, yet less unifying mechanisms of phenol inhibition; the observed burst could conceivably result from partial mixed phenol inhibition or from phenol-induced nonproductive substrate binding. Because antibodies often use tyrosines in antigen binding, tyrosyl catalytic antibodies may be commonly encountered in the future.

Catalytic antibodies are introduced as an important new class of biomolecules for molecular recognition in biosensors in which the binding sites are continually regenerated by the catalytic reaction of the substrate. Consequently, molecular recognition by catalytic antibodies can yield reversible immunoblosensors. In this example, a prototype potentiometric biosensor is described in which a micro-pH electrode is modified with a catalytic antibody that catalyzes the hydrolysis of phenyl acetate, producing hydrogen ions that can be monitored by the electrode. The reversible response is linear with the log of substrate concentration over a range of 20-500 microM with a detection limit of 5 microM under the conditions of this study. Alternative applications of catalytic antibodies in other biosensor configurations are discussed.

Reporter-gene assays that employ the Escherichia coli lacZ gene are ubiquitously employed in biological research. However, we were not able to readily identify a quantitative method that worked reliably with yeast (Saccharomyces cerevisiae) cells and that was compatible with high-throughput screening and robotic liquid handling tools. We have therefore adapted a commercially available assay employing a 6-O-beta-galactopyranosyl-luciferin substrate to provide the required sensitivity with minimal sample handling times. Our assay uses only one-tenth of the reagents suggested by the reagent manufacturer (Promega) for equivalent assays with mammalian cell cultures and produces rapid, sensitive and reproducible analysis with as little as 1 microl yeast cell culture and with < 100 cells. We demonstrate that the assay is compatible with yeast strains generated by the systematic yeast deletion project and functions equally well with genomically integrated or plasmid-encoded lacZ reporters and with cells grown in complex or defined media. The high-sensitivity, miniaturized format reduced sample handling required will make this assay useful for a wide range of applications.

Costello syndrome (CS) is a disorder caused by HRAS gene mutations. Substitutions in glycine residues at positions 12 or 13 of the HRAS gene lead to aberrant activation of the guanine-nucleotide binding protein, HRAS. Homozygous expression of mutant HRAS predisposes pediatric patients to early onset of cancer, especially embryonal rhabdomyosarcoma (ERMS), while heterozygosity at this locus results in non-malignant abnormalities. Since the mid 1970’s there has been only a modest increase in the 5-year survival rate for ERMS in the United States to no more than 70%, so new treatment strategies are needed. To address this need, we have acquired, propagated, and banked patient-derived ERMS cells harboring a homozygous p.G12A substitution and have used these cells to develop a high-throughput screening (HTS) assay to discover compounds that selectively kill cells bearing homozygous HRAS mutation. The assay employs luminescent detection of ATP using CellTiter-Glo to quantify cell viability. Cell number, growth conditions, and compound delivery were optimized to obtain a robust 72-hr cell growth assay. The potent cytotoxic agent dactinomycin was used to ensure reproducible sensitivity of the assay. The assay was validated for HTS by pilot screening of a library of 2,000 known drugs and bioactive natural products, and hits were confirmed by retesting in dose-response. Confirmed active compounds were tested for selectivity for ERMS harboring homozygous mutant HRAS in a panel of patient-derived and purchased cell lines: RD, an ERMS line not driven by HRAS mutation; RH36, a mutant HRAS (p.Q61K) ERMS line that is not derived from CS; fibroblasts derived from CS patients with heterozygous HRAS mutations (CS242, p.G12A and CS066, p.G12S), and normal fibroblasts. In future work, compounds identified by HTS as displaying selectivity for homozygous mutant HRAS-expressing ERMS cells will be evaluated in pairwise combinations and also in combination with currently used chemotherapeutics. Targeting mutant HRAS-activated cell growth could direct the design of a novel therapeutic strategy for treatment of children with ERMS. Citation Format: Donna M. Cartledge, Katherine M. Drake, Karen W. Gripp, Katia Sol-Church, E. Anders Kolb, Andrew D. Napper. High-throughput screen to identify compounds targeted to embryonal rhabdomyosarcoma harboring mutant HRAS using cells derived from a pediatric Costello syndrome patient. [abstract]. In: Proceedings of the 104th Annual Meeting of the American Association for Cancer Research; 2013 Apr 6-10; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2013;73(8 Suppl):Abstract nr 4547. doi:10.1158/1538-7445.AM2013-4547

MLL-rearranged leukemia is an aggressive form of cancer characterized by specific chromosomal rearrangements of the mixed lineage leukemia (MLL) gene. In normal cells, MLL, a histone methyltransferase, methylates the lysine 4 residue on histone 3 (H3K4) and thereby regulates the expression of genes which control hematopoietic development. In leukemic cells, chromosomal rearrangements fuse the N-terminus of MLL with one of more than 50 partner proteins, giving rise to oncogenic fusion proteins that lack H3K4 methyltransferase activity. These fusion proteins inappropriately recruit DOT1L, which methylates H3K79, producing an abnormal methylation pattern and activation of genes normally suppressed during differentiation. Elevated expression of HOXA9 in particular leads to a block in normal hematopoetic differentiation and the onset of leukemia. A solid body of evidence has shown DOT1L to be crucial for the development and maintenance of MLL-rearranged leukemias; consequently, it is a promising target for therapeutic intervention. We have focused on developing an assay suitable for high-throughput screening (HTS) to detect compounds that inhibit DOT1L enzymatic activity. DOT1L catalyzes the transfer of a methyl group from S-adenosylmethionine (SAM) to H3K79 and generates S-adenosylhomocysteine (SAH) as a byproduct. The enzyme displays a very low rate of methylation of histone octamers or nucleosomes and no activity towards individual histones or histone-derived peptides. Thus DOT1L HTS requires a highly sensitive assay that is able to accommodate protein-protein and protein-DNA complexes as substrates. An assay using radiolabeled SAM has sufficient sensitivity, but is relatively laborious and not well-suited for HTS. We tested several previously reported non-radioactive assay formats, but none enabled robust detection of DOT1L activity. Therefore, we developed a coupled enzyme assay in which SAH is converted to homocysteine and adenosine using SAH hydrolase. In a second enzyme-coupled step, adenosine kinase generates AMP and ADP from the SAH-derived adenosine and added ATP. Finally, ADP is detected using the luciferase-based Promega ADP-Glo assay. The luminescent signal provides a highly sensitive measure of ADP at a concentration as low as 10 nM. The readout is directly proportional to SAH production; thus DOT1L inhibitors are identified by a reduction in this signal. We report development of the assay, including careful optimization of the activity of adenosine kinase to enhance sensitivity, and assay validation demonstrating DOT1L inhibition by the non-selective SAM-competitive inhibitor sinefungin, and pilot HTS of a 2,000-compound library. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 103rd Annual Meeting of the American Association for Cancer Research; 2012 Mar 31-Apr 4; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2012;72(8 Suppl):Abstract nr 3878. doi:1538-7445.AM2012-3878

Chromosomal translocations of the mixed lineage leukemia (MLL) gene occur in 5-10% of pediatric acute lymphoblastic leukemia (ALL) and acute myeloid leukemia (AML), including up to 80% of infant ALL and most treatment-related secondary AML. These MLL rearrangements result in a portion of the MLL protein being fused to one of a number of transcription factors. The prognosis for MLL-rearranged (MLL-R) leukemia is much poorer than for leukemia devoid of MLL fusions; five-year survival of Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 102nd Annual Meeting of the American Association for Cancer Research; 2011 Apr 2-6; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2011;71(8 Suppl):Abstract nr 1375. doi:10.1158/1538-7445.AM2011-1375