The design of potent protein kinase C (PK-C) ligands with low nanomolar binding affinities was accomplished by the combined use of pharmacophore- and receptor-guided approaches based on the structure of the physiological enzyme activator, diacylglycerol (DAG). Earlier use of the former approach, which was based on the structural equivalence of DAG and phorbol ester pharmacophores, identified a fixed template for the construction of a semirigid "recognition domain" that contained the three principal pharmacophores of DAG constrained into a lactone ring (DAG-lactones). In the present work, the pharmacophore-guided approach was refined to a higher level based on the X-ray structure of the C1b domain of PK-Cdelta complexed with phorbol-13-O-acetate. A systematic search that involved modifying the DAG-lactone template with a combination of linear or branched acyl and alpha-alkylidene chains, which functioned as variable hydrophobic "affinity domains", helped identify compounds that optimized hydrophobic contacts with a group of conserved hydrophobic amino acids located on the top half of the C1 domain where the phorbol binds. The hydrophilic/hydrophobic balance of the molecules was estimated by the octanol/water partition coefficients (log P) calculated according to a fragment-based approach. The presence of branched alpha-alkylidene or acyl chains was of critical importance to reach low nanomolar binding affinities for PK-C. These branched chains appear to facilitate important van der Waals contacts with hydrophobic segments of the protein and help promote the activation of PK-C through critical membrane interactions. Molecular modeling of these DAG-lactones into an empty C1b domain using the program AutoDock 2.4 suggests the existence of competing binding modes (sn-1 and sn-2) depending on which carbonyl is directly involved in binding to the protein. Inhibition of epidermal growth factor (EGF) binding, an indirect PK-C mediated response, was realized with some DAG-lactones at a dose 10-fold higher than with the standard phorbol-12, 13-dibutyrate (PDBU). Through the National Cancer Institute (NCI) 60-cell line in vitro screen, DAG-lactone 31 was identified as a very selective and potent antitumor agent. The NCI's computerized, pattern-recognition program COMPARE, which analyzes the degree of similarity of mean-graph profiles produced by the screen, corroborated our principles of drug design by matching the profile of compound 31 with that of the non-tumor-promoting antitumor phorbol ester, prostratin. The structural simplicity and the degree of potency achieved with some of the DAG-lactones described here should dispel the myth that chemical complexity and pharmacological activity go hand in hand. Even as a racemate, DAG-lactone 31 showed low namomolar binding affinity for PK-C and displayed selective antitumor activity at equivalent nanomolar levels. Our present approach should facilitate the generation of multiple libraries of structurally similar DAG-lactones to help exploit molecular diversity for PK-C and other...

No abstract

No abstract

No abstract

Phorbol esters, the archetypical (PKC) activators, induce apoptosis in androgen-sensitive LNCaP prostate cancer cells. In this study we evaluate the effect of a novel class of PKC ligands, the diacylglycerol (DAG)-lactones, as inducers of apoptosis in LNCaP cells. These unique ligands were designed using novel pharmacophore-and receptorguided approaches to achieve highly potent DAG surrogates. Two of these compounds, HK434 and HK654, induced apoptosis in LNCaP cells with much higher potency than oleoyl-acetyl-glycerol or phorbol 12,13-dibutyrate. Moreover, different PKC isozymes were found to mediate the apoptotic effect of phorbol 12-myristate 13-acetate (PMA) and HK654 in LNCaP cells. Using PKC inhibitors and dominant negative PKC isoforms, we found that both PKC␣ and PKC␦ mediated the apoptotic effect of PMA, whereas only PKC␣ was involved in the effect of the DAG-lactone. The PKC␣ selectivity of HK654 in LNCaP cells contrasts with similar potencies in vitro for binding and activation of PKC␣ and PKC␦. Consistent with the differences in isoform dependence in intact cells, PMA and HK654 show marked differences in their abilities to translocate PKC isozymes. Both PMA and HK654 induce a marked redistribution of PKC␣ to the plasma membrane. On the other hand, unlike PMA, HK654 translocates PKC␦ predominantly to the nuclear membrane. Thus, DAG-lactones have a unique profile of activation of PKC isozymes for inducing apoptosis in LNCaP cells and represent the first example of a selective activator of a classical PKC in cellular models. An attractive hypothesis is that selective activation of PKC isozymes by pharmacological agents in cells can be achieved by differential intracellular targeting of each PKC.The phorbol esters and related diterpenes are natural compounds that have been for many years the preferred pharmacological tools for studying protein kinase C (PKC), 1 a key family of kinases implicated in growth factor-and G-proteincoupled receptor signaling. These compounds mimic the action of the lipid second messenger diacylglycerol (DAG), a relatively simple and highly flexible molecule generated by cellular phospholipases. The higher potency of phorbol esters and their greater stability compared with the second messenger DAG makes these agents the preferred activators of PKC (1, 2). Phorbol esters regulate a variety of cellular functions, including cell cycle progression, differentiation, cytoskeleton remodeling, and malignant transformation. Although phorbol esters promote mitogenesis in several cell types, accumulating data indicate that activation of PKC leads to inhibition of cell growth in many cells (3-6). Interestingly, phorbol esters induce apoptosis in several cell lines, including thymocytes, breast cancer cells, and prostate cancer cells (7-12).The heterogeneity of effects of the phorbol esters is related to the presence of multiple phorbol ester/DAG receptors, including PKC isozymes and "nonkinase" PKC receptors. The PKC family comprises at least 10 related kinases with differential expression, ...

Machine learning on (homomorphic) encrypted data is a cryptographic method for analyzing private and/or sensitive data while keeping privacy. In the training phase, it takes as input an encrypted training data and outputs an encrypted model without ever decrypting. In the prediction phase, it uses the encrypted model to predict results on new encrypted data. In each phase, no decryption key is needed, and thus the data privacy is ultimately guaranteed. It has many applications in various areas such as finance, education, genomics, and medical field that have sensitive private data. While several studies have been reported on the prediction phase, few studies have been conducted on the training phase.In this paper, we present an efficient algorithm for logistic regression on homomorphic encrypted data, and evaluate our algorithm on real financial data consisting of 422,108 samples over 200 features. Our experiment shows that an encrypted model with a sufficient Kolmogorov Smirnow statistic value can be obtained in ∼17 hours in a single machine. We also evaluate our algorithm on the public MNIST dataset, and it takes ∼2 hours to learn an encrypted model with 96.4% accuracy. Considering the inefficiency of homomorphic encryption, our result is encouraging and demonstrates the practical feasibility of the logistic regression training on large encrypted data, for the first time to the best of our knowledge.

No abstract

Cardiovascular disease and osteoporosis are thought to share common risk factors, and metabolic syndrome (MS) is composed of major risk factors for cardiovascular disease. This study was performed to investigate the relationships between specific MS components and bone mineral density (BMD). BMD was measured at the femoral neck of Korean men aged 40 years or more (n = 1,780) and postmenopausal women (n = 1,108) using dual-energy X-ray absorptiometry. We identified subjects with MS as defined by two criteria, International Diabetes Federation (IDF) and American Heart Association/National Heart, Lung, and Blood Institute (AHA/NHLBI). Body fat and lean mass were measured via bioimpedance analysis. The prevalence of MS was 19.8% and 7.7% in men and 20.8% and 11.6% in postmenopausal women according to the AHA/NHLBI definition and the IDF definition, respectively. After multivariate adjustment, femoral neck BMD was significantly lower in subjects with MS regardless of diagnostic criteria. BMD decreased as the number of MS components increased (P < 0.001 for trends in both sexes). Among MS components, waist circumference was the most important factor in this negative association. When multiple linear regression models were applied to each 5-kg weight stratum to test for a linear trend, waist circumference and fat mass were negatively associated with BMD and lean mass was positively associated with BMD in men but not in women. MS was associated with a lower BMD in Korean men and postmenopausal women, suggesting that visceral fat may lead to bone loss, especially in men.