Since it was recently reported that an antibody for proprotein convertase subtilisin/kexin type 9 (PCSK9) reduces the risk of cardiovascular events in a clinical context, PCSK9 inhibition is thought to be an attractive therapy for dyslipidemia. In the present study, we created a novel small biologic alternative to PCSK9 antibodies called DS-9001a, comprising an albumin binding domain fused to an artificial lipocalin mutein (ABD-fused Anticalin protein), which can be produced by a microbial production system. DS-9001a strongly interfered with PCSK9 binding to low-density-lipoprotein receptor (LDL-R) and PCSK9-mediated degradation of LDL-R. In cynomolgus monkeys, single DS-9001a administration significantly reduced the serum LDL-C level up to 21 days (62.4% reduction at the maximum). Moreover, DS-9001a reduced plasma non-high-density-lipoprotein cholesterol and oxidized LDL levels, and their further reductions were observed when atorvastatin and DS-9001a were administered in combination in human cholesteryl ester transfer protein/ApoB double transgenic mice. Additionally, their reductions on the combination of atorvastatin and DS-9001a were more pronounced than those on the combination of atorvastatin and anacetrapib. Besides its favorable pharmacologic profile, DS-9001a has a lower molecular weight (about 22 kDa), yielding a high stoichiometric drug concentration that might result in a smaller administration volume than that in existing antibody therapy. Since bacterial production systems are viewed as more suited to mass production at low cost, DS-9001a may provide a new therapeutic option to treat patients with dyslipidemia. In addition, considering the growing demand for antibody-like drugs, ABD-fused Anticalin proteins could represent a promising new class of small biologic molecules.

SUMMARYA novel online speaker clustering method based on a generative model is proposed. It employs an incremental variant of variational Bayesian learning and provides probabilistic (non-deterministic) decisions for each input utterance, on the basis of the history of preceding utterances. It can be expected to be robust against errors in cluster estimation and the classification of utterances, and hence to be applicable to many real-time applications. Experimental results show that it produces 50% fewer classification errors than does a conventional online method. They also show that it is possible to reduce the number of speech recognition errors by combining the method with unsupervised speaker adaptation.

No abstract