Search

This paper investigates how to leverage a DurIAN-based average model to enable a new speaker to have both accurate pronunciation and fluent cross-lingual speaking with very limited monolingual data. A weakness of the recently proposed end-to-end text-to-speech (TTS) systems is that robust alignment is hard to achieve, which hinders it to scale well with very limited data. To cope with this issue, we introduce AdaDurIAN by training an improved DurIAN-based average model and leverage it to few-shot learning with the shared speaker-independent content encoder across different speakers. Several few-shot learning tasks in our experiments show AdaDurIAN can outperform the baseline end-to-end system by a large margin. Subjective evaluations also show that AdaDurIAN yields higher mean opinion score (MOS) of naturalness and more preferences of speaker similarity. In addition, we also apply AdaDurIAN to emotion transfer tasks and demonstrate its promising performance., Comment: Submitted to InterSpeech 2020

Background Aberrant expression of the RON receptor tyrosine kinase is a pathogenic feature and a validated drug target in various types of cancers. Currently, therapeutic antibodies targeting RON for cancer therapy are under intensive evaluation. Here we report the development and validation of a novel humanized anti-RON antibody-drug conjugate for cancer therapy. Methods Antibody humanization was achieved by grafting sequences of complementarity-determining regions from mouse monoclonal antibody Zt/g4 into human IgG1/κ acceptor frameworks. The selected humanized Zt/g4 subclone H1L3 was conjugated with monomethyl auristatin E using a dipeptide linker to form H-Zt/g4-MMAE. Pharmacokinetic analysis of H-Zt/g4-MMAE was determined using hydrophobic interaction chromatography and a MMAE ADC ELISA kit. Biochemical and biological assays were used for measuring RON expression, internalization, cell viability and death. Therapeutic efficacies of H-Zt/g4-MMAE were validated in vivo using three pancreatic cancer xenograft models. Toxicological activities of H-Zt/g4-MMAE were determined in mouse and cynomolgus monkey. Results H-Zt/g4-MMAE had a drug to antibody ratio of 3.77:1 and was highly stable in human plasma with a dissociation rate less than 5% within a 20 day period. H-Zt/g4-MMAE displayed a favorable pharmacokinetic profile in both mouse and cynomolgus monkey. In vitro, H-Zt/g4-MMAE induced RON internalization, which results in killing of pancreatic cancer cells with IC50 values at 10–20 nM. In vivo, H-Zt/g4-MMAE inhibited pancreatic cancer xenograft growth with tumoristatic concentrations at 1~3 mg/kg bodyweight. Significantly, H-Zt/g4-MMAE eradicated tumors across multiple xenograft models regardless their chemoresistant and metastatic statuses. Moreover, H-Zt/g4-MMAE inhibited and eradicated xenografts mediated by pancreatic cancer stem-like cells and by primary cells from patient-derived tumors. Toxicologically, H-Zt/g4-MMAE is well tolerated in mice up to 60 mg/kg. In cynomolgus monkey, H-Zt/g4-MMAE up to 30 mg/kg had a manageable and reversible toxicity profile. Conclusions H-Zt/g4-MMAE is superior in eradication of pancreatic cancer xenografts with favorable pharmacokinetic profiles and manageable toxicological activities. These findings warrant the transition of H-Zt/g4-MMAE into clinical trials in the future. Electronic supplementary material The online version of this article (10.1186/s40425-019-0525-0) contains supplementary material, which is available to authorized users.

Tables S1. Pathological and Biological Features of Primary PDAC Cell Lines from Patient-Derived Xenograft Tumors*. Table S2. Adverse Effects of H-Zt/g4-MMAE on blood leukocyte and erythrocytes in Cynomolgus monkey. Table S3. Effect of H-Zt/g4-MMAE in vivo on various enzymatic activities in blood samples collected from cynomolgus monkeys. (PDF 663 kb)

Figure S1. Immune reactivity of H-Zt/g4 to RON in multiple tissues from human and cynomolgus monkey. Figure S2. Effect of H-Zt/g4-MMAE on eradication of RON-expressing cells in PDAC xenograft tumors. Figure S3. Induction in vitro by H-Zt/g4-MMAE of death of PDAC stem-like cells. Figure S4. Effect of H-Zt/g4-MMAE on elimination of RON-expressing cells in PSC + 24/44/ESA cell-mediated xenografts. Figure S5A. Cell surface RON expression by a panel of primary PDX cell lines. Figure S5B. Induction of cell surface RON internalization by H-Zt/g4. Figure S5C. Effect of H-Zt/g4-MMAE on viability of primary PDAC cell lines derived from patient’ tumors. Figure S6. Histological examination of multiple organs and/or tissues from cynomolgus monkey treated with H-Zt/g4-MMAE. Figure S7. Histological examination of multiple organs and/or tissues from cynomolgus monkey treated with H-Zt/g4-MMAE. (PDF 2315 kb)