In recent years, Microcystis aeruginosa blooms have occurred throughout the world, causing huge economic losses and destroying aquatic ecosystems. It is necessary to develop effective and ecofriendly methods to control M. aeruginosa blooms. Here, we report a high algicidal activity of prodigiosin (PG) against M. aeruginosa as well as the algicidal mechanism. PG showed high algicidal activity against M. aeruginosa, with a 50% lethal dose (LD50) of 5.87 μg/mL in 72 h. A combination of methods, including propidium iodide and Annexin V-fluorescein staining assays and light and electron microscopy indicated the existence of two modes of cell death with features similar to those in eukaryotic programmed cell death: necrotic-like and apoptotic-like. Biochemical and physiological analyses showed that PG generates reactive oxygen species (ROS), which induce lipid peroxidation, damage the membrane system and destroy the function of the photosystem. A proteomics analysis revealed that many proteins were differentially expressed in response to PG stress and that most of these proteins were involved in important metabolic processes, which may trigger necrotic-like or apoptotic-like cell death. The present study sheds light on the multiple toxicity mechanisms of PG on M. aeruginosa and its potential for controlling the occurrence of M. aeruginosa blooms in lakes.

blooms occur frequently in the Yangtze River estuary and the adjacent East China Sea. These blooms have damaged marine ecosystems and caused enormous economic losses over the past 2 decades. Thus, highly efficient, low-cost, ecofriendly approaches must be developed to control blooms. In this study, a bacterial strain (strain Y42) was identified as sp. and was used to lyse The supernatant of the strain Y42 culture was able to lyse, and the algicidal activity of this Y42 supernatant was stable with different temperatures and durations of light exposure and over a wide pH range. In addition to , Y42 showed high algicidal activity against, , and, suggesting that it targets primarily Pyrrophyta. To clarify the algicidal effects of Y42, we assessed algal lysis and determined the chlorophyll contents, photosynthetic activity, and malondialdehyde contents of after exposure to the Y42 supernatant. Scanning electron microscopy and transmission electron microscopy analyses showed that the Y42 supernatant disrupted membrane integrity and caused algal cell breakage at the megacytic zone. Photosynthetic pigment loss and significant declines in both photosynthetic efficiency and the electron transport rate indicated that the Y42 supernatant damaged the photosynthetic system of Malondialdehyde overproduction indicated that the Y42 supernatant caused lipid peroxidation and oxidative damage to membrane systems in the algal cell, ultimately leading to death. The findings of this study reveal the potential of Y42 to remove algal cells from blooms. is one of the most common dinoflagellate species that form harmful algal blooms, which frequently cause serious ecological pollution and pose health hazards to humans and other animals. Screening for bacteria with high algicidal activity against and studying their algicidal processes and characteristics will contribute to an understanding of their algicidal effects and provide a theoretical basis for preventing algal blooms and reducing their harm to the environment. This study reports the algicidal activity and characteristics of against The stability of the algicidal activity of in different environments (including different temperature, pH, and sunlight conditions) indicates its potential for use in the control of blooms.

Pre-trained language models have achieved state-of-the-art results in various Natural Language Processing (NLP) tasks. has shown that scaling up pre-trained language models can further exploit their enormous potential. A unified framework named ERNIE 3.0 [2] was recently proposed for pre-training large-scale knowledge enhanced models and trained a model with 10 billion parameters. ERNIE 3.0 outperformed the state-of-the-art models on various NLP tasks. In order to explore the performance of scaling up ERNIE 3.0, we train a hundred-billion-parameter model called ERNIE 3.0 Titan with up to 260 billion parameters on the PaddlePaddle [3] platform. Furthermore, we design a self-supervised adversarial loss and a controllable language modeling loss to make ERNIE 3.0 Titan generate credible and controllable texts. To reduce the computation overhead and carbon emission, we propose an online distillation framework for ERNIE 3.0 Titan, where the teacher model will teach students and train itself simultaneously. ERNIE 3.0 Titan is the largest Chinese dense pre-trained model so far. Empirical results show that the ERNIE 3.0 Titan outperforms the state-of-the-art models on 68 NLP datasets.

No abstract

No abstract