Through the analysis of hundreds of full-length cDNAs from fifteen species representing all major orders of dinoflagellates, we demonstrate that nuclear-encoded mRNAs in all species, from ancestral to derived lineages, are trans-spliced with the addition of the 22-nt conserved spliced leader (SL), DCCGUAGCCAUUUUGGCUCAAG (D ‫؍‬ U, A, or G), to the 5 end. SL trans-splicing has been documented in a limited but diverse number of eukaryotes, in which this process makes it possible to translate polycistronically transcribed nuclear genes. In SL trans-splicing, SL-donor transcripts (SL RNAs) contain two functional domains: an exon that provides the SL for mRNA and an intron that contains a spliceosomal (Sm) binding site. In dinoflagellates, SL RNAs are unusually short at 50 -60 nt, with a conserved Sm binding motif (AUUUUGG) located in the SL (exon) rather than the intron. The initiation nucleotide is predominantly U or A, an unusual feature that may affect capping, and hence the translation and stability of the recipient mRNA. The core SL element was found in mRNAs coding for a diverse array of proteins. Among the transcripts characterized were three homologs of Sm-complex subunits, indicating that the role of the Sm binding site is conserved, even if the location on the SL is not. Because association with an Sm-complex often signals nuclear import for U-rich small nuclear RNAs, it is unclear how this Sm binding site remains on mature mRNAs without impeding cytosolic localization or translation of the latter.inoflagellates are unicellular eukaryotes that contribute significantly to marine primary production, coral reef growth, and marine toxins. They are members of the Alveolata, which also include ciliates and apicomplexa (1). Dinoflagellate genomes are enormous (3-200 pg of DNA per cell) and lack typical histones, with chromosomes permanently condensed, nuclear membranes remaining intact in mitosis, and mitotic spindle being extranuclear (for review see ref.2). The mechanism of gene regulation is largely unknown. Sporadic investigations have shown that a relatively small fraction of genes are under transcriptional control (3-6) and that introns are not common (7,8). The few comprehensive studies of dinoflagellate gene structures reveal genes with high copy number and arrangement in polycistronic or otherwise tandem arrays (e.g., refs. 7-10).Spliced leader (SL) trans-splicing has been found in a disjointed group of eukaryotes, in which a short RNA fragment (i.e., SL, Ϸ15-50 nt) from a small noncoding RNA (SL RNA) is transplanted to the 5Ј end of independently transcribed pre-mRNAs to yield mature mRNAs. This process converts a polycistronic transcript into translatable monocistronic mRNAs. SL trans-splicing has been well studied in Euglenozoa. It has been detected in nematodes, Platyhelminthes, cnidarians, rotifers, ascidians, and appendicularia (for review, see 11-13). SL RNA contains two functional domains: an exon (i.e., SL) that is spliced to an mRNA and an intron that contains a spliceosomal (Sm) binding site be...

Dinoflagellates are important components of marine ecosystems and essential coral symbionts, yet little is known about their genomes. We report here on the analysis of a high-quality assembly from the 1180-megabase genome of Symbiodinium kawagutii. We annotated protein-coding genes and identified Symbiodinium-specific gene families. No whole-genome duplication was observed, but instead we found active (retro)transposition and gene family expansion, especially in processes important for successful symbiosis with corals. We also documented genes potentially governing sexual reproduction and cyst formation, novel promoter elements, and a microRNA system potentially regulating gene expression in both symbiont and coral. We found biochemical complementarity between genomes of S. kawagutii and the anthozoan Acropora, indicative of host-symbiont coevolution, providing a resource for studying the molecular basis and evolution of coral symbiosis.

No abstract

BackgroundLignin materials are abundant and among the most important potential sources for biofuel production. Development of an efficient lignin degradation process has considerable potential for the production of a variety of chemicals, including bioethanol. However, lignin degradation using current methods is inefficient. Given their immense environmental adaptability and biochemical versatility, bacterial could be used as a valuable tool for the rapid degradation of lignin. Kraft lignin (KL) is a polymer by-product of the pulp and paper industry resulting from alkaline sulfide treatment of lignocellulose, and it has been widely used for lignin-related studies.ResultsBeta-proteobacterium Cupriavidus basilensis B-8 isolated from erosive bamboo slips displayed substantial KL degradation capability. With initial concentrations of 0.5–6 g L-1, at least 31.3% KL could be degraded in 7 days. The maximum degradation rate was 44.4% at the initial concentration of 2 g L-1. The optimum pH and temperature for KL degradation were 7.0 and 30°C, respectively. Manganese peroxidase (MnP) and laccase (Lac) demonstrated their greatest level of activity, 1685.3 U L-1 and 815.6 U L-1, at the third and fourth days, respectively. Many small molecule intermediates were formed during the process of KL degradation, as determined using GC-MS analysis. In order to perform metabolic reconstruction of lignin degradation in this bacterium, a draft genome sequence for C. basilensis B-8 was generated. Genomic analysis focused on the catabolic potential of this bacterium against several lignin-derived compounds. These analyses together with sequence comparisons predicted the existence of three major metabolic pathways: β-ketoadipate, phenol degradation, and gentisate pathways.ConclusionThese results confirmed the capability of C. basilensis B-8 to promote KL degradation. Whole genomic sequencing and systematic analysis of the C. basilensis B-8 genome identified degradation steps and intermediates from this bacterial-mediated KL degradation method. Our findings provide a theoretical basis for research into the mechanisms of lignin degradation as well as a practical basis for biofuel production using lignin materials.

Vertebrate ancestors appeared in a uniform, shallow water environment, but modern species flourish in highly variable niches. A striking array of phenotypes exhibited by contemporary animals is assumed to have evolved by accumulating a series of selectively advantageous mutations. However, the experimental test of such adaptive events at the molecular level is remarkably difficult. One testable phenotype, dim-light vision, is mediated by rhodopsins. Here, we engineered 11 ancestral rhodopsins and show that those in early ancestors absorbed light maximally ( max) at 500 nm, from which contemporary rhodopsins with variable maxs of 480 -525 nm evolved on at least 18 separate occasions. These highly environment-specific adaptations seem to have occurred largely by amino acid replacements at 12 sites, and most of those at the remaining 191 (Ϸ94%) sites have undergone neutral evolution. The comparison between these results and those inferred by commonly-used parsimony and Bayesian methods demonstrates that statistical tests of positive selection can be misleading without experimental support and that the molecular basis of spectral tuning in rhodopsins should be elucidated by mutagenesis analyses using ancestral pigments. molecular adaptation ͉ rhodopsin

Recent studies have shown that adversarial examples in stateof-the-art image classifiers trained by deep neural networks (DNN) can be easily generated when the target model is transparent to an attacker, known as the white-box setting. However, when attacking a deployed machine learning service, one can only acquire the input-output correspondences of the target model; this is the so-called black-box attack setting. The major drawback of existing black-box attacks is the need for excessive model queries, which may give a false sense of model robustness due to inefficient query designs. To bridge this gap, we propose a generic framework for query-efficient blackbox attacks. Our framework, AutoZOOM, which is short for Autoencoder-based Zeroth Order Optimization Method, has two novel building blocks towards efficient black-box attacks: (i) an adaptive random gradient estimation strategy to balance query counts and distortion, and (ii) an autoencoder that is either trained offline with unlabeled data or a bilinear resizing operation for attack acceleration. Experimental results suggest that, by applying AutoZOOM to a state-of-the-art black-box attack (ZOO), a significant reduction in model queries can be achieved without sacrificing the attack success rate and the visual quality of the resulting adversarial examples. In particular, when compared to the standard ZOO method, AutoZOOM can consistently reduce the mean query counts in finding successful adversarial examples (or reaching the same distortion level) by at least 93% on MNIST, CIFAR-10 and ImageNet datasets, leading to novel insights on adversarial robustness.

Recent studies have revealed the vulnerability of deep neural networks: A small adversarial perturbation that is imperceptible to human can easily make a well-trained deep neural network misclassify. This makes it unsafe to apply neural networks in security-critical applications. In this paper, we propose a new defense algorithm called Random Self-Ensemble (RSE) by combining two important concepts: randomness and ensemble. To protect a targeted model, RSE adds random noise layers to the neural network to prevent the strong gradient-based attacks, and ensembles the prediction over random noises to stabilize the performance. We show that our algorithm is equivalent to ensemble an infinite number of noisy models f without any additional memory overhead, and the proposed training procedure based on noisy stochastic gradient descent can ensure the ensemble model has a good predictive capability. Our algorithm significantly outperforms previous defense techniques on real data sets. For instance, on CIFAR-10 with VGG network (which has 92% accuracy without any attack), under the strong C&W attack within a certain distortion tolerance, the accuracy of unprotected model drops to less than 10%, the best previous defense technique has 48% accuracy, while our method still has 86% prediction accuracy under the same level of attack. Finally, our method is simple and easy to integrate into any neural network.

Polyurethanes (PUs) have many applications resulting from their preeminent properties, but being commonly used toxic catalysts, and the lack of processability for PU thermosets cause limitations. Herein, we report a new class of the PU-like dynamic covalent polymers, poly(oxime-urethanes) (POUs), which are prepared from the uncatalyzed polyaddition of multifunctional oximes and hexamethylene diisocyanate (HDI) at ambient temperature. Kinetics studies reveal that almost complete polymerization (∼99% conversion) can be achieved in 3 h at 30 °C in dichloromethane (DCM), the most effective among the solvents evaluated, producing linear POUs with comparable molecular weights to the catalyzed PUs. We find that the oxime-carbamate structures are reversible at about 100 °C through oxime-enabled transcarbamoylation via a thermally dissociative mechanism. The cross-linked POUs based on oxime-carbamate bonds show efficient catalyst-free healable/recyclable properties. Density functional theory (DFT) calculations suggest that the fast oxime-urethanation and the mild thermoreversible nature are mediated by the characteristic nitrone tautomer of the oxime. Given widespread urethane-containing materials, POUs are of promising potential in applications because of the excellent mechanical performances, facile preparation, and dynamic property without using catalysts.