POLLEN WALL ABORTION 1 is essential for pollen wall development in rice.

The integrity of pollen wall structures is essential for pollen development and maturity in rice (Oryza sativa L.). In this study, we isolated and characterized the rice male-sterile mutant pollen wall abortion 1 (pwa1), which exhibits a defective pollen wall (DPW) structure and has sterile pollen. Map-based cloning, genetic complementation, and gene knockout experiments revealed that PWA1 corresponds to the gene LOC_Os01g55094 encoding a coiled-coil domain-containing protein. PWA1 localized to the nucleus, and PWA1 was expressed in the tapetum and microspores. PWA1 interacted with the transcription factor TAPETUM DEGENERATION RETARDATION (TDR)-INTERACTING PROTEIN2 (TIP2, also named bHLH142) in vivo and in vitro. The tip2-1 mutant, which we obtained by clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR-associated protein 9-mediated gene editing, showed delayed tapetum degradation, sterile pollen, and DPWs. We determined that TIP2/bHLH142 regulates PWA1 expression by binding to its promoter. Analysis of the phenotype of the tip2-1 pwa1 double mutant indicated that TIP2/bHLH142 functions upstream of PWA1. Further studies suggested that PWA1 has transcriptional activation activity and participates in pollen intine development through the ß-glucosidase Os12BGlu38. Therefore, we identified a sterility factor, PWA1, and uncovered a regulatory network underlying the formation of the pollen wall and mature pollen in rice.

OsCCRL1 is Essential for Phenylpropanoid Metabolism in Rice Anthers.

Phenylpropanoid metabolism and timely tapetal degradation are essential for anther and pollen development, but the underlying mechanisms are unclear. In the current study, to investigate this, we identified and analyzed the male-sterile mutant, osccrl1 (cinnamoyl coA reductase-like 1), which exhibited delayed tapetal programmed cell death (PCD) and defective mature pollen. Map-based cloning, genetic complementation, and gene knockout revealed that OsCCRL1 corresponds to the gene LOC_Os09g32020.2, a member of SDR (short-chain dehydrogenase/reductase) family enzyme. OsCCRL1 was preferentially expressed in the tapetal cells and microspores, and localized to the nucleus and cytoplasm in both rice protoplasts and Nicotiana benthamiana leaves. The osccrl1 mutant exhibited reduced CCRs enzyme activity, less lignin accumulation, delayed tapetum degradation, and disrupted phenylpropanoid metabolism. Furthermore, an R2R3 MYB transcription factor OsMYB103/OsMYB80/OsMS188/BM1, involved in tapetum and pollen development, regulates the expression of OsCCRL1. Finally, the osmyb103 osccrl1 double mutants, exhibited the same phenotype as the osmyb103 single mutant, further indicating that OsMYB103/OsMYB80/OsMS188/BM1 functions upstream of OsCCRL1. These findings help to clarify the role of phenylpropanoid metabolism in male sterility and the regulatory network underlying the tapetum degradation.

CartoonLossGAN: Learning Surface and Coloring of Images for Cartoonization.

Cartoonization as a special type of artistic style transfer is a difficult image processing task. The current existing artistic style transfer methods cannot generate satisfactory cartoon-style images due to that artistic style images often have delicate strokes and rich hierarchical color changes while cartoon-style images have smooth surfaces without obvious color changes, and sharp edges. To this end, we propose a cartoon loss based generative adversarial network (CartoonLossGAN) for cartoonization. Particularly, we first reuse the encoder part of the discriminator to build a compact generative adversarial network (GAN) based cartoonization architecture. Then we propose a novel cartoon loss function for the architecture. It can imitate the process of sketching to learn the smooth surface of the cartoon image, and imitate the coloring process to learn the coloring of the cartoon image. Furthermore, we also propose an initialization strategy, which is used in the scenario of reusing the discriminator to make our model training easier and more stable. Extensive experimental results demonstrate that our proposed CartoonLossGAN can generate fantastic cartoon-style images, and outperforms four representative methods.

VERAM: View-Enhanced Recurrent Attention Model for 3D Shape Classification.

Multi-view deep neural network is perhaps the most successful approach in 3D shape classification. However, the fusion of multi-view features based on max or average pooling lacks a view selection mechanism, limiting its application in, e.g., multi-view active object recognition by a robot. This paper presents VERAM, a view-enhanced recurrent attention model capable of actively selecting a sequence of views for highly accurate 3D shape classification. VERAM addresses an important issue commonly found in existing attention-based models, i.e., the unbalanced training of the subnetworks corresponding to next view estimation and shape classification. The classification subnetwork is easily overfitted while the view estimation one is usually poorly trained, leading to a suboptimal classification performance. This is surmounted by three essential view-enhancement strategies: 1) enhancing the information flow of gradient backpropagation for the view estimation subnetwork, 2) devising a highly informative reward function for the reinforcement training of view estimation and 3) formulating a novel loss function that explicitly circumvents view duplication. Taking grayscale image as input and AlexNet as CNN architecture, VERAM with 9 views achieves instance-level and class-level accuracy of 95.5 and 95.3 percent on ModelNet10, 93.7 and 92.1 percent on ModelNet40, both are the state-of-the-art performance under the same number of views.