Summary Co-transcriptional histone methylations by Set1 and Set2 have been shown to affect histone acetylation at promoters and 3′ regions of genes, respectively. While histone H3K4 trimethylation (H3K4me3) is thought to promote nucleosome acetylation and remodeling near promoters, we show here that H3K4 dimethylation (H3K4me2) by Set1 leads to reduced histone acetylation levels near 5′ ends of genes. H3K4me2 recruits the Set3 complex via the Set3 PHD finger, localizing the Hos2 and Hst1 subunits to deacetylate histones in 5′ transcribed regions. Cells lacking the Set1-Set3 complex pathway are sensitive to mycophenolic acid and have reduced polymerase levels at a Set3 target gene, suggesting a positive role in transcription. We propose that Set1 establishes two distinct chromatin zones on genes: H3K4me3 leads to high levels of acetylation and low nucleosome density at promoters, while H3K4me2 just downstream recruits the Set3 complex to suppress nucleosome acetylation and remodeling.

The discriminative power of modern deep learning models for 3D human action recognition is growing ever so potent. In conjunction with the recent resurgence of 3D human action representation with 3D skeletons, the quality and the pace of recent progress have been significant. However, the inner workings of state-of-the-art learning based methods in 3D human action recognition still remain mostly black-box. In this work, we propose to use a new class of models known as Temporal Convolutional Neural Networks (TCN) for 3D human action recognition. Compared to popular LSTM-based Recurrent Neural Network models, given interpretable input such as 3D skeletons, TCN provides us a way to explicitly learn readily interpretable spatio-temporal representations for 3D human action recognition. We provide our strategy in re-designing the TCN with interpretability in mind and how such characteristics of the model is leveraged to construct a powerful 3D activity recognition method. Through this work, we wish to take a step towards a spatio-temporal model that is easier to understand, explain and interpret. The resulting model, Res-TCN, achieves state-of-the-art results on the largest 3D human action recognition dataset, NTU-RGBD.

Summary Packaging of DNA into chromatin has a profound impact on gene expression. To understand how changes in chromatin influence transcription, we analyzed 165 mutants of chromatin machinery components in Saccharomyces cerevisiae. mRNA expression patterns change in 80% of mutants, always with specific effects, even for loss of widespread histone marks. The data is assembled into a network of chromatin interaction pathways. The network is function-based, has a branched, interconnected topology and lacks strict one-to-one relationships between complexes. Chromatin pathways are not separate entities for different gene sets, but share many components. The study evaluates which interactions are important for which genes and predicts new interactions, for example between Paf1C and Set3C, as well as a role for Mediator in subtelomeric silencing. The results indicate the presence of gene-dependent effects that go beyond context-dependent binding of chromatin factors and provide a framework for understanding how specificity is achieved through regulating chromatin.

Set3 complex (Set3C) binds histone H3 dimethylated at lysine 4 (H3K4me2) to mediate deacetylation of histones in 5′ transcribed regions. To discern how Set3C affects gene expression, genome-wide transcription was analyzed in yeast undergoing a series of carbon source shifts. Deleting SET3 primarily caused changes during transition periods, as genes were induced or repressed. Surprisingly, a majority of Set3-affected genes are overlapped by non-coding RNA (ncRNA) transcription. Many Set3-repressed genes have H3K4me2 instead of me3 over promoter regions, due either to reduced H3K4me3 or ncRNA transcription from distal or anti-sense promoters. The resulting deacetylation by Set3C slows gene induction. Set3C represses internal cryptic promoters, but in different regions of genes than the Set2/Rpd3S pathway. Finally, Set3C stimulates some genes by repressing an overlapping antagonistic anti-sense transcript. These results suggest that Set3C and overlapping non-coding transcription can cooperate to fine tune the response kinetics of individual genes.

Summary Various factors differentially recognize trimethylated histone H3 lysine 4 (H3K4me3) near promoters, H3K4me2 just downstream, and promoter-distal H3K4me1 to modulate gene expression. This methylation “gradient” is thought to result from preferential binding of the H3K4 methyltransferase Set1/COMPASS to promoter-proximal RNA polymerase II. However, other studies have suggested that location-specific cues allosterically activate Set1. ChIP-Seq experiments show that H3K4 methylation patterns on active genes are not universal or fixed, and change in response to both transcription elongation rate and frequency, as well as reduced COMPASS activity. Fusing Set1 to RNA polymerase II results in H3K4me2 throughout transcribed regions, and similarly extended H3K4me3 on highly transcribed genes. Tethered Set1 still requires histone H2B ubiquitylation for activity. These results show higher-level methylations reflect not only Set1/COMPASS recruitment, but also multiple rounds of transcription. This model provides a simple explanation for non-canonical methylation patterns at some loci or in certain COMPASS mutants.