Self-organizing maps with variable neighborhoods facilitate learning of chromatin accessibility signal shapes associated with regulatory elements.

BACKGROUND: Assigning chromatin states genome-wide (e.g. promoters, enhancers, etc.) is commonly performed to improve functional interpretation of these states. However, computational methods to assign chromatin state suffer from the following drawbacks: they typically require data from multiple assays, which may not be practically feasible to obtain, and they depend on peak calling algorithms, which require careful parameterization and often exclude the majority of the genome. To address these drawbacks, we propose a novel learning technique built upon the Self-Organizing Map (SOM), Self-Organizing Map with Variable Neighborhoods (SOM-VN), to learn a set of representative shapes from a single, genome-wide, chromatin accessibility dataset to associate with a chromatin state assignment in which a particular RE is prevalent. These shapes can then be used to assign chromatin state using our workflow. RESULTS: We validate the performance of the SOM-VN workflow on 14 different samples of varying quality, namely one assay each of A549 and GM12878 cell lines and two each of H1 and HeLa cell lines, primary B-cells, and brain, heart, and stomach tissue. We show that SOM-VN learns shapes that are (1) non-random, (2) associated with known chromatin states, (3) generalizable across sets of chromosomes, and (4) associated with magnitude and multimodality. We compare the accuracy of SOM-VN chromatin states against the Clustering Aggregation Tool (CAGT), an unsupervised method that learns chromatin accessibility signal shapes but does not associate these shapes with REs, and we show that overall precision and recall is increased when learning shapes using SOM-VN as compared to CAGT. We further compare enhancer state assignments from SOM-VN in signals above a set threshold to enhancer state assignments from Predicting Enhancers from ATAC-seq Data (PEAS), a deep learning method that assigns enhancer chromatin states to peaks. We show that the precision-recall area under the curve for the assignment of enhancer states is comparable to PEAS. CONCLUSIONS: Our work shows that the SOM-VN workflow can learn relationships between REs and chromatin accessibility signal shape, which is an important step toward the goal of assigning and comparing enhancer state across multiple experiments and phenotypic states.

Vpu is the main determinant for tetraspanin downregulation in HIV-1-infected cells.

UNLABELLED: Tetraspanins constitute a family of cellular proteins that organize various membrane-based processes. Several members of this family, including CD81, are actively recruited by HIV-1 Gag to viral assembly and release sites. Despite their enrichment at viral exit sites, the overall levels of tetraspanins are decreased in HIV-1-infected cells. Here, we identify Vpu as the main viral determinant for tetraspanin downregulation. We also show that reduction of CD81 levels by Vpu is not a by-product of CD4 or BST-2/tetherin elimination from the surfaces of infected cells and likely occurs through an interaction between Vpu and CD81. Finally, we document that Vpu-mediated downregulation of CD81 from the surfaces of infected T cells can contribute to preserving the infectiousness of viral particles, thus revealing a novel Vpu function that promotes virus propagation by modulating the host cell environment. IMPORTANCE: The HIV-1 accessory protein Vpu has previously been shown to downregulate various host cell factors, thus helping the virus to overcome restriction barriers, evade immune attack, and maintain the infectivity of viral particles. Our study identifies tetraspanins as an additional group of host factors whose expression at the surfaces of infected cells is lowered by Vpu. While the downregulation of these integral membrane proteins, including CD81 and CD82, likely affects more than one function of HIV-1-infected cells, we document that Vpu-mediated lowering of CD81 levels in viral particles can be critical to maintaining their infectiousness.

Ezrin is a component of the HIV-1 virological presynapse and contributes to the inhibition of cell-cell fusion.

UNLABELLED: During cell-to-cell transmission of HIV-1, viral and cellular proteins transiently accumulate at the contact zone between infected (producer) and uninfected (target) cells, forming the virological synapse. Rearrangements of the cytoskeleton in producer and target cells are required for proper targeting of viral and cellular components during synapse formation, yet little is known about how these processes are regulated, particularly within the producer cell. Since ezrin-radixin-moesin (ERM) proteins connect F-actin with integral and peripheral membrane proteins, are incorporated into virions, and interact with cellular components of the virological presynapse, we hypothesized that they play roles during the late stage of HIV-1 replication. Here we document that phosphorylated (i.e., active) ezrin specifically accumulates at the HIV-1 presynapse in T cell lines and primary CD4(+) lymphocytes. To investigate whether ezrin supports virus transmission, we sought to ablate ezrin expression in producer cells. While cells did not tolerate a complete knockdown of ezrin, even a modest reduction of ezrin expression (~50%) in HIV-1-producing cells led to the release of particles with impaired infectivity. Further, when cocultured with uninfected target cells, ezrin-knockdown producer cells displayed reduced accumulation of the tetraspanin CD81 at the synapse and fused more readily with target cells, thus forming syncytia. Such an outcome likely is not optimal for virus dissemination, as evidenced by the fact that, in vivo, only relatively few infected cells form syncytia. Thus, ezrin likely helps secure efficient virus spread not only by enhancing virion infectivity but also by preventing excessive membrane fusion at the virological synapse. IMPORTANCE: While viruses, in principal, can propagate through successions of syncytia, HIV-1-infected cells in the majority of cases do not fuse with potential target cells during viral transmission. This mode of spread is coresponsible for key features of HIV-1 pathogenesis, including killing of bystander cells and establishment of latently infected T lymphocytes. Here we identify the ERM protein family member ezrin as a cellular factor that contributes to the inhibition of cell-cell fusion and thus to suppressing excessive syncytium formation. Our analyses further suggest that ezrin, which connects integral membrane proteins with actin, functions in concert with CD81, a member of the tetraspanin family of proteins. Additional evidence, documented here and elsewhere, suggests that ezrin and CD81 cooperate to prevent cytoskeleton rearrangements that need to take place during the fusion of cellular membranes.

Clustering and mobility of HIV-1 Env at viral assembly sites predict its propensity to induce cell-cell fusion.

HIV-1 Env mediates virus attachment to and fusion with target cell membranes, and yet, while Env is still situated at the plasma membrane of the producer cell and before its incorporation into newly formed particles, Env already interacts with the viral receptor CD4 on target cells, thus enabling the formation of transient cell contacts that facilitate the transmission of viral particles. During this first encounter with the receptor, Env must not induce membrane fusion, as this would prevent the producer cell and the target cell from separating upon virus transmission, but how Env's fusion activity is controlled remains unclear. To gain a better understanding of the Env regulation that precedes viral transmission, we examined the nanoscale organization of Env at the surface of producer cells. Utilizing superresolution microscopy (stochastic optical reconstruction microscopy [STORM]) and fluorescence recovery after photobleaching (FRAP), we quantitatively assessed the clustering and dynamics of Env upon its arrival at the plasma membrane. We found that Gag assembly induced the aggregation of small Env clusters into larger domains and that these domains were completely immobile. Truncation of the cytoplasmic tail (CT) of Env abrogated Gag's ability to induce Env clustering and restored Env mobility at assembly sites, both of which correlated with increased Env-induced fusion of infected and uninfected cells. Hence, while Env trapping by Gag secures Env incorporation into viral particles, Env clustering and its sequestration at assembly sites likely also leads to the repression of its fusion function, and thus, by preventing the formation of syncytia, Gag helps to secure efficient transfer of viral particles to target cells.

Rous sarcoma virus gag has no specific requirement for phosphatidylinositol-(4,5)-bisphosphate for plasma membrane association in vivo or for liposome interaction in vitro.

The MA domain of the retroviral Gag protein mediates interactions with the plasma membrane, which is the site of productive virus release. HIV-1 MA has a phosphatidylinositol-(4,5)-bisphosphate [PI(4,5)P2] binding pocket; depletion of this phospholipid from the plasma membrane compromises Gag membrane association and virus budding. We used multiple methods to examine the possible role of PI(4,5)P2 in Gag-membrane interaction of the alpharetrovirus Rous sarcoma virus (RSV). In contrast to HIV-1, which was tested in parallel, neither membrane localization of RSV Gag-GFP nor release of virus-like particles was affected by phosphatase-mediated depletion of PI(4,5)P2 in transfected avian cells. In liposome flotation experiments, RSV Gag required acidic lipids for binding but showed no specificity for PI(4,5)P2. Mono-, di-, and triphosphorylated phosphatidylinositol phosphate (PIP) species as well as high concentrations of phosphatidylserine (PS) supported similar levels of flotation. A mutation that increases the overall charge of RSV MA also enhanced Gag membrane binding. Contrary to previous reports, we found that high concentrations of PS, in the absence of PIPs, also strongly promoted HIV-1 Gag flotation. Taken together, we interpret these results to mean that RSV Gag membrane association is driven by electrostatic interactions and not by any specific association with PI(4,5)P2.

NEIL3 Repairs Telomere Damage during S Phase to Secure Chromosome Segregation at Mitosis.

Oxidative damage to telomere DNA compromises telomere integrity. We recently reported that the DNA glycosylase NEIL3 preferentially repairs oxidative lesions in telomere sequences in vitro. Here, we show that loss of NEIL3 causes anaphase DNA bridging because of telomere dysfunction. NEIL3 expression increases during S phase and reaches maximal levels in late S/G2. NEIL3 co-localizes with TRF2 and associates with telomeres during S phase, and this association increases upon oxidative stress. Mechanistic studies reveal that NEIL3 binds to single-stranded DNA via its intrinsically disordered C terminus in a telomere-sequence-independent manner. Moreover, NEIL3 is recruited to telomeres through its interaction with TRF1, and this interaction enhances the enzymatic activity of purified NEIL3. Finally, we show that NEIL3 interacts with AP Endonuclease 1 (APE1) and the long-patch base excision repair proteins PCNA and FEN1. Taken together, we propose that NEIL3 protects genome stability through targeted repair of oxidative damage in telomeres during S/G2 phase.