CellNeighborEX: deciphering neighbor-dependent gene expression from spatial transcriptomics data.

Cells have evolved their communication methods to sense their microenvironments and send biological signals. In addition to communication using ligands and receptors, cells use diverse channels including gap junctions to communicate with their immediate neighbors. Current approaches, however, cannot effectively capture the influence of various microenvironments. Here, we propose a novel approach to investigate cell neighbor-dependent gene expression (CellNeighborEX) in spatial transcriptomics (ST) data. To categorize cells based on their microenvironment, CellNeighborEX uses direct cell location or the mixture of transcriptome from multiple cells depending on ST technologies. For each cell type, CellNeighborEX identifies diverse gene sets associated with partnering cell types, providing further insight. We found that cells express different genes depending on their neighboring cell types in various tissues including mouse embryos, brain, and liver cancer. Those genes are associated with critical biological processes such as development or metastases. We further validated that gene expression is induced by neighboring partners via spatial visualization. The neighbor-dependent gene expression suggests new potential genes involved in cell-cell interactions beyond what ligand-receptor co-expression can discover.

Vesalius: high-resolution in silico anatomization of spatial transcriptomic data using image analysis.

Characterization of tissue architecture promises to deliver insights into development, cell communication, and disease. In silico spatial domain retrieval methods have been developed for spatial transcriptomics (ST) data assuming transcriptional similarity of neighboring barcodes. However, domain retrieval approaches with this assumption cannot work in complex tissues composed of multiple cell types. This task becomes especially challenging in cellular resolution ST methods. We developed Vesalius to decipher tissue anatomy from ST data by applying image processing technology. Vesalius uniquely detected territories composed of multiple cell types and successfully recovered tissue structures in high-resolution ST data including in mouse brain, embryo, liver, and colon. Utilizing this tissue architecture, Vesalius identified tissue morphology-specific gene expression and regional specific gene expression changes for astrocytes, interneuron, oligodendrocytes, and entorhinal cells in the mouse brain.

A cis-acting mechanism mediates transcriptional memory at Polycomb target genes in mammals.

Epigenetic inheritance of gene expression states enables a single genome to maintain distinct cellular identities. How histone modifications contribute to this process remains unclear. Using global chromatin perturbations and local, time-controlled modulation of transcription, we establish the existence of epigenetic memory of transcriptional activation for genes that can be silenced by the Polycomb group. This property emerges during cell differentiation and allows genes to be stably switched after a transient transcriptional stimulus. This transcriptional memory state at Polycomb targets operates in cis; however, rather than relying solely on read-and-write propagation of histone modifications, the memory is also linked to the strength of activating inputs opposing Polycomb proteins, and therefore varies with the cellular context. Our data and computational simulations suggest a model whereby transcriptional memory arises from double-negative feedback between Polycomb-mediated silencing and active transcription. Transcriptional memory at Polycomb targets thus depends not only on histone modifications but also on the gene-regulatory network and underlying identity of a cell.

Hybrid protein assembly-histone modification mechanism for PRC2-based epigenetic switching and memory.

The histone modification H3K27me3 plays a central role in Polycomb-mediated epigenetic silencing. H3K27me3 recruits and allosterically activates Polycomb Repressive Complex 2 (PRC2), which adds this modification to nearby histones, providing a read/write mechanism for inheritance through DNA replication. However, for some PRC2 targets, a purely histone-based system for epigenetic inheritance may be insufficient. We address this issue at the Polycomb target FLOWERING LOCUS C (FLC) in Arabidopsis thaliana, as a narrow nucleation region of only ~three nucleosomes within FLC mediates epigenetic state switching and subsequent memory over many cell cycles. To explain the memory's unexpected persistence, we introduce a mathematical model incorporating extra protein memory storage elements with positive feedback that persist at the locus through DNA replication, in addition to histone modifications. Our hybrid model explains many features of epigenetic switching/memory at FLC and encapsulates generic mechanisms that may be widely applicable.

Using computational modelling to reveal mechanisms of epigenetic Polycomb control.

The Polycomb system is essential for stable gene silencing in many organisms. This regulation is achieved in part through addition of the histone modifications H3K27me2/me3 by Polycomb Repressive Complex 2 (PRC2). These modifications are believed to be the causative epigenetic memory elements of PRC2-mediated silencing. As these marks are stored locally in the chromatin, PRC2-based memory is a cis-acting system. A key feature of stable epigenetic memory in cis is PRC2-mediated, self-reinforcing feedback from K27-methylated histones onto nearby histones in a read-write paradigm. However, it was not clear under what conditions such feedback can lead to stable memory, able, for example, to survive the perturbation of histone dilution at DNA replication. In this context, computational modelling has allowed a rigorous exploration of possible underlying memory mechanisms and has also greatly accelerated our understanding of switching between active and silenced states. Specifically, modelling has predicted that switching and memory at Polycomb loci is digital, with a locus being either active or inactive, rather than possessing intermediate, smoothly varying levels of activation. Here, we review recent advances in models of Polycomb control, focusing on models of epigenetic switching through nucleation and spreading of H3K27me2/me3. We also examine models that incorporate transcriptional feedback antagonism and those including bivalent chromatin states. With more quantitative experimental data on histone modification kinetics, as well as single-cell resolution data on transcription and protein levels for PRC2 targets, we anticipate an expanded need for modelling to help dissect increasingly interconnected and complex memory mechanisms.

Exploring the Link between Nucleosome Occupancy and DNA Methylation.

Near promoters, both nucleosomes and CpG sites form characteristic spatial patterns. Previously, nucleosome depleted regions were observed upstream of transcription start sites and nucleosome occupancy was reported to correlate both with CpG density and the level of CpG methylation. Several studies imply a causal link where CpG methylation might induce nucleosome formation, whereas others argue the opposite, i.e., that nucleosome occupancy might influence CpG methylation. Correlations are indeed evident between nucleosomes, CpG density and CpG methylation-at least near promoter sites. It is however less established whether there is an immediate causal relation between nucleosome occupancy and the presence of CpG sites-or if nucleosome occupancy could be influenced by other factors. In this work, we test for such causality in human genomes by analyzing the three quantities both near and away from promoter sites. For data from the human genome we compare promoter regions with given CpG densities with genomic regions without promoters but of similar CpG densities. We find the observed correlation between nucleosome occupancy and CpG density, respectively CpG methylation, to be specific to promoter regions. In other regions along the genome nucleosome occupancy is statistically independent of the positioning of CpGs or their methylation levels. Anti-correlation between CpG density and methylation level is however similarly strong in both regions. On promoters, nucleosome occupancy is more strongly affected by the level of gene expression than CpG density or CpG methylation-calling into question any direct causal relation between nucleosome occupancy and CpG organization. Rather, our results suggest that for organisms with cytosine methylation nucleosome occupancy might be primarily linked to gene expression, with no strong impact on methylation.

Nanog, Oct4 and Tet1 interplay in establishing pluripotency.

A few central transcription factors inside mouse embryonic stem (ES) cells and induced pluripotent stem (iPS) cells are believed to control the cells' pluripotency. Characterizations of pluripotent state were put forward on both transcription factor and epigenetic levels. Whereas core players have been identified, it is desirable to map out gene regulatory networks which govern the reprogramming of somatic cells as well as the early developmental decisions. Here we propose a multiple level model where the regulatory network of Oct4, Nanog and Tet1 includes positive feedback loops involving DNA-demethylation around the promoters of Oct4 and Tet1. We put forward a mechanistic understanding of the regulatory dynamics which account for i) Oct4 overexpression is sufficient to induce pluripotency in somatic cell types expressing the other Yamanaka reprogramming factors endogenously; ii) Tet1 can replace Oct4 in reprogramming cocktail; iii) Nanog is not necessary for reprogramming however its over-expression leads to enhanced self-renewal; iv) DNA methylation is the key to the regulation of pluripotency genes; v) Lif withdrawal leads to loss of pluripotency. Overall, our paper proposes a novel framework combining transcription regulation with DNA methylation modifications which, takes into account the multi-layer nature of regulatory mechanisms governing pluripotency acquisition through reprogramming.

Stabilization of epigenetic states of CpG islands by local cooperation.

DNA methylation of CpG sites is an important epigenetic mark in mammals. Active promoters are often associated with unmethylated CpG sites, whereas methylated CpG sites correlate with silenced promoters. Methylation of CpG sites must be generally described as a dynamical process that is mediated by methylation enzymes, such as DNMT1 and DNMT3a/b. However, there are several models of how CpG sites can be protected from methylation and thereby remain unmethylated. In this paper we examine the combination of both: the positive feedbacks of DNA methylation and a short range counterpart which in turn protects-and thereby maintains-the unmethylated state. The emergent dynamics is provided by collaborative, re-enforcing feedbacks in favor of methylated CpG islands and cooperative protection of one CpG site by another in favor of unmethylated CpG sites. Our results suggest that this synthesis of mechanisms provides equally robust maintenance of both the unmethylated and methylated states of CpG islands.

DNA methylation in human epigenomes depends on local topology of CpG sites.

In vertebrates, methylation of cytosine at CpG sequences is implicated in stable and heritable patterns of gene expression. The classical model for inheritance, in which individual CpG sites are independent, provides no explanation for the observed non-random patterns of methylation. We first investigate the exact topology of CpG clustering in the human genome associated to CpG islands. Then, by pooling genomic CpG clusters on the basis of short distances between CpGs within and long distances outside clusters, we show a strong dependence of methylation on the number and density of CpG organization. CpG clusters with fewer, or less densely spaced, CpGs are predominantly hyper-methylated, while larger clusters are predominantly hypo-methylated. Intermediate clusters, however, are either hyper- or hypo-methylated but are rarely found in intermediate methylation states. We develop a model for spatially-dependent collaboration between CpGs, where methylated CpGs recruit methylation enzymes that can act on CpGs over an extended local region, while unmethylated CpGs recruit demethylation enzymes that act more strongly on nearby CpGs. This model can reproduce the effects of CpG clustering on methylation and produces stable and heritable alternative methylation states of CpG clusters, thus providing a coherent model for methylation inheritance and methylation patterning.

Collaboration between CpG sites is needed for stable somatic inheritance of DNA methylation states.

Inheritance of 5-methyl cytosine modification of CpG (CG/CG) DNA sequences is needed to maintain early developmental decisions in vertebrates. The standard inheritance model treats CpGs as independent, with methylated CpGs maintained by efficient methylation of hemimethylated CpGs produced after DNA replication, and unmethylated CpGs maintained by an absence of de novo methylation. By stochastic simulations of CpG islands over multiple cell cycles and systematic sampling of reaction parameters, we show that the standard model is inconsistent with many experimental observations. In contrast, dynamic collaboration between CpGs can provide strong error-tolerant somatic inheritance of both hypermethylated and hypomethylated states of a cluster of CpGs, reproducing observed stable bimodal methylation patterns. Known recruitment of methylating enzymes by methylated CpGs could provide the necessary collaboration, but we predict that recruitment of demethylating enzymes by unmethylated CpGs strengthens inheritance and allows CpG islands to remain hypomethylated within a sea of hypermethylation.

Improved contact prediction in proteins: using pseudolikelihoods to infer Potts models.

Spatially proximate amino acids in a protein tend to coevolve. A protein's three-dimensional (3D) structure hence leaves an echo of correlations in the evolutionary record. Reverse engineering 3D structures from such correlations is an open problem in structural biology, pursued with increasing vigor as more and more protein sequences continue to fill the data banks. Within this task lies a statistical inference problem, rooted in the following: correlation between two sites in a protein sequence can arise from firsthand interaction but can also be network-propagated via intermediate sites; observed correlation is not enough to guarantee proximity. To separate direct from indirect interactions is an instance of the general problem of inverse statistical mechanics, where the task is to learn model parameters (fields, couplings) from observables (magnetizations, correlations, samples) in large systems. In the context of protein sequences, the approach has been referred to as direct-coupling analysis. Here we show that the pseudolikelihood method, applied to 21-state Potts models describing the statistical properties of families of evolutionarily related proteins, significantly outperforms existing approaches to the direct-coupling analysis, the latter being based on standard mean-field techniques. This improved performance also relies on a modified score for the coupling strength. The results are verified using known crystal structures of specific sequence instances of various protein families. Code implementing the new method can be found at http://plmdca.csc.kth.se/.