baredSC: Bayesian approach to retrieve expression distribution of single-cell data

Background The number of studies using single-cell RNA sequencing (scRNA-seq) is constantly growing. This powerful technique provides a sampling of the whole transcriptome of a cell. However, sparsity of the data can be a major hurdle when studying the distribution of the expression of a specific gene or the correlation between the expressions of two genes. Results We show that the main technical noise associated with these scRNA-seq experiments is due to the sampling, i.e., Poisson noise. We present a new tool named baredSC, for Bayesian Approach to Retrieve Expression Distribution of Single-Cell data, which infers the intrinsic expression distribution in scRNA-seq data using a Gaussian mixture model. baredSC can be used to obtain the distribution in one dimension for individual genes and in two dimensions for pairs of genes, in particular to estimate the correlation in the two genes’ expressions. We apply baredSC to simulated scRNA-seq data and show that the algorithm is able to uncover the expression distribution used to simulate the data, even in multi-modal cases with very sparse data. We also apply baredSC to two real biological data sets. First, we use it to measure the anti-correlation between Hoxd13 and Hoxa11, two genes with known genetic interaction in embryonic limb. Then, we study the expression of Pitx1 in embryonic hindlimb, for which a trimodal distribution has been identified through flow cytometry. While other methods to analyze scRNA-seq are too sensitive to sampling noise, baredSC reveals this trimodal distribution. Conclusion baredSC is a powerful tool which aims at retrieving the expression distribution of few genes of interest from scRNA-seq data.

baredSC: Bayesian Approach to Retrieve Expression Distribution of Single-Cell

The number of studies using single-cell RNA sequencing (scRNA-seq) is constantly growing. This powerful technique provides a sampling of the whole transcriptome of a cell. However, the commonly used droplet-based method often produces very sparse samples. Sparsity can be a major hurdle when studying the distribution of the expression of a specific gene or the correlation between the expressions of two genes. We show that the main technical noise associated with these scRNA-seq experiments is due to the sampling (i.e. Poisson noise). We developed a new tool named baredSC, for Bayesian Approach to Retrieve Expression Distribution of Single-Cell, which infers the intrinsic expression distribution in noisy single-cell data using a Gaussian mixture model (GMM). baredSC can be used to obtain the distribution in one dimension for individual genes and in two dimensions for pairs of genes, in particular to estimate the correlation in the two genes' expressions. We apply baredSC to simulated scRNA-seq data and show that the algorithm is able to uncover the expression distribution used to simulate the data, even in multi-modal cases with very sparse data. We also apply baredSC to two real biological data sets. First, we use it to measure the anti-correlation between Hoxd13 and Hoxa11, two genes with known genetic interaction in embryonic limb. Then, we study the expression of Pitx1 in embryonic hindlimb, for which a trimodal distribution has been identified through flow cytometry. While other methods to analyze scRNA-seq are too sensitive to sampling noise, baredSC reveals this trimodal distribution.

Histone Epigenetic Signatures in Embryonic Limb Interdigital Cells Fated to Die

During limb formation in vertebrates with free digits, the interdigital mesoderm is eliminated by a massive degeneration process that involves apoptosis and cell senescence. The degradation process is preceded by intense DNA damage in zones located close to methylated DNA, accompanied by the activation of the DNA repair response. In this study, we show that trimethylated histone 3 (H3K4me3, H3K9me3, and H3K27me3) overlaps with zones positive for 5mC in the nuclei of interdigital cells. This pattern contrasts with the widespread distribution of acetylated histones (H3K9ac and H4ac) and the histone variant H3.3 throughout the nucleoplasm. Consistent with the intense labeling of acetylated histones, the histone deacetylase genes Hdac1, Hdac2, Hdac3, and Hdac8, and at a more reduced level, Hdac10, are expressed in the interdigits. Furthermore, local treatments with the histone deacetylase inhibitor trichostatin A, which promotes an open chromatin state, induces massive cell death and transcriptional changes reminiscent of, but preceding, the physiological process of interdigit remodeling. Together, these findings suggest that the epigenetic profile of the interdigital mesoderm contributes to the sensitivity to DNA damage that precedes apoptosis during tissue regression.

Modeling the bistable transition between cell phenotypes during limb precartilage condensation

Previous work showed that Gal-8 and Gal-1A, two proteins belonging to the galactoside-binding galectin family, are the earliest determinants of the patterning of the skeletal elements of embryonic chicken limbs, and further, that their experimentally determined interactions in the embryonic limb bud can be interpreted through a reaction-diffusion-adhesion framework. Here, we use an ordinary differential equation-based approach to analyze the intrinsic switching modality of the galectin reaction network and characterize the states of the network independent of the diffusive and adhesive arms of the patterning mechanism. We identify two steady states: where the concentrations of both the galectins are respectively, negligible, and very high. We provide an explicit Lyapunov function, which shows that there are no periodic solutions. In an extension of the model with sigmoidal galectin production terms, we show that an analogous bistable switch-like system arises via saddle-node bifurcation from a monostable one. Our model therefore predicts that the galectin network may exist in low expression and high expression states separated in space or time without any intermediate states. We verify these predictions in experiments performed with high density micromass cultures of chick limb mesenchymal cells and observe that cells inside and outside the precartilage protocondensations exhibit distinct behaviors with respect to galectin expression, motility, and spreading behavior on their substratum. The interactional complexity of the Gal-1 and -8-based patterning network is therefore sufficient to partition the mesenchymal cell population into two discrete cell types, which can be spatially patterned when incorporated into an adhesion and diffusion-enabled system.

RhoA/Rock activation represents a new mechanism for inactivating Wnt/β-catenin signaling in the aging-associated bone loss

The Wnt/β-catenin signaling pathway appears to be particularly important for bone homeostasis, whereas nuclear accumulation of β-catenin requires the activation of Rac1, a member of the Rho small GTPase family. The aim of the present study was to investigate the role of RhoA/Rho kinase (Rock)-mediated Wnt/β-catenin signaling in the regulation of aging-associated bone loss. We find that Lrp5/6-dependent and Lrp5/6-independent RhoA/Rock activation by Wnt3a activates Jak1/2 to directly phosphorylate Gsk3β at Tyr216, resulting in Gsk3β activation and subsequent β-catenin destabilization. In line with these molecular events, RhoA loss- or gain-of-function in mouse embryonic limb bud ectoderms interacts genetically with Dkk1 gain-of-function to rescue the severe limb truncation phenotypes or to phenocopy the deletion of β-catenin, respectively. Likewise, RhoA loss-of-function in pre-osteoblasts robustly increases bone formation while gain-of-function decreases it. Importantly, high RhoA/Rock activity closely correlates with Jak and Gsk3β activities but inversely correlates with β-catenin signaling activity in bone marrow mesenchymal stromal cells from elderly male humans and mice, whereas systemic inhibition of Rock therefore activates the β-catenin signaling to antagonize aging-associated bone loss. Taken together, these results identify RhoA/Rock-dependent Gsk3β activation and subsequent β-catenin destabilization as a hitherto uncharacterized mechanism controlling limb outgrowth and bone homeostasis.

Hidden limbs in the “limbless skink” Brachymeles lukbani: developmental observations

Reduced limbs and limblessness have evolved independently in many lizard clades. Skinks exhibit a wide range of limb-reduced morphologies, but only some species have been used to study the embryology of limb reduction (i.g., digit reduction in Chalcides and limb reduction in Scelotes). The genus Brachymeles, a Southeast Asian clade of skinks, includes species with a range of limb morphologies, from pentadactyl to functionally as well as structurally limbless species. Adults of the small, snake-like species Brachymeles lukbani show no sign of external limbs in the adult except for small depressions where they might be expected to occur. Embryos of B. lukbani in early stages of development, on the other hand, show a truncated but well-developed limb with a stylopod and a zeugopod, but no signs of an autopod. As development proceeds, the limb’s small size persists even while the embryo elongates. These observations are made based on external morphology. We used florescent whole-mount immunofluorescence to visualize the morphology of skeletal elements and muscles within the embryonic limb of B. lukabni. Early stages have a humerus and separated ulna and radius cartilages; associated with these structures are dorsal and ventral muscle masses as those found in the embryos of other limbed species. While the limb remains small, the pectoral girdle grows in proportion to the rest of the body, with well-developed skeletal elements and their associated muscles. In later stages of development, the small limb is still present under the skin but there are few indications of its presence, save for the morphology of the scale covering it. The adult morphology consists of a well-developed pectoral girdle, small humerus, extremely reduced ulna and radius, and well-developed limb musculature connected to the pectoral girdle. These muscles form in association with a developing limb during embryonic stages, a hint that “limbless” lizards that possesses these muscles may have or have had at least transient developing limbs, as we find in B. lukbani. Overall, the observed pattern of ontogenetic reduction, leading to an externally limbless adult in which a limb rudiment is hidden and covered under the trunk skin, is a situation called cryptomelia. The results of this work add to our growing understanding of clade-specific patterns of limb reduction and the convergent evolution of limbless phenotypes through different developmental processes.

A gene desert required for regulatory control of pleiotropic Shox2 expression and embryonic survival

ABSTRACTThe Shox2 homeodomain transcriptional regulator is known for its critical functions during mouse embryogenesis, enabling accurate development of limbs, craniofacial structures, neural populations and the cardiac conduction system. At the genomic level, the Shox2 gene is flanked by an extensive gene desert, a continuous non-coding genomic region spanning over 500 kilobases that contains a multitude of evolutionarily conserved elements with predicted cis-regulatory activities. However, the transcriptional enhancer potential of the vast majority of these elements in combination with the biological necessity of the gene desert have not yet been explored. Using transgenic reporter assays in mouse embryos to validate an extensive set of stringent epigenomic enhancer predictions, we identify several novel gene desert enhancers with distinct tissue-specific activities in Shox2 expressing tissues. 4C-seq chromatin conformation capture further uncovers a repertoire of gene desert enhancers with overlapping activities in the proximal limb, in a compartment essential for Shox2-mediated stylopod formation. Leveraging CRISPR/Cas9 to delete the gene desert region contained in the Shox2 topologically associated domain (TAD), we demonstrate that this complex cis-regulatory platform is essential for embryonic survival and required for control of region-specific Shox2 expression in multiple developing tissues. While transcription of Shox2 in the embryonic limb is only moderately affected by gene desert loss, Shox2 expression in craniofacial and cardiac domains is nearly abolished. In particular, Shox2 transcripts in the sinus venosus (SV) encompassing the sinoatrial node (SAN) were depleted in embryos lacking the gene desert, likely accounting for the embryonic lethality due to Shox2-dependency of the SAN pacemaker. Finally, we discover a 1.5kb SV enhancer within the deleted gene desert region, which may act as a genomic module controlling the development of the cardiac conduction system. In summary, our results identify a gene desert indispensable for pleiotropic patterning and highlight the importance of these extensive regulatory landscapes for embryonic development and viability.