Xiuwei Zhang

Accounting for technical noise in single-cell RNA-seq experiments

Single-cell RNA-seq can yield valuable insights about the variability within a population of seemingly homogeneous cells. We developed a quantitative statistical method to distinguish true biological variability from the high levels of technical noise in single-cell experiments. Our approach quantifies the statistical significance of observed cell-to-cell variability in expression strength on a gene-by-gene basis. We validate our approach using two independent data sets from Arabidopsis thaliana and Mus musculus.

Single-Cell RNA Sequencing Reveals T Helper Cells Synthesizing Steroids De Novo to Contribute to Immune Homeostasis

Summary T helper 2 (Th2) cells regulate helminth infections, allergic disorders, tumor immunity, and pregnancy by secreting various cytokines. It is likely that there are undiscovered Th2 signaling molecules. Although steroids are known to be immunoregulators, de novo steroid production from immune cells has not been previously characterized. Here, we demonstrate production of the steroid pregnenolone by Th2 cells in vitro and in vivo in a helminth infection model. Single-cell RNA sequencing and quantitative PCR analysis suggest that pregnenolone synthesis in Th2 cells is related to immunosuppression. In support of this, we show that pregnenolone inhibits Th cell proliferation and B cell immunoglobulin class switching. We also show that steroidogenic Th2 cells inhibit Th cell proliferation in a Cyp11a1 enzyme-dependent manner. We propose pregnenolone as a “lymphosteroid,” a steroid produced by lymphocytes. We speculate that this de novo steroid production may be an intrinsic phenomenon of Th2-mediated immune responses to actively restore immune homeostasis.

Refining transcriptional regulatory networks using network evolutionary models and gene histories.

BackgroundComputational inference of transcriptional regulatory networks remains a challenging problem, in part due to the lack of strong network models. In this paper we present evolutionary approaches to improve the inference of regulatory networks for a family of organisms by developing an evolutionary model for these networks and taking advantage of established phylogenetic relationships among these organisms. In previous work, we used a simple evolutionary model and provided extensive simulation results showing that phylogenetic information, combined with such a model, could be used to gain significant improvements on the performance of current inference algorithms.ResultsIn this paper, we extend the evolutionary model so as to take into account gene duplications and losses, which are viewed as major drivers in the evolution of regulatory networks. We show how to adapt our evolutionary approach to this new model and provide detailed simulation results, which show significant improvement on the reference network inference algorithms. Different evolutionary histories for gene duplications and losses are studied, showing that our adapted approach is feasible under a broad range of conditions. We also provide results on biological data (cis-regulatory modules for 12 species of Drosophila), confirming our simulation results.

Evolution of oligomeric state through allosteric pathways that mimic ligand binding.

Introduction Evolution and design of protein complexes are frequently viewed through the lens of amino acid mutations at protein interfaces, but we showed previously that residues distant from interfaces are also commonly involved in the evolution of alternative quaternary structures. We hypothesized that in these protein families, the difference in oligomeric state is due to a change in intersubunit geometry. The indirect mutations would act by changing protein conformation and dynamics, similar to the way in which allosteric small molecules introduce functional conformational change. We refer to these substitutions as “allosteric mutations.” Allosteric mutations change oligomeric state by employing the same conformational dynamics as ligands. PyrR homologs differ by mutations, all of which are outside the tetrameric interface. A subset of these allosteric mutations can be used to engineer a shift in oligomeric state in the ancestral PyrR. Allosteric mutations act by introducing conformational change in a manner analogous to that of the allosteric ligands. Rationale In this work, we investigate the mechanism of action of allosteric mutations on oligomeric state in the PyrR family of pyrimidine operon attenuators. In this family, an entirely sequence-conserved helix that forms a tetrameric interface in the thermophilic ortholog (BcPyrR) switches to being solvent-exposed in the mesophilic ortholog (BsPyrR). This results in a homodimeric structure in which the two subunits are clearly rotated relative to their orientation in the tetramer. What is the origin of this rotation and the change in quaternary structure? To dissect the role of the 49 substitutions between BsPyrR and BcPyrR, we used ancestral sequence reconstruction in combination with structural and biophysical methods to identify a set of allosteric mutations that are responsible for this shift in conformation. We compared the conformational changes introduced by the mutations to the protein motion during allosteric regulation by guanosine monophosphate (GMP). Results We identified 11 key mutations controlling oligomeric state, all distant from the interfaces and outside ligand-binding pockets. We confirmed the role of these allosteric mutations by engineering a shift in oligomeric state in an inferred ancestral PyrR protein (intermediate in sequence between the extant orthologs). We further used the inferred ancestral states and their mutants to show that the allosteric mutations are part of a downhill adaptation of the PyrR proteins to lower temperatures. We compared the x-ray crystal structures of ancestral and engineered PyrR proteins to the free and GMP-bound structure of the mesophilic BsPyrR, which shifts its equilibrium from dimer to tetramer upon ligand binding. Binding of the allosteric molecule introduces a change in intersubunit geometry that is equivalent to the evolutionary difference in intersubunit geometry between the dimeric and tetrameric homologs. We further find that the difference in oligomeric state is coupled to the difference in intrinsic dynamics of the dimers. Finally, we used the residue-residue contact network approach to show that the residues corresponding to the allosteric mutations undergo large contact rewiring when the intersubunit geometry and, in turn, oligomeric state change, either by GMP binding or by the introduction of allosteric mutations. Conclusion We show that evolution employs the intrinsic dynamics of this protein to toggle a conformational switch in a manner similar to that of small molecules. Shifting the relative populations of different states by subtle modifications is a process central to protein function and, as shown here, also to protein evolution. This suggests that we can learn from evolution and design proteins with multiple conformational states. Evolution and design of protein complexes are almost always viewed through the lens of amino acid mutations at protein interfaces. We showed previously that residues not involved in the physical interaction between proteins make important contributions to oligomerization by acting indirectly or allosterically. In this work, we sought to investigate the mechanism by which allosteric mutations act, using the example of the PyrR family of pyrimidine operon attenuators. In this family, a perfectly sequence-conserved helix that forms a tetrameric interface is exposed as solvent-accessible surface in dimeric orthologs. This means that mutations must be acting from a distance to destabilize the interface. We identified 11 key mutations controlling oligomeric state, all distant from the interfaces and outside ligand-binding pockets. Finally, we show that the key mutations introduce conformational changes equivalent to the conformational shift between the free versus nucleotide-bound conformations of the proteins. Mutations can alter protein conformations in the same way that allosteric small molecules do. Controlling the state of dynamic proteins Small molecules that change the oligomeric state of proteins by binding to a site distant from the interface are called allosteric. They often act by taking advantage of intrinsic protein dynamics and stabilizing a particular conformation of the protein. Perica et al. show that mutations can similarly act at a distance to change protein conformation. They identified 11 mutations in an RNA- binding protein that determine whether it is stable as a dimer or a tetramer. Examination of ancestral sequences showed that the allosteric mutations are part of a downhill adaptation to lower environmental temperatures. This mechanism for modulating the oligomeric state is probably common in evolution. Science, this issue 10.1126/science.1254346

Evolution of protein structures and interactions from the perspective of residue contact networks

Here we review mechanisms of protein evolution leading to structural changes in protein complexes. These mechanisms include mutations directly within protein interfaces, as well as the effects of mutations that propagate from distant regions of the protein. We also discuss the constraints protein complex structures impose on sequence evolution. We interpret, wherever possible, these mechanisms using amino acid residue contact networks. Many insights into protein evolution come from studies of monomers, and these results facilitate our understanding of evolution of protein complexes. Finally, we highlight the potential of formalizing a phylogenetic framework to integrate residue evolution, structure evolution, and to quantify changes in residue contact networks in protein families.

Single-cell analysis of CD4+ T-cell differentiation reveals three major cell states and progressive acceleration of proliferation

Differentiation of lymphocytes is frequently accompanied by cell cycle changes, interplay that is of central importance for immunity but is still incompletely understood. Here, we interrogate and quantitatively model how proliferation is linked to differentiation in CD4+ T cells. We perform ex vivo single-cell RNA-sequencing of CD4+ T cells during a mouse model of infection that elicits a type 2 immune response and infer that the differentiated, cytokine-producing cells cycle faster than early activated precursor cells. To dissect this phenomenon quantitatively, we determine expression profiles across consecutive generations of differentiated and undifferentiated cells during Th2 polarization in vitro. We predict three discrete cell states, which we verify by single-cell quantitative PCR. Based on these three states, we extract rates of death, division and differentiation with a branching state Markov model to describe the cell population dynamics. From this multi-scale modelling, we infer a significant acceleration in proliferation from the intermediate activated cell state to the mature cytokine-secreting effector state. We confirm this acceleration both by live imaging of single Th2 cells and in an ex vivo Th1 malaria model by single-cell RNA-sequencing. The link between cytokine secretion and proliferation rate holds both in Th1 and Th2 cells in vivo and in vitro, indicating that this is likely a general phenomenon in adaptive immunity.BackgroundDifferentiation of lymphocytes is frequently accompanied by cell cycle changes, interplay that is of central importance for immunity but is still incompletely understood. Here, we interrogate and quantitatively model how proliferation is linked to differentiation in CD4+ T cells.ResultsWe perform ex vivo single-cell RNA-sequencing of CD4+ T cells during a mouse model of infection that elicits a type 2 immune response and infer that the differentiated, cytokine-producing cells cycle faster than early activated precursor cells. To dissect this phenomenon quantitatively, we determine expression profiles across consecutive generations of differentiated and undifferentiated cells during Th2 polarization in vitro. We predict three discrete cell states, which we verify by single-cell quantitative PCR. Based on these three states, we extract rates of death, division and differentiation with a branching state Markov model to describe the cell population dynamics. From this multi-scale modelling, we infer a significant acceleration in proliferation from the intermediate activated cell state to the mature cytokine-secreting effector state. We confirm this acceleration both by live imaging of single Th2 cells and in an ex vivo Th1 malaria model by single-cell RNA-sequencing.ConclusionThe link between cytokine secretion and proliferation rate holds both in Th1 and Th2 cells in vivo and in vitro, indicating that this is likely a general phenomenon in adaptive immunity.

Boosting the Performance of Inference Algorithms for Transcriptional Regulatory Networks Using a Phylogenetic Approach

Inferring transcriptional regulatory networks from gene-expression data remains a challenging problem, in part because of the noisy nature of the data and the lack of strong network models. Time-series expression data have shown promise and recent work by Babu on the evolution of regulatory networks in E. coliand S. cerevisiaeopened another avenue of investigation. In this paper we take the evolutionary approach one step further, by developing ML-based refinement algorithms that take advantage of established phylogenetic relationships among a group of related organisms and of a simple evolutionary model for regulatory networks to improve the inference of these networks for these organisms from expression data gathered under similar conditions. We use simulations with different methods for generating gene-expression data, different phylogenies, and different evolutionary rates, and use different network inference algorithms, to study the performance of our algorithmic boosters. The results of simulations (including various tests to exclude confounding factors) demonstrate clear and significant improvements (in both specificity and sensitivity) on the performance of current inference algorithms. Thus gene-expression studies across a range of related organisms could yield significantly more accurate regulatory networks than single-organism studies.

Refining Regulatory Networks through Phylogenetic Transfer of Information

The experimental determination of transcriptional regulatory networks in the laboratory remains difficult and time-consuming, while computational methods to infer these networks provide only modest accuracy. The latter can be attributed partly to the limitations of a single-organism approach. Computational biology has long used comparative and evolutionary approaches to extend the reach and accuracy of its analyses. In this paper, we describe ProPhyC, a probabilistic phylogenetic model and associated inference algorithms, designed to improve the inference of regulatory networks for a family of organisms by using known evolutionary relationships among these organisms. ProPhyC can be used with various network evolutionary models and any existing inference method. Extensive experimental results on both biological and synthetic data confirm that our model (through its associated refinement algorithms) yields substantial improvement in the quality of inferred networks over all current methods. We also compare ProPhyC with a transfer learning approach we design. This approach also uses phylogenetic relationships while inferring regulatory networks for a family of organisms. Using similar input information but designed in a very different framework, this transfer learning approach does not perform better than ProPhyC, which indicates that ProPhyC makes good use of the evolutionary information.

Analysis of coevolving gene families using mutually exclusive orthologous modules.

Abstract Coevolutionary networks can encapsulate information about the dynamics of presence and absence of gene families in organisms. Analysis of such networks should reveal fundamental principles underlying the evolution of cellular systems and the functionality of sets of genes. In this study, we describe a new approach for analyzing coevolutionary networks. Our method detects Mutually Exclusive Orthologous Modules (MEOMs). A MEOM is composed of two sets of gene families, each including gene families that tend to appear in the same organisms, such that the two sets tend to mutually exclude each other (if one set appears in a certain organism the second set does not). Thus, a MEOM reflects the evolutionary replacement of one set of genes by another due to reasons such as lineage/environmental specificity, incompatibility, or functional redundancy. We use our method to analyze a coevolutionary network that is based on 383 microorganisms from the three domains of life. As we demonstrate, our method is useful for detecting meaningful evolutionary clades of organisms as well as sets of proteins that interact with each other. Among our results, we report that: 1) MEOMs tend to include gene families whose cellular functions involve transport, energy production, metabolism, and translation, suggesting that changes in the metabolic environments that require adaptation to new sources of energy are central triggers of complex/pathway replacement in evolution. 2) Many MEOMs are related to outer membrane proteins, such proteins are involved in interaction with the environment and could thus be replaced as a result of adaptation. 3) MEOMs tend to separate organisms with large phylogenetic distance but they also separate organisms that live in different ecological niches. 4) Strikingly, although many MEOMs can be identified, there are much fewer cases where the two cliques in the MEOM completely mutually exclude each other, demonstrating the flexibility of protein evolution. 5) CO dehydrogenase and thymidylate synthase and the glycine cleavage genes mutually exclude each other in archaea; this may represent an alternative route for generation of methyl donors for thymidine synthesis.

Improving inference of transcriptional regulatory networks based on network evolutionary models

Computational inference of transcriptional regulatory networks remains a challenging problem, in part due to the lack of strong network models. In this paper we present evolutionary approaches to improve the inference of regulatory networks for a family of organisms by developing an evolutionary model for these networks and taking advantage of established phylogenetic relationships among these organisms. In previous work, we used a simple evolutionary model for regulatory networks and provided extensive simulation results showing that phylogenetic information, combined with such a model, could be used to gain significant improvements on the performance of current inference algorithms. In this paper, we extend the evolutionary model so as to take into account gene duplications and losses, which are viewed as major drivers in the evolution of regulatory networks. We show how to adapt our evolutionary approach to this new model and provide detailed simulation results, which show significant improvement on the reference network inference algorithms. We also provide results on biological data (cis-regulatory modules for 12 species of Drosophila), confirming our simulation results.