Modeling stochastic gene expression: implications for haploinsufficiency

Haploinsufficient Transcription Factors in Myeloid Neoplasms

Many transcription factors (TFs) function as tumor suppressor genes with heterozygous phenotypes, yet haploinsufficiency generally has an underappreciated role in neoplasia. This is no less true in myeloid cells, which are normally regulated by a delicately balanced and interconnected transcriptional network. Detailed understanding of TF dose in this circuitry sheds light on the leukemic transcriptome. In this review, we discuss the emerging features of haploinsufficient transcription factors (HITFs). We posit that: (a) monoallelic and biallelic losses can have distinct cellular outcomes; (b) the activity of a TF exists in a greater range than the traditional Mendelian genetic doses; and (c) how a TF is deleted or mutated impacts the cellular phenotype. The net effect of a HITF is a myeloid differentiation block and increased intercellular heterogeneity in the course of myeloid neoplasia.

Genes for highly abundant proteins in Escherichia coli avoid 5' codons that promote ribosomal initiation

In many species highly expressed genes (HEGs) over-employ the synonymous codons that match the more abundant iso-acceptor tRNAs. Bacterial transgene codon randomization experiments report, however, that enrichment with such "translationally optimal" codons has little to no effect on the resultant protein level. By contrast, consistent with the view that ribosomal initiation is rate limiting, synonymous codon usage following the 5' ATG greatly influences protein levels, at least in part by modifying RNA stability. For the design of bacterial transgenes, for simple codon based in silico inference of protein levels and for understanding selection on synonymous mutations, it would be valuable to computationally determine initiation optimality (IO) scores for codons for any given species. One attractive approach is to characterize the 5' codon enrichment of HEGs compared with the most lowly expressed genes, just as translational optimality scores of codons have been similarly defined employing the full gene body. Here we determine the viability of this approach employing a unique opportunity: for Escherichia coli there is both the most extensive protein abundance data for native genes and a unique large-scale transgene codon randomization experiment enabling objective definition of the 5' codons that cause, rather than just correlate with, high protein abundance (that we equate with initiation optimality, broadly defined). Surprisingly, the 5' ends of native genes that specify highly abundant proteins avoid such initiation optimal codons. We find that this is probably owing to conflicting selection pressures particular to native HEGs, including selection favouring low initiation rates, this potentially enabling high efficiency of ribosomal usage and low noise. While the classical HEG enrichment approach does not work, rendering simple prediction of native protein abundance from 5' codon content futile, we report evidence that initiation optimality scores derived from the transgene experiment may hold relevance for in silico transgene design for a broad spectrum of bacteria.

Out-of-equilibrium gene expression fluctuations in the presence of extrinsic noise

Cell-to-cell variability in protein concentrations is strongly affected by extrinsic noise, especially for highly expressed genes. Extrinsic noise can be due to fluctuations of several possible cellular factors connected to cell physiology and to the level of key enzymes in the expression process. However, how to identify the predominant sources of extrinsic noise in a biological system is still an open question. This work considers a general stochastic model of gene expression with extrinsic noise represented as fluctuations of the different model rates, and focuses on the out-of-equilibrium expression dynamics. Combining analytical calculations with stochastic simulations, we characterize how extrinsic noise shapes the protein variability during gene activation or inactivation, depending on the prevailing source of extrinsic variability, on its intensity and timescale. In particular, we show that qualitatively different noise profiles can be identified depending on which are the fluctuating parameters. This indicates an experimentally accessible way to pinpoint the dominant sources of extrinsic noise using time-coarse experiments.

Transcription factor fluctuations underlie cell-to-cell variability in a signaling pathway response

Stochastic differences among clonal cells can initiate cell fate decisions in development or cause cell-to-cell differences in the responses to drugs or extracellular ligands. One hypothesis is that some of this phenotypic variability is caused by stochastic fluctuations in the activities of transcription factors (TFs). We tested this hypothesis in NIH3T3-CG cells using the response to Hedgehog signaling as a model cellular response. Here, we present evidence for the existence of distinct fast- and slow-responding substates in NIH3T3-CG cells. These two substates have distinct expression profiles, and fluctuations in the Prrx1 TF underlie some of the differences in expression and responsiveness between fast and slow cells. Our results show that fluctuations in TFs can contribute to cell-to-cell differences in Hedgehog signaling.

Out of Line or Altered States? Neural Progenitors as a Target in a Polygenic Neurodevelopmental Disorder

The genesis of a mature complement of neurons is thought to require, at least in part, precursor cell lineages in which neural progenitors have distinct identities recognized by exclusive expression of one or a few molecular markers. Nevertheless, limited progenitor types distinguished by specific markers and lineal progression through such subclasses cannot easily yield the magnitude of neuronal diversity in most regions of the nervous system. The late Verne Caviness, to whom this edition of Developmental Neuroscience is dedicated, recognized this mismatch. In his pioneering work on the histogenesis of the cerebral cortex, he acknowledged the additional flexibility required to generate multiple classes of cortical projection and interneurons. This flexibility may be accomplished by establishing cell states in which levels rather than binary expression or repression of individual genes vary across each progenitor's shared transcriptome. Such states may reflect local, stochastic signaling via soluble factors or coincidence of cell surface ligand/receptor pairs in subsets of neighboring progenitors. This probabilistic, rather than determined, signaling could modify transcription levels via multiple pathways within an apparently uniform population of progenitors. Progenitor states, therefore, rather than lineal relationships between types may underlie the generation of neuronal diversity in most regions of the nervous system. Moreover, mechanisms that influence variation required for flexible progenitor states may be targets for pathological changes in a broad range of neurodevelopmental disorders, especially those with polygenic origins.

A domestic plant differs from its wild relative along multiple axes of within-plant trait variability and diversity

For 10,000 years humans have altered plant traits through domestication and ongoing crop improvement, shaping plant form and function in agroecosystems. To date, studies have focused on how these processes shape whole-plant or average traits; however, plants also have characteristic levels of trait variability among their repeated parts, which can be heritable and mediate critical ecological interactions. Here, we examine an underappreciated scale of trait variation-among leaves, within plants-that may have changed through the process of domestication and improvement. Variability at this scale may itself be a target of selection, or be shaped as a by-product of the domestication process. We explore how levels of among-leaf trait variability differ between cultivars and wild relatives of alfalfa (Medicago sativa), a key forage crop with a 7,000-year domestication history. We grew individual plants from 30 wild populations and 30 cultivars, and quantified variability in a broad suite of physical, nutritive, and chemical leaf traits, including measures of chemical dissimilarity (beta diversity) among leaves within each plant. We find that trait variability has changed over the course of domestication, with effects often larger than changes in trait means. Domestic alfalfa had elevated among-leaf variability in SLA, trichomes, and C:N; increased diversity in defensive compounds; and reduced variability in phytochemical composition. We also elucidate fundamental relationships between trait means and variability, and between overall production of secondary metabolites and patterns of chemical diversity. We conclude that within-plant variability is an overlooked dimension of trait diversity in a globally critical agricultural crop. Trait variability is actually higher in cultivated plants compared to wild progenitors for multiple nutritive, physical, and chemical traits, highlighting a scale of variation that may mitigate loss of trait diversity at other scales in alfalfa agroecosystems, and in other crops with similar histories of domestication and improvement.

The good, the bad, and the ugly: Evolutionary and pathological aspects of gene dosage alterations

Diploid organisms contain a maternal and a paternal genome complement that is thought to provide robustness and allow developmental progression despite genetic perturbations that occur in heterozygosity. However, changes affecting gene dosage from the chromosome down to the individual gene level possess a significant pathological potential and can lead to developmental disorders (DDs). This indicates that expression from a balanced gene complement is highly relevant for proper cellular and organismal function in eukaryotes. Paradoxically, gene and whole chromosome duplications are a principal driver of evolution, while heteromorphic sex chromosomes (XY and ZW) are naturally occurring aneuploidies important for sex determination. Here, we provide an overview of the biology of gene dosage at the crossroads between evolutionary benefit and pathogenicity during disease. We describe the buffering mechanisms and cellular responses to alterations, which could provide a common ground for the understanding of DDs caused by copy number alterations.

Multiple de novo gene variations in a progeroid phenotype case report: haploinsufficiency mechanisms

We are presenting the case of a 6-year-old male patient with progeroid phenotype and severe developmental delay referred to Genetic clinic. Given the complex phenotype an extensive metabolic and genetic evaluation was performed including a whole exome sequencing analysis that showed genetic variants in TTR, RELN, MYH6, PHIP, and SYNE2 genes. Patients' mother and brother were analyzed for the genetic variants in MYH6, PHIP and RELN. Both had same variants on PHIP and RELN as our patient, with no apparent phenotypical consequences. Physical examination was remarkable for dysmorphism including plagiocephaly, low set and abnormally shaped ears, up slanted palpebral fissures, hypoplastic alae nasi, and a head circumference two standard deviations below the 3^rd percentile (microcephaly). Other characteristics include wrinkled skin, a broad forehead, sparse eyelashes in lower eyelid, short palpebral fissures, upturned nares, thick lips, right occipital plagiocephaly, overfolded helix and prominent anti-helix, protuberant chest, scaphoid abdomen, digitalized thumbs, and kyphosis due to low muscle tone. The patient presented abnormal EEG with evidence of epileptic discharges. A temporal bone CT showed plagiocephaly with flattening of the right occipital bone. Brain MRI showed callosal agenesis with bilateral colpocephaly with temporal horn dilatation, parahippocampal atrophy, lissencephaly and midbrain hypoplasia. The combination of de novo gene variants mentioned above has never been reported nor correlated as the result of haploinsufficiency mechanisms. Thus, we propose haploinsufficiency and loss of heterozygosity as etiological reasons for this patient phenotype. Further proteomic studies are needed to allocate the extense of genetic influence within the clinical manifestations.

Heterogeneity is not always a source of noise: Stochastic gene expression in regulatory heterozygote

Zygosity of diploid genome (i.e., degree to which two parental alleles of a gene have varied genetic sequences) adds another dimension to stochastic gene expression. The allelic imbalance in chromatin accessibility or divergence in regulatory sequences leads to fitness effects but the quantitative aspects thereof are largely left unexplored. We investigate diploid gene expression systems with homozygous (the same) and heterozygous (varied) combination of alleles in cis-regulatory sequences, not in structural gene loci, and characterize the zygosity-associated stochastic fluctuations in protein abundance. An emerging feat of heterozygosity is its counterintuitive capacity for genetic noise control. Especially when the noise is dominantly contributed to by the fluctuations in duty cycle ("reliability") rather than in process speed ("productivity") of gene expression machinery, its interallelic discrepancy acts to reduce the gene expression noise. These findings offer a novel insight into the rich repertoire of balancing selection enriched in the regulatory elements of immune response genes.

Noise suppression in stochastic genetic circuits using PID controllers

Inside individual cells, protein population counts are subject to molecular noise due to low copy numbers and the inherent probabilistic nature of biochemical processes. We investigate the effectiveness of proportional, integral and derivative (PID) based feedback controllers to suppress protein count fluctuations originating from two noise sources: bursty expression of the protein, and external disturbance in protein synthesis. Designs of biochemical reactions that function as PID controllers are discussed, with particular focus on individual controllers separately, and the corresponding closed-loop system is analyzed for stochastic controller realizations. Our results show that proportional controllers are effective in buffering protein copy number fluctuations from both noise sources, but this noise suppression comes at the cost of reduced static sensitivity of the output to the input signal. In contrast, integral feedback has no effect on the protein noise level from stochastic expression, but significantly minimizes the impact of external disturbances, particularly when the disturbance comes at low frequencies. Counter-intuitively, integral feedback is found to amplify external disturbances at intermediate frequencies. Next, we discuss the design of a coupled feedforward-feedback biochemical circuit that approximately functions as a derivate controller. Analysis using both analytical methods and Monte Carlo simulations reveals that this derivative controller effectively buffers output fluctuations from bursty stochastic expression, while maintaining the static input-output sensitivity of the open-loop system. In summary, this study provides a systematic stochastic analysis of biochemical controllers, and paves the way for their synthetic design and implementation to minimize deleterious fluctuations in gene product levels.

In the noisy cellular environment, biochemical species such as genes, RNAs and proteins that often occur at low molecular counts, are subject to considerable stochastic fluctuations in copy numbers over time. How cellular biochemical processes function reliably in the face of such randomness is an intriguing fundamental problem. Increasing evidence suggests that random fluctuations (noise) in protein copy numbers play important functional roles, such as driving genetically identical cells to different cell fates. Moreover, many disease states have been attributed to elevated noise levels in specific proteins. Here we systematically investigate design of biochemical systems that function as proportional, integral and derivative-based feedback controllers to suppress protein count fluctuations arising from bursty expression of the protein and external disturbance in protein synthesis. Our results show that different controllers are effective in buffering different noise components, and identify ranges of feedback gain for minimizing deleterious fluctuations in protein levels.

Transcriptional dysregulation in developing trigeminal sensory neurons in the LgDel mouse model of DiGeorge 22q11.2 deletion syndrome

LgDel mice, which model the heterozygous deletion of genes at human chromosome 22q11.2 associated with DiGeorge/22q11.2 deletion syndrome (22q11DS), have cranial nerve and craniofacial dysfunction as well as disrupted suckling, feeding and swallowing, similar to key 22q11DS phenotypes. Divergent trigeminal nerve (CN V) differentiation and altered trigeminal ganglion (CNgV) cellular composition prefigure these disruptions in LgDel embryos. We therefore asked whether a distinct transcriptional state in a specific population of early differentiating LgDel cranial sensory neurons, those in CNgV, a major source of innervation for appropriate oropharyngeal function, underlies this departure from typical development. LgDel versus wild-type (WT) CNgV transcriptomes differ significantly at E10.5 just after the ganglion has coalesced. Some changes parallel altered proportions of cranial placode versus cranial neural crest-derived CNgV cells. Others are consistent with a shift in anterior-posterior patterning associated with divergent LgDel cranial nerve differentiation. The most robust quantitative distinction, however, is statistically verifiable increased variability of expression levels for most of the over 17 000 genes expressed in common in LgDel versus WT CNgV. Thus, quantitative expression changes of functionally relevant genes and increased stochastic variation across the entire CNgV transcriptome at the onset of CN V differentiation prefigure subsequent disruption of cranial nerve differentiation and oropharyngeal function in LgDel mice.

Epigenetics drive the evolution of sex chromosomes in animals and plants

We review how epigenetics affect sex chromosome evolution in animals and plants. In a few species, sex is determined epigenetically through the action of Y-encoded small RNAs. Epigenetics is also responsible for changing the sex of individuals through time, even in species that carry sex chromosomes, and could favour species adaptation through breeding system plasticity. The Y chromosome accumulates repeats that become epigenetically silenced which leads to an epigenetic conflict with the expression of Y genes and could accelerate Y degeneration. Y heterochromatin can be lost through ageing, which activates transposable elements and lowers male longevity. Y chromosome degeneration has led to the evolution of meiotic sex chromosome inactivation in eutherians (placentals) and marsupials, and dosage compensation mechanisms in animals and plants. X-inactivation convergently evolved in eutherians and marsupials via two independently evolved non-coding RNAs. In Drosophila, male X upregulation by the male specific lethal (MSL) complex can spread to neo-X chromosomes through the transposition of transposable elements that carry an MSL-binding motif. We discuss similarities and possible differences between plants and animals and suggest future directions for this dynamic field of research. This article is part of the theme issue 'How does epigenetics influence the course of evolution?'

KFGRNI: A robust method to inference gene regulatory network from time-course gene data based on ensemble Kalman filter

A central problem of systems biology is the reconstruction of Gene Regulatory Networks (GRNs) by the use of time series data. Although many attempts have been made to design an efficient method for GRN inference, providing a best solution is still a challenging task. Existing noise, low number of samples, and high number of nodes are the main reasons causing poor performance of existing methods. The present study applies the ensemble Kalman filter algorithm to model a GRN from gene time series data. The inference of a GRN is decomposed with p genes into p subproblems. In each subproblem, the ensemble Kalman filter algorithm identifies the weight of interactions for each target gene. With the use of the ensemble Kalman filter, the expression pattern of the target gene is predicted from the expression patterns of all the remaining genes. The proposed method is compared with several well-known approaches. The results of the evaluation indicate that the proposed method improves inference accuracy and demonstrates better regulatory relations with noisy data.

MLL4 is required after implantation, whereas MLL3 becomes essential during late gestation

Methylation of histone 3 lysine 4 (H3K4) is a major epigenetic system associated with gene expression. In mammals there are six H3K4 methyltransferases related to yeast Set1 and fly Trithorax, including two orthologs of fly Trithorax-related: MLL3 and MLL4. Exome sequencing has documented high frequencies of MLL3 and MLL4 mutations in many types of human cancer. Despite this emerging importance, the requirements of these paralogs in mammalian development have only been incompletely reported. Here, we examined the null phenotypes to establish that MLL3 is first required for lung maturation, whereas MLL4 is first required for migration of the anterior visceral endoderm that initiates gastrulation in the mouse. This collective cell migration is preceded by a columnar-to-squamous transition in visceral endoderm cells that depends on MLL4. Furthermore, Mll4 mutants display incompletely penetrant, sex-distorted, embryonic haploinsufficiency and adult heterozygous mutants show aspects of Kabuki syndrome, indicating that MLL4 action, unlike MLL3, is dosage dependent. The highly specific and discordant functions of these paralogs in mouse development argues against their action as general enhancer factors.

DNA methylation ambiguity in the Fibrillin-1 (FBN1) CpG island shore possibly involved in Marfan syndrome

Fibrillin-1 (FBN1) is responsible for haploinsufficient and autosomal dominant Marfan syndrome. Even in the same Marfan pedigree, penetrance and expressivity in heterozygous individuals can differ and result in variable disease onset and severity. Thus, other factors in addition to mutations in FBN1 are likely to contribute to the disease. In this study, we examined the regulation of FBN1 in porcine Marfan syndrome model, focusing on DNA methylation patterns distinguishable as wild-type (WT) and FBN1 null (KO) alleles in heterozygous cells. Most importantly, the ratio of the transcriptionally active hypomethylated WT allele was altered during cellular passage and highly correlated with FBN1 mRNA level compared with that in the KO allele. Transcribed FBN1 RNA from the KO allele was abolished after splicing coupled with translational initiation, suggesting that the functional FBN1 mRNA levels were affected by DNA methylation of the WT allele.

The regulatory landscapes of developmental genes

Regulatory landscapes have been defined in vertebrates as large DNA segments containing diverse enhancer sequences that produce coherent gene transcription. These genomic platforms integrate multiple cellular signals and hence can trigger pleiotropic expression of developmental genes. Identifying and evaluating how these chromatin regions operate may be difficult as the underlying regulatory mechanisms can be as unique as the genes they control. In this brief article and accompanying poster, we discuss some of the ways in which regulatory landscapes operate, illustrating these mechanisms using genes important for vertebrate development as examples. We also highlight some of the techniques available to researchers for analysing regulatory landscapes.

Dosage sensitivity of X-linked genes in human embryonic single cells

BACKGROUND

During the evolution of mammalian sex chromosomes, the degeneration of Y-linked homologs has led to a dosage imbalance between X-linked and autosomal genes. The evolutionary resolution to such dosage imbalance, as hypothesized by Susumu Ohno fifty years ago, should be doubling the expression of X-linked genes. Recent studies have nevertheless shown that the X to autosome expression ratio equals ~ 1 in haploid human parthenogenetic embryonic stem (pES) cells and ~ 0.5 in diploid pES cells, suggesting no doubled expression for X-linked genes and refuting Ohno's hypothesis.

RESULTS

Here, by reanalyzing an RNA-seq-based single-cell transcriptome dataset of human embryos, we found that from the 8-cell stage until the time-point just prior to implantation, the expression levels of X-linked genes are not two-fold upregulated in male cells and gradually decrease from two-fold in female cells. Additional analyses of gene expression noise further suggest that the dosage sensitivity of X-linked genes is weaker than that of autosomal genes in differentiated female cells, which contradicts a key assumption in Ohno's hypothesis, that most X-linked genes are dosage sensitive. Moreover, the dosage-sensitive housekeeping genes are preferentially located on autosomes, implying selection against X-linkage for dosage-sensitive genes.

CONCLUSIONS

We observed dosage imbalance between X-linked and autosomal genes, as well as relatively high expression noise from X-linked genes. These results collectively suggest that X-linked genes are less dosage sensitive than autosomal genes, putting one primary assumption of Ohno's hypothesis in question.

Do GWAS and studies of heterozygotes for NPC1 and/or NPC2 explain why NPC disease cases are so rare?

Early onset Niemann-Pick C diseases are extremely rare, especially Niemann-Pick C2. Perhaps unusually for autosomal recessive diseases, heterozygotes for mutations in NPC1 manifest many biological variations. NPC2 deficiency has large effects on fertility. These features of NPC1 and NPC2 are reviewed in regard to possible negative selection for heterozygotes carrying null and hypomorphic alleles.

Dosage-Dependent Expression Variation Suppressed on the Drosophila Male X Chromosome

DNA copy number variation is associated with many high phenotypic heterogeneity disorders. We systematically examined the impact of Drosophila melanogaster deletions on gene expression profiles to ask whether increased expression variability owing to reduced gene dose might underlie this phenotypic heterogeneity. Indeed, we found that one-dose genes have higher gene expression variability relative to two-dose genes. We then asked whether this increase in variability could be explained by intrinsic noise within cells due to stochastic biochemical events, or whether expression variability is due to extrinsic noise arising from more complex interactions. Our modeling showed that intrinsic gene expression noise averages at the organism level and thus cannot explain increased variation in one-dose gene expression. Interestingly, expression variability was related to the magnitude of expression compensation, suggesting that regulation, induced by gene dose reduction, is noisy. In a remarkable exception to this rule, the single X chromosome of males showed reduced expression variability, even compared with two-dose genes. Analysis of sex-transformed flies indicates that X expression variability is independent of the male differentiation program. Instead, we uncovered a correlation between occupancy of the chromatin-modifying protein encoded by males absent on the first (mof) and expression variability, linking noise suppression to the specialized X chromosome dosage compensation system. MOF occupancy on autosomes in both sexes also lowered transcriptional noise. Our results demonstrate that gene dose reduction can lead to heterogeneous responses, which are often noisy. This has implications for understanding gene network regulatory interactions and phenotypic heterogeneity. Additionally, chromatin modification appears to play a role in dampening transcriptional noise.

Mechanisms of Mendelian dominance

Genetic dominance has long been considered as a qualitative reflection of interallelic interactions. Dominance arises from many multiple sources whose unifying theme is the existence of non-linear relationships between the genotypic and phenotypic values. One of the clearest examples are dominant negative mutations (DNMs) in which a defective subunit poisons a macromolecular complex. Dominance can also be due to the presence of a heterozygous null allele, as is the case of haploinsufficiency. Dominance can also be influenced by epistatic (interloci) interactions. For instance, a pre-existing genetic variant can make possible the expression of a pathogenic variant in a seemingly "dominant" fashion. Such interactions, which can make an individual more or less sensitive to a particular pathogenic variant, will also be discussed here.

A novel frameshift GRN mutation results in frontotemporal lobar degeneration with a distinct clinical phenotype in two siblings: case report and literature review

Background

Progranulin gene (GRN) mutations are major causes of frontotemporal lobar degeneration. To date, 68 pathogenic GRN mutations have been identified. However, very few of these mutations have been reported in Asians. Moreover, some GRN mutations manifest with familial phenotypic heterogeneity. Here, we present a novel GRN mutation resulting in frontotemporal lobar degeneration with a distinct clinical phenotype, and we review reports of GRN mutations associated with familial phenotypic heterogeneity.

Case presentation

We describe the case of a 74-year-old woman with left frontotemporal lobe atrophy who presented with progressive anarthria and non-fluent aphasia. Her brother had been diagnosed with corticobasal syndrome (CBS) with right-hand limb-kinetic apraxia, aphasia, and a similar pattern of brain atrophy. Laboratory blood examinations did not reveal abnormalities that could have caused cognitive dysfunction. In the cerebrospinal fluid, cell counts and protein concentrations were within normal ranges, and concentrations of tau protein and phosphorylated tau protein were also normal. Since similar familial cases due to mutation of GRN and microtubule-associated protein tau gene (MAPT) were reported, we performed genetic analysis. No pathological mutations of MAPT were identified, but we identified a novel GRN frameshift mutation (c.1118_1119delCCinsG: p.Pro373ArgX37) that resulted in progranulin haploinsufficiency.

Conclusion

This is the first report of a GRN mutation associated with familial phenotypic heterogeneity in Japan. Literature review of GRN mutations associated with familial phenotypic heterogeneity revealed no tendency of mutation sites. The role of progranulin has been reported in this and other neurodegenerative diseases, and the analysis of GRN mutations may lead to the discovery of a new therapeutic target.

Establishment of DNA methylation patterns of the Fibrillin1 (FBN1) gene in porcine embryos and tissues

DNA methylation in transcriptional regulatory regions is crucial for gene expression. The DNA methylation status of the edges of CpG islands, called CpG island shore, is involved in tissue/cell-type-specific gene expression. Haploinsufficiency diseases are caused by inheritance of one mutated null allele and are classified as autosomal dominant. However, in the same pedigree, phenotypic variances are observed despite the inheritance of the identical mutated null allele, including Fibrillin1 (FBN1), which is responsible for development of the haploinsufficient Marfan disease. In this study, we examined the relationship between gene expression and DNA methylation patterns of the FBN1 CpG island shore focusing on transcriptionally active hypomethylated alleles (Hypo-alleles). No difference in the DNA methylation level of FBN1 CpG island shore was observed in porcine fetal fibroblast (PFF) and the liver, whereas FBN1 expression was higher in PFF than in the liver. However, Hypo-allele ratio of the FBN1 CpG island shore in PFF was higher than that in the liver, indicating that Hypo-allele ratio of the FBN1 CpG island shore likely correlated with FBN1 expression level. In addition, oocyte-derived DNA hypermethylation in preimplantation embryos was erased until the blastocyst stage, and re-methylation of the FBN1 CpG island shore was observed with prolonged in vitro culture of blastocysts. These results suggest that the establishment of the DNA methylation pattern within the FBN1 CpG island shore occurs after the blastocyst stage, likely during organogenesis. In conclusion, Hypo-allele ratios of the FBN1 CpG island shore correlated with FBN1 expression levels in porcine tissues.

Analytic framework for a stochastic binary biological switch

We propose and solve analytically a stochastic model for the dynamics of a binary biological switch, defined as a DNA unit with two mutually exclusive configurations, each one triggering the expression of a different gene. Such a device has the potential to be used as a memory unit for biological computing systems designed to operate in noisy environments. We discuss a recent implementation of this switch in living cells, the recombinase addressable data (RAD) module. In order to understand the behavior of a RAD module we compute the exact time-dependent joint distribution of the two expressed genes starting in one state and evolving to another asymptotic state. We consider two operating regimes of the RAD module, a fast and a slow stochastic switching regime. The fast regime is aggregative and produces unimodal distributions, whereas the slow regime is separative and produces bimodal distributions. Both regimes can serve to prepare pure memory states when all cells are expressing the same gene. The slow regime can also separate mixed states by producing two subpopulations, each one expressing a different gene. Compared to the genetic toggle switch based on positive feedback, the RAD module ensures more rapid memory operations for the same quality of the separation between binary states. Our model provides a simplified phenomenological framework for studying RAD memory devices and our analytic solution can be further used to clarify theoretical concepts in biocomputation and for optimal design in synthetic biology.

Macromolecular Crowding Regulates the Gene Expression Profile by Limiting Diffusion

We seek to elucidate the role of macromolecular crowding in transcription and translation. It is well known that stochasticity in gene expression can lead to differential gene expression and heterogeneity in a cell population. Recent experimental observations by Tan et al. have improved our understanding of the functional role of macromolecular crowding. It can be inferred from their observations that macromolecular crowding can lead to robustness in gene expression, resulting in a more homogeneous cell population. We introduce a spatial stochastic model to provide insight into this process. Our results show that macromolecular crowding reduces noise (as measured by the kurtosis of the mRNA distribution) in a cell population by limiting the diffusion of transcription factors (i.e. removing the unstable intermediate states), and that crowding by large molecules reduces noise more efficiently than crowding by small molecules. Finally, our simulation results provide evidence that the local variation in chromatin density as well as the total volume exclusion of the chromatin in the nucleus can induce a homogenous cell population.

The cellular nucleus is packed with macromolecules such as DNAs and proteins, which leaves limited space for other molecules to move around. Recent experimental results by C. Tan et al. have shown that macromolecular crowding can regulate gene expression, resulting in a more homogenous cell population. We introduce a computational model to uncover the mechanism by which macromolecular crowding functions. Our results suggest that macromolecular crowding limits the diffusion of the transcription factors and attenuates the transcriptional bursting, which leads to a more homogenous cell population. Regulation of gene expression noise by macromolecules depends on the size of the crowders, i.e. larger macromolecules can reduce the noise more effectively than smaller macromolecules. We also demonstrate that local variation of chromatin density can affect the noise of gene expression. This shows the importance of the chromatin structure in gene expression regulation.

Analysis of allelic expression patterns in clonal somatic cells by single-cell RNA-seq

Cellular heterogeneity can emerge from the expression of only one parental allele. However, it has remained controversial whether, or to what degree, random monoallelic expression of autosomal genes (aRME) is mitotically inherited (clonal) or stochastic (dynamic) in somatic cells, particularly in vivo. Here we used allele-sensitive single-cell RNA-seq on clonal primary mouse fibroblasts and freshly isolated human CD8⁺ T cells to dissect clonal and dynamic monoallelic expression patterns. Dynamic aRME affected a considerable portion of the cells' transcriptomes, with levels dependent on the cells' transcriptional activity. Notably, clonal aRME was detected, but it was surprisingly scarce (<1% of genes) and mainly affected the most weakly expressed genes. Consequently, the overwhelming majority of aRME occurs transiently within individual cells, and patterns of aRME are thus primarily scattered throughout somatic cell populations rather than, as previously hypothesized, confined to patches of clonally related cells.

Demonstration of Autosomal Monoallelic Expression in Thyroid Tissue Assessed by Whole-Exome and Bulk RNA Sequencing

BACKGROUND

Congenital hypothyroidism due to thyroid dysgenesis (CHTD) is a disorder with a prevalence of 1/4000 live births, the cause of which remains unknown. The most common diagnostic category is thyroid ectopy, which occurs in up to 80% of CHTD cases. CHTD is predominantly not inherited and has a high discordance rate (>92%) between monozygotic (MZ) twins. The sporadic nature of CHTD might be explained by somatic events such as autosomal monoallelic expression (AME), given that genes expressed in a monoallelic way are more vulnerable to otherwise benign monoallelict genetic or epigenetic mutations.

OBJECTIVE

The aim of this study was to search for complete (90%) AME in normal and dysgenetic thyroid tissues.

METHODS

Aggregated analysis of whole-exome and bulk RNA sequencing was performed on two ectopic thyroids, four normal thyroids, and the human thyroid cell line Nthy-ori.

RESULTS

A median of 5062 (range 2081-5270) genes per sample showed sufficient numbers of heterozygous single nucleotide polymorphisms to be informative. The median monoallelic expression represented 22 (range 16-32) of the informative genes for each thyroid sample. Examples of genes displaying AME are FCGBP, ZNF331, USP10, BCLAF1, and some HLA genes; these genes are involved in epithelial-mesenchymal transition, cell migration, cancer, and immunity.

CONCLUSIONS

AME may account for the high discordance rate observed between MZ twins and for the sporadic nature of CHTD. These findings also have implications for other pathologies, including cancers and autoimmune disorders of the thyroid.

Random monoallelic expression of autosomal genes: stochastic transcription and allele-level regulation

Random monoallelic expression (RME) of genes represents a striking example of how stochastic molecular processes can result in cellular heterogeneity. Recent transcriptome-wide studies have revealed both mitotically stable and cell-to-cell dynamic forms of autosomal RME, with the latter presumably resulting from burst-like stochastic transcription. Here, we discuss the distinguishing features of these two forms of RME and revisit literature on their nature, pervasiveness and regulation. Finally, we explore how RME may contribute to phenotypic variation, including the incomplete penetrance and variable expressivity often seen in genetic disease.

Post-transcriptional regulation tends to attenuate the mRNA noise and to increase the mRNA gain

Post-transcriptional regulation is ubiquitous in prokaryotic and eukaryotic cells, but how it impacts gene expression remains to be fully explored. Here, we analyze a simple gene model in which we assume that mRNAs are produced in a constitutive manner but are regulated post-transcriptionally by a decapping enzyme that switches between the active state and the inactive state. We derive the analytical mRNA distribution governed by a chemical master equation, which can be well used to analyze the mechanism of how post-transcription regulation influences the mRNA expression level including the mRNA noise. We demonstrate that the mean mRNA level in the stochastic case is always higher than that in the deterministic case due to the stochastic effect of the enzyme, but the size of the increased part depends mainly on the switching rates between two enzyme states. More interesting is that we find that in contrast to transcriptional regulation, post-transcriptional regulation tends to attenuate noise in mRNA. Our results provide insight into the role of post-transcriptional regulation in controlling the transcriptional noise.

X chromosome inactivation and active X upregulation in therian mammals: facts, questions, and hypotheses

X chromosome inactivation is a mechanism that modulates the expression of X-linked genes in eutherian females (XX). Ohno proposed that to achieve a proper balance between X-linked and autosomal genes, those on the active X should also undergo a 2-fold upregulation. Although some support for Ohno's hypothesis has been provided through the years, recent genomic studies testing this hypothesis have brought contradictory results and fueled debate. Thus far, there are as many results in favor as against Ohno's hypothesis, depending on the nature of the datasets and the various assumptions and thresholds involved in the analyses. However, they have confirmed the importance of dosage balance between X-linked and autosomal genes involved in stoichiometric relationships. These facts as well as questions and hypotheses are discussed below.

Effects of elongation delay in transcription dynamics

In the transcription process, elongation delay is induced by the movement of RNA polymerases (RNAP) along the DNA sequence, and can result in changes in the transcription dynamics. This paper studies the transcription dynamics that involved the elongation delay and effects of cell division and DNA replication. The stochastic process of gene expression is modeled with delay chemical master equation with periodic coefficients, and is studied numerically through the stochastic simulation algorithm with delay. We show that the average transcription level approaches to a periodic dynamics over cell cycles at homeostasis, and the elongation delay can reduce the transcription level and increase the transcription noise. Moreover, the transcription elongation can induce bimodal distribution of mRNA levels that can be measured by the techniques of flow cytometry.

Parallel solutions for voxel-based simulations of reaction-diffusion systems

There is an increasing awareness of the pivotal role of noise in biochemical processes and of the effect of molecular crowding on the dynamics of biochemical systems. This necessity has given rise to a strong need for suitable and sophisticated algorithms for the simulation of biological phenomena taking into account both spatial effects and noise. However, the high computational effort characterizing simulation approaches, coupled with the necessity to simulate the models several times to achieve statistically relevant information on the model behaviours, makes such kind of algorithms very time-consuming for studying real systems. So far, different parallelization approaches have been deployed to reduce the computational time required to simulate the temporal dynamics of biochemical systems using stochastic algorithms. In this work we discuss these aspects for the spatial TAU-leaping in crowded compartments (STAUCC) simulator, a voxel-based method for the stochastic simulation of reaction-diffusion processes which relies on the Sτ-DPP algorithm. In particular we present how the characteristics of the algorithm can be exploited for an effective parallelization on the present heterogeneous HPC architectures.

Transcription factors: specific DNA binding and specific gene regulation

Specific recognition of cis-regulatory regions is essential for correct gene regulation in response to developmental and environmental signals. Such DNA sequences are recognized by transcription factors (TFs) that recruit the transcriptional machinery. Achievement of specific sequence recognition is not a trivial problem; many TFs recognize similar consensus DNA-binding sites and a genome can harbor thousands of consensus or near-consensus sequences, both functional and nonfunctional. Although genomic technologies have provided large-scale snapshots of TF binding, a full understanding of the mechanistic and quantitative details of specific recognition in the context of gene regulation is lacking. Here, we explore the various ways in which TFs recognizing similar consensus sites distinguish their own targets from a large number of other sequences to ensure specific cellular responses.

Deafness gene expression patterns in the mouse cochlea found by microarray analysis

BACKGROUND

Tonotopy is one of the most fundamental principles of auditory function. While gradients in various morphological and physiological characteristics of the cochlea have been reported, little information is available on gradient patterns of gene expression. In addition, the audiograms in autosomal dominant non syndromic hearing loss can be distinctive, however, the mechanism that accounts for that has not been clarified. We thought that it is possible that tonotopic gradients of gene expression within the cochlea account for the distinct audiograms.

METHODOLOGY/PRINCIPAL FINDINGS

We compared expression profiles of genes in the cochlea between the apical, middle, and basal turns of the mouse cochlea by microarray technology and quantitative RT-PCR. Of 24,547 genes, 783 annotated genes expressed more than 2-fold. The most remarkable finding was a gradient of gene expression changes in four genes (Pou4f3, Slc17a8, Tmc1, and Crym) whose mutations cause autosomal dominant deafness. Expression of these genes was greater in the apex than in the base. Interestingly, expression of the Emilin-2 and Tectb genes, which may have crucial roles in the cochlea, was also greater in the apex than in the base.

CONCLUSIONS/SIGNIFICANCE

This study provides baseline data of gradient gene expression in the cochlea. Especially for genes whose mutations cause autosomal dominant non syndromic hearing loss (Pou4f3, Slc17a8, Tmc1, and Crym) as well as genes important for cochlear function (Emilin-2 and Tectb), gradual expression changes may help to explain the various pathological conditions.

Role of mef2ca in developmental buffering of the zebrafish larval hyoid dermal skeleton

Phenotypic robustness requires a process of developmental buffering that is largely not understood, but which can be disrupted by mutations. Here we show that in mef2ca(b1086) loss of function mutant embryos and early larvae, development of craniofacial hyoid bones, the opercle (Op) and branchiostegal ray (BR), becomes remarkably unstable; the large magnitude of the instability serves as a positive attribute to learn about features of this developmental buffering. The OpBR mutant phenotype variably includes bone expansion and fusion, Op duplication, and BR homeosis. Formation of a novel bone strut, or a bone bridge connecting the Op and BR together occurs frequently. We find no evidence that the phenotypic stability in the wild type is provided by redundancy between mef2ca and its co-ortholog mef2cb, or that it is related to the selector (homeotic) gene function of mef2ca. Changes in dorsal-ventral patterning of the hyoid arch also might not contribute to phenotypic instability in mutants. However, subsequent development of the bone lineage itself, including osteoblast differentiation and morphogenetic outgrowth, shows marked variation. Hence, steps along the developmental trajectory appear differentially sensitive to the loss of buffering, providing focus for the future study.

Alternative-splicing-mediated gene expression

Alternative splicing (AS) is a fundamental process during gene expression and has been found to be ubiquitous in eukaryotes. However, how AS impacts gene expression levels both quantitatively and qualitatively remains to be fully explored. Here, we analyze two common models of gene expression, each incorporating a simple splice mechanism that a pre-mRNA is spliced into two mature mRNA isoforms in a probabilistic manner. In the constitutive expression case, we show that the steady-state molecular numbers of two mature mRNA isoforms follow mutually independent Poisson distributions. In the bursting expression case, we demonstrate that the tail decay of the steady-state distribution for both mature mRNA isoforms that in general are not mutually independent can be characterized by the product of mean burst size and splicing probability. In both cases, we find that AS can efficiently modulate both the variability (measured by variance) and the noise level of the total mature mRNA, and in particular, the latter is always lower than the noise level of the pre-mRNA, implying that AS always reduces the noise. These results altogether reveal that AS is a mechanism of efficiently controlling the gene expression noise.

Quantifying intrinsic and extrinsic variability in stochastic gene expression models

Genetically identical cell populations exhibit considerable intercellular variation in the level of a given protein or mRNA. Both intrinsic and extrinsic sources of noise drive this variability in gene expression. More specifically, extrinsic noise is the expression variability that arises from cell-to-cell differences in cell-specific factors such as enzyme levels, cell size and cell cycle stage. In contrast, intrinsic noise is the expression variability that is not accounted for by extrinsic noise, and typically arises from the inherent stochastic nature of biochemical processes. Two-color reporter experiments are employed to decompose expression variability into its intrinsic and extrinsic noise components. Analytical formulas for intrinsic and extrinsic noise are derived for a class of stochastic gene expression models, where variations in cell-specific factors cause fluctuations in model parameters, in particular, transcription and/or translation rate fluctuations. Assuming mRNA production occurs in random bursts, transcription rate is represented by either the burst frequency (how often the bursts occur) or the burst size (number of mRNAs produced in each burst). Our analysis shows that fluctuations in the transcription burst frequency enhance extrinsic noise but do not affect the intrinsic noise. On the contrary, fluctuations in the transcription burst size or mRNA translation rate dramatically increase both intrinsic and extrinsic noise components. Interestingly, simultaneous fluctuations in transcription and translation rates arising from randomness in ATP abundance can decrease intrinsic noise measured in a two-color reporter assay. Finally, we discuss how these formulas can be combined with single-cell gene expression data from two-color reporter experiments for estimating model parameters.

Random or stochastic monoallelic expressed genes are enriched for neurodevelopmental disorder candidate genes

Random or stochastic monoallelic expressed genes (StMA genes) represent a unique form of monoallelic expression where allelic choice is made at random early in development. The consequential clonal diversity provides opportunity for functional heterozygosity in tissues such as the brain, and can impact on both development and disease. We investigate the relationship of StMA expressed genes previously identified in clonal neural stem cells with the neurodevelopmental disorders autism and schizophrenia. We found that StMA genes show an overrepresentation of schizophrenia risk candidates identified by genome wide association studies from the genetic association database. Similar suggestive enrichment was also found for genes from the NHGRI genome-wide association study catalog and a psychiatric genetics consortium schizophrenia dataset although these latter more robust gene lists did not achieve statistical significance. We also examined multiple sources of copy number variation (CNV) datasets from autism and schizophrenia cohorts. After taking into account total gene numbers and CNV size, both autism and schizophrenia associated CNVs appeared to show an enrichment of StMA genes relative to the control CNV datasets. Since the StMA genes were originally identified in neural stem cells, bias due to the neural transcriptome is possible. To address this, we randomly sampled neural stem cell expressed genes and repeated the tests. After a significant number of iterations, neural stem cell expressed genes did not show an overrepresentation in autism or schizophrenia CNV datasets. Therefore, irrespective of the neural derived transcriptome, StMA genes originally identified in neural stem cells show an overrepresentation in CNVs associated with autism and schizophrenia. If this association is functional, then the regulation (or dysregulation) of this form of allelic expression status within tissues such as the brain may be a contributory risk factor for neurodevelopmental disorders and may also influence disease discordance sometimes observed in monozygotic twins.

'Particle genetics': treating every cell as unique

Genotype-phenotype relations are usually inferred from a deterministic point of view. For example, quantitative trait loci (QTL), which describe regions of the genome associated with a particular phenotype, are based on a mean trait difference between genotype categories. However, living systems comprise huge numbers of cells (the 'particles' of biology). Each cell can exhibit substantial phenotypic individuality, which can have dramatic consequences at the organismal level. Now, with technology capable of interrogating individual cells, it is time to consider how genotypes shape the probability laws of single cell traits. The possibility of mapping single cell probabilistic trait loci (PTL), which link genomic regions to probabilities of cellular traits, is a promising step in this direction. This approach requires thinking about phenotypes in probabilistic terms, a concept that statistical physicists have been applying to particles for a century. Here, I describe PTL and discuss their potential to enlarge our understanding of genotype-phenotype relations.

Selection is no more efficient in haploid than in diploid life stages of an angiosperm and a moss

The masking hypothesis predicts that selection is more efficient in haploids than in diploids, because dominant alleles can mask the deleterious effects of recessive alleles in diploids. However, gene expression breadth and noise can potentially counteract the effect of masking on the rate at which genes evolve. Land plants are ideal to ask whether masking, expression breadth, or expression noise dominate in their influence on the rate of molecular evolution, because they have a biphasic life cycle in which the duration and complexity of the haploid and diploid phase varies among organisms. Here, we generate and compile genome-wide gene expression, sequence divergence, and polymorphism data for Arabidopsis thaliana and for the moss Funaria hygrometrica to show that the evolutionary rates of haploid- and diploid-specific genes contradict the masking hypothesis. Haploid-specific genes do not evolve more slowly than diploid-specific genes in either organism. Our data suggest that gene expression breadth influence the evolutionary rate of phase-specific genes more strongly than masking. Our observations have implications for the role of haploid life stages in the purging of deleterious mutations, as well as for the evolution of ploidy.

Consequences of mRNA transport on stochastic variability in protein levels

Homogeneous cell populations can exhibit considerable cell-to-cell variability in protein levels arising from the stochastic nature of the gene-expression process. In particular, transcriptional bursting of mRNAs from the promoter has been implicated as a major source of stochasticity in the expression of many genes. In eukaryotes, transcribed pre-mRNAs have to be exported outside the nucleus and in many cases, export rates can be slow and comparable to mRNA turnover rates. We investigate whether such export processes can be effective mechanisms in buffering protein levels from transcriptional bursting of pre-mRNAs in the nucleus. For a stochastic gene-expression model with both transcriptional bursting and export, we derive an exact solution of the steady-state probability-generating function for both the nuclear and the cytoplasmic mRNA levels. These formulas reveal that decreasing export rates can dramatically reduce variability in cytoplasmic mRNA levels. However, our results also show that decreasing export rates enhance mRNA autocorrelation times, which function to increase heterogeneity in protein levels. Our overall analysis concludes that under physiologically relevant parameter regimes, a pre-mRNA export step can decrease steady-state variability at the mRNA level but not at the protein level. Finally, we reinforce previous observations that saturation in the pre-mRNA transport machinery can be an important mechanism in suppressing protein variability from underlying transcriptional bursts.

A survey of dominant mutations in Arabidopsis thaliana

Following the recent publication of a comprehensive dataset of 2400 genes with a loss-of-function mutant phenotype in Arabidopsis (Arabidopsis thaliana), questions remain concerning the diversity of dominant mutations in Arabidopsis. Most of these dominant phenotypes are expected to result from inappropriate gene expression, novel protein function, or disrupted protein complexes. This review highlights the major classes of dominant mutations observed in model organisms and presents a collection of 200 Arabidopsis genes associated with a dominant or semidominant phenotype. Emphasis is placed on mutants identified through forward genetic screens of mutagenized or activation-tagged populations. These datasets illustrate the variety of genetic changes and protein functions that underlie dominance in Arabidopsis and may ultimately contribute to phenotypic variation in flowering plants.

Inference of gene regulatory networks from genome-wide knockout fitness data

Motivation: Genome-wide fitness is an emerging type of high-throughput biological data generated for individual organisms by creating libraries of knockouts, subjecting them to broad ranges of environmental conditions, and measuring the resulting clone-specific fitnesses. Since fitness is an organism-scale measure of gene regulatory network behaviour, it may offer certain advantages when insights into such phenotypical and functional features are of primary interest over individual gene expression. Previous works have shown that genome-wide fitness data can be used to uncover novel gene regulatory interactions, when compared with results of more conventional gene expression analysis. Yet, to date, few algorithms have been proposed for systematically using genome-wide mutant fitness data for gene regulatory network inference.

Results: In this article, we describe a model and propose an inference algorithm for using fitness data from knockout libraries to identify underlying gene regulatory networks. Unlike most prior methods, the presented approach captures not only structural, but also dynamical and non-linear nature of biomolecular systems involved. A state–space model with non-linear basis is used for dynamically describing gene regulatory networks. Network structure is then elucidated by estimating unknown model parameters. Unscented Kalman filter is used to cope with the non-linearities introduced in the model, which also enables the algorithm to run in on-line mode for practical use. Here, we demonstrate that the algorithm provides satisfying results for both synthetic data as well as empirical measurements of GAL network in yeast Saccharomyces cerevisiae and TyrR–LiuR network in bacteria Shewanella oneidensis.

Availability: MATLAB code and datasets are available to download at http://www.duke.edu/∼lw174/Fitness.zip and http://genomics.lbl.gov/supplemental/fitness-bioinf/

Contact:wangx@ee.columbia.edu or mssamoilov@lbl.gov

Supplementary information:Supplementary data are available at Bioinformatics online

Modeling regulation of zinc uptake via ZIP transporters in yeast and plant roots

In yeast (Saccharomyces cerevisiae) and plant roots (Arabidopsis thaliana) zinc enters the cells via influx transporters of the ZIP family. Since zinc is both essential for cell function and toxic at high concentrations, tight regulation is essential for cell viability. We provide new insight into the underlying mechanisms, starting from a general model based on ordinary differential equations and adapting it to the specific cases of yeast and plant root cells. In yeast, zinc is transported by the transporters ZRT1 and ZRT2, which are both regulated by the zinc-responsive transcription factor ZAP1. Using biological data, parameters were estimated and analyzed, confirming the different affinities of ZRT1 and ZRT2 reported in the literature. Furthermore, our model suggests that the positive feedback in ZAP1 production has a stabilizing function at high influx rates. In plant roots, various ZIP transporters play a role in zinc uptake. Their regulation is largely unknown, but bZIP transcription factors are thought to be involved. We set up three putative models based on: an activator only, an activator with dimerization and an activator-inhibitor pair. These were fitted to measurements and analyzed. Simulations show that the activator-inhibitor model outperforms the other two in providing robust and stable homeostasis at reasonable parameter ranges.

Negative feedback through mRNA provides the best control of gene-expression noise

Genetically identical cell populations exposed to the same environment can exhibit considerable cell-to-cell variation in the levels of specific proteins. This variation or expression noise arises from the inherent stochastic nature of biochemical reactions that constitute gene expression. Negative feedback loops are common motifs in gene networks that reduce expression noise and intercellular variability in protein levels. Using stochastic models of gene expression we here compare different feedback architectures in their ability to reduce stochasticity in protein levels. A mathematically controlled comparison shows that in physiologically relevant parameter regimes, feedback regulation through the mRNA provides the best suppression of expression noise. Consistent with our theoretical results we find negative feedback loops though the mRNA in essential eukaryotic genes, where feedback is mediated via intron-derived microRNAs. Finally, we find that contrary to previous results, protein-mediated translational regulation may not always provide significantly better noise suppression than protein-mediated transcriptional regulation.