RNA-SEQUENCING APPLICATIONS: GENE EXPRESSION QUANTIFICATION AND METHYLATOR PHENOTYPE IDENTIFICATION

My dissertation focuses on two aspects of RNA sequencing technology. The first is the methodology for modeling the overdispersion inherent in RNA-seq data for differential expression analysis. This aspect is addressed in three sections. The second aspect is the application of RNA-seq data to identify the CpG island methylator phenotype (CIMP) by integrating datasets of mRNA expression level and DNA methylation status. Section 1: The cost of DNA sequencing has reduced dramatically in the past decade. Consequently, genomic research increasingly depends on sequencing technology. However it remains elusive how the sequencing capacity influences the accuracy of mRNA expression measurement. We observe that accuracy improves along with the increasing sequencing depth. To model the overdispersion, we use the beta-binomial distribution with a new parameter indicating the dependency between overdispersion and sequencing depth. Our modified beta-binomial model performs better than the binomial or the pure beta-binomial model with a lower false discovery rate. Section 2: Although a number of methods have been proposed in order to accurately analyze differential RNA expression on the gene level, modeling on the base pair level is required. Here, we find that the overdispersion rate decreases as the sequencing depth increases on the base pair level. Also, we propose four models and compare them with each other. As expected, our beta binomial model with a dynamic overdispersion rate is shown to be superior. Section 3: We investigate biases in RNA-seq by exploring the measurement of the external control, spike-in RNA. This study is based on two datasets with spike-in controls obtained from a recent study. We observe an undiscovered bias in the measurement of the spike-in transcripts that arises from the influence of the sample transcripts in RNA-seq. Also, we find that this influence is related to the local sequence of the random hexamer that is used in priming. We suggest a model of the inequality between samples and to correct this type of bias. Section 4: The expression of a gene can be turned off when its promoter is highly methylated. Several studies have reported that a clear threshold effect exists in gene silencing that is mediated by DNA methylation. It is reasonable to assume the thresholds are specific for each gene. It is also intriguing to investigate genes that are largely controlled by DNA methylation. These genes are called “L-shaped” genes. We develop a method to determine the DNA methylation threshold and identify a new CIMP of BRCA. In conclusion, we provide a detailed understanding of the relationship between the overdispersion rate and sequencing depth. And we reveal a new bias in RNA-seq and provide a detailed understanding of the relationship between this new bias and the local sequence. Also we develop a powerful method to dichotomize methylation status and consequently we identify a new CIMP of breast cancer with a distinct classification of molecular characteristics and clinical features.

Epigenetic variability in the genetically uniform forest tree species Pinus pinea L

There is an increasing interest in understanding the role of epigenetic variability in forest species and how it may contribute to their rapid adaptation to changing environments. In this study we have conducted a genome-wide analysis of cytosine methylation pattern in Pinus pinea, a species characterized by very low levels of genetic variation and a remarkable degree of phenotypic plasticity. DNA methylation profiles of different vegetatively propagated trees from representative natural Spanish populations of P. pinea were analyzed with the Methylation Sensitive Amplified Polymorphism (MSAP) technique. A high degree of cytosine methylation was detected (64.36% of all scored DNA fragments). Furthermore, high levels of epigenetic variation were observed among the studied individuals. This high epigenetic variation found in P. pinea contrasted with the lack of genetic variation based on Amplified Fragment Length Polymorphism (AFLP) data. In this manner, variable epigenetic markers clearly discriminate individuals and differentiates two well represented populations while the lack of genetic variation revealed with the AFLP markers fail to differentiate at both, individual or population levels. In addition, the use of different replicated trees allowed identifying common polymorphic methylation sensitive MSAP markers among replicates of a given propagated tree. This set of MSAPs allowed discrimination of the 70% of the analyzed trees.

Incidence and functional consequences of hMLH1 promoter hypermethylation in colorectal carcinoma

Inactivation of the genes involved in DNA mismatch repair is associated with microsatellite instability (MSI) in colorectal cancer. We report that hypermethylation of the 5′ CpG island of hMLH1 is found in the majority of sporadic primary colorectal cancers with MSI, and that this methylation was often, but not invariably, associated with loss of hMLH1 protein expression. Such methylation also occurred, but was less common, in MSI− tumors, as well as in MSI+ tumors with known mutations of a mismatch repair gene (MMR). No hypermethylation of hMSH2 was found. Hypermethylation of colorectal cancer cell lines with MSI also was frequently observed, and in such cases, reversal of the methylation with 5-aza-2′-deoxycytidine not only resulted in reexpression of hMLH1 protein, but also in restoration of the MMR capacity in MMR-deficient cell lines. Our results suggest that microsatellite instability in sporadic colorectal cancer often results from epigenetic inactivation of hMLH1 in association with DNA methylation.

Epigenetic phenotypes distinguish microsatellite-stable and -unstable colorectal cancers

Aberrant DNA methylation is a common phenomenon in human cancer, but its patterns, causes, and consequences are poorly defined. Promoter methylation of the DNA mismatch repair gene MutL homologue (MLH1) has been implicated in the subset of colorectal cancers that shows microsatellite instability (MSI). The present analysis of four MspI/HpaII sites at the MLH1 promoter region in a series of 89 sporadic colorectal cancers revealed two main methylation patterns that closely correlated with the MSI status of the tumors. These sites were hypermethylated in tumor tissue relative to normal mucosa in most MSI(+) cases (31/51, 61%). By contrast, in the majority of MSI(−) cases (20/38, 53%) the same sites showed methylation in normal mucosa and hypomethylation in tumor tissue. Hypermethylation displayed a direct correlation with increasing age and proximal location in the bowel and was accompanied by immunohistochemically documented loss of MLH1 protein both in tumors and in normal tissue. Similar patterns of methylation were observed in the promoter region of the calcitonin gene that does not have a known functional role in tumorigenesis. We propose a model of carcinogenesis where different epigenetic phenotypes distinguish the colonic mucosa in individuals who develop MSI(+) and MSI(−) tumors. These phenotypes may underlie the different developmental pathways that are known to occur in these tumors.

Histone underacetylation is an ancient component of mammalian X chromosome inactivation

Underacetylation of histone H4 is thought to be involved in the molecular mechanism of mammalian X chromosome inactivation, which is an important model system for large-scale genetic control in eukaryotes. However, it has not been established whether histone underacetylation plays a critical role in the multistep inactivation pathway. Here we demonstrate differential histone H4 acetylation between the X chromosomes of a female marsupial, Macropus eugenii. Histone underacetylation is the only molecular aspect of X inactivation known to be shared by marsupial and eutherian mammals. Its strong evolutionary conservation implies that, unlike DNA methylation, histone underacetylation was a feature of dosage compensation in a common mammalian ancestor, and is therefore likely to play a central role in X chromosome inactivation in all mammals.

A model system to study genomic imprinting of human genes

Somatic-cell hybrids have been shown to maintain the correct epigenetic chromatin states to study developmental globin gene expression as well as gene expression on the active and inactive X chromosomes. This suggests the potential use of somatic-cell hybrids containing either a maternal or a paternal human chromosome as a model system to study known imprinted genes and to identify as-yet-unknown imprinted genes. Testing gene expression by using reverse transcription followed by PCR, we show that functional imprints are maintained at four previously characterized 15q11–q13 loci in hybrids containing a single human chromosome 15 and at two chromosome 11p15 loci in hybrids containing a single chromosome 11. In contrast, three γ-aminobutyric acid type A receptor subunit genes in 15q12–q13 are nonimprinted. Furthermore, we have found that differential DNA methylation imprints at the SNRPN promoter and at a CpG island in 11p15 are also maintained in somatic-cell hybrids. Somatic-cell hybrids therefore are a valid and powerful system for studying known imprinted genes as well as for rapidly identifying new imprinted genes.

Nuclear Accumulation of S-Adenosylhomocysteine Hydrolase in Transcriptionally Active Cells during Development of Xenopus laevis

The oocyte nuclear antigen of the monoclonal antibody 32-5B6 of Xenopus laevis is subject to regulated nuclear translocation during embryogenesis. It is distributed in the cytoplasm during oocyte maturation, where it remains during cleavage and blastula stages, before it gradually reaccumulates in the nuclei during gastrulation. We have now identified this antigen to be the enzyme S-adenosylhomocysteine hydrolase (SAHH). SAHH is the only enzyme that cleaves S-adenosylhomocysteine, a reaction product and an inhibitor of all S-adenosylmethionine-dependent methylation reactions. We have compared the spatial and temporal patterns of nuclear localization of SAHH and of nuclear methyltransferase activities during embryogenesis and in tissue culture cells. Nuclear localization of Xenopus SAHH did not temporally correlate with DNA methylation. However, we found that SAHH nuclear localization coincides with high rates of mRNA synthesis, a subpopulation colocalizes with RNA polymerase II, and inhibitors of SAHH reduce both methylation and synthesis of poly(A)+ RNA. We therefore propose that accumulation of SAHH in the nucleus may be required for efficient cap methylation in transcriptionally active cells. Mutation analysis revealed that the C terminus and the N terminus are both required for efficient nuclear translocation in tissue culture cells, indicating that more than one interacting domain contributes to nuclear accumulation of Xenopus SAHH.

V(D)J recombination is not activated by demethylation of the kappa locus

V(D)J recombination is thought to be regulated by changes in the accessibility of target sites, such as modulation of methylation. To test whether demethylation of the kappa locus can activate recombination, we generated two recombinationally active B cell lines in which the gene for maintenance of genomic DNA methylation, Dnmt1, was flanked with loxP sites. Transduction with a retrovirus expressing both the cre recombinase and green fluorescent protein allowed us to purify recombinationally active cells devoid of methylation. Loss of methylation of the kappa locus was not sufficient to activate recombination, although transcription was activated in one line. It appears that demethylation of the kappa locus is not the rate-limiting step for altering accessibility and thus regulated demethylation does not generate specificity of recombination.

Expression and parent-of-origin effects for FIS2, MEA, and FIE in the endosperm and embryo of developing Arabidopsis seeds

The promoters of MEA (FIS1), FIS2, and FIE (FIS3), genes that repress seed development in the absence of pollination, were fused to β-glucuronidase (GUS) to study their activity pattern. The FIS2∷GUS product is found in the embryo sac, in each of the polar cell nuclei, and in the central cell nucleus. After pollination, the maternally derived FIS2∷GUS protein occurs in the nuclei of the cenocytic endosperm. Before cellularization of the endosperm, activity is terminated in the micropylar and central nuclei of the endosperm and subsequently in the nuclei of the chalazal cyst. MEA∷GUS has a pattern of activity similar to that of FIS2∷GUS, but FIE∷GUS protein is found in many tissues, including the prepollination embryo sac, and in embryo and endosperm postpollination. The similarity in mutant phenotypes; the activity of FIE, MEA, and FIS2 in the same cells in the embryo sac; and the fact that MEA and FIE proteins interact in a yeast two-hybrid system suggest that these proteins operate in the same system of control of seed development. Maternal and not paternal FIS2∷GUS, MEA∷GUS, and FIE∷GUS show activity in early endosperm, so these genes may be imprinted. When fis2, mea, and fie mutants are pollinated, seed development is arrested at the heart embryo stage. The seed arrest of mea and fis2 is avoided when they are fertilized by a low methylation parent. The wild-type alleles of MEA or FIS2 are not required. The parent-of-origin-determined differential activity of MEA, FIS2, and FIE is not dependent on DNA methylation, but methylation does control some gene(s) that have key roles in seed development.

Complete genome sequence of Caulobacter crescentus

The complete genome sequence of Caulobacter crescentus was determined to be 4,016,942 base pairs in a single circular chromosome encoding 3,767 genes. This organism, which grows in a dilute aquatic environment, coordinates the cell division cycle and multiple cell differentiation events. With the annotated genome sequence, a full description of the genetic network that controls bacterial differentiation, cell growth, and cell cycle progression is within reach. Two-component signal transduction proteins are known to play a significant role in cell cycle progression. Genome analysis revealed that the C. crescentus genome encodes a significantly higher number of these signaling proteins (105) than any bacterial genome sequenced thus far. Another regulatory mechanism involved in cell cycle progression is DNA methylation. The occurrence of the recognition sequence for an essential DNA methylating enzyme that is required for cell cycle regulation is severely limited and shows a bias to intergenic regions. The genome contains multiple clusters of genes encoding proteins essential for survival in a nutrient poor habitat. Included are those involved in chemotaxis, outer membrane channel function, degradation of aromatic ring compounds, and the breakdown of plant-derived carbon sources, in addition to many extracytoplasmic function sigma factors, providing the organism with the ability to respond to a wide range of environmental fluctuations. C. crescentus is, to our knowledge, the first free-living α-class proteobacterium to be sequenced and will serve as a foundation for exploring the biology of this group of bacteria, which includes the obligate endosymbiont and human pathogen Rickettsia prowazekii, the plant pathogen Agrobacterium tumefaciens, and the bovine and human pathogen Brucella abortus.

Posttranscriptional Gene Silencing in Transgenic Sugarcane. Dissection of Homology-Dependent Virus Resistance in a Monocot That Has a Complex Polyploid Genome1

RNA-mediated, posttranscriptional gene silencing has been determined as the molecular mechanism underlying transgenic virus resistance in many plant virus-dicot host plant systems. In this paper we show that transgenic virus resistance in sugarcane (Saccharum spp. hybrid) is based on posttranscriptional gene silencing. The resistance is derived from an untranslatable form of the sorghum mosaic potyvirus strain SCH coat protein (CP) gene. Transgenic sugarcane plants challenged with sorghum mosaic potyvirus strain SCH had phenotypes that ranged from fully susceptible to completely resistant, and a recovery phenotype was also observed. Clones derived from the same transformation event or obtained after vegetative propagation could display different levels of virus resistance, suggesting the involvement of a quantitative component in the resistance response. Most resistant plants displayed low or undetectable steady-state CP transgene mRNA levels, although nuclear transcription rates were high. Increased DNA methylation was observed in the transcribed region of the CP transgenes in most of these plants. Collectively, these characteristics indicate that an RNA-mediated, homology-dependent mechanism is at the base of the virus resistance. This work extends posttranscriptional gene silencing and homology-dependent virus resistance, so far observed only in dicots, to an agronomically important, polyploid monocot.

Selective association of the methyl-CpG binding protein MBD2 with the silent p14/p16 locus in human neoplasia

DNA methylation of tumor suppressor genes is a common feature of human cancer. The cyclin-dependent kinase inhibitor gene p16/Ink4A is hypermethylated in a wide range of malignant tissues and the p14/ARF gene located 20 kb upstream on chromosome 9p21 is also methylated in carcinomas. p14/ARF (ARF, alternative reading frame) does not inhibit the activities of cyclins or cyclin-dependent kinase complexes; however, the importance of the two gene products in the etiology of cancer resides in their involvement in two major cell cycle regulatory pathways: p53 and the retinoblastoma protein, Rb, respectively. Distinct first exons driven from separate promoters are spliced onto the common exons 2 and 3 and the resulting proteins are translated in different reading frames. Both genes are expressed in normal cells but can be alternatively or coordinately silenced when their CpG islands are hypermethylated. Herein, we examined the presence of methyl-CpG binding proteins associated with aberrantly methylated promoters, the distribution of acetylated histones H3 and H4 by chromatin immunoprecipitation assays, and the effect of chemical treatment with 5-aza-2′-deoxycytidine (5aza-dC) and trichostatin A on gene induction in colon cell lines by quantitative reverse transcriptase–PCR. We observed that the methyl-CpG binding protein MBD2 is targeted to methylated regulatory regions and excludes the acetylated histones H3 and H4, resulting in a localized inactive chromatin configuration. When methylated, the genes can be induced by 5aza-dC but the combined action of 5aza-dC and trichostatin A results in robust gene expression. Thus, methyl-CpG binding proteins and histone deacetylases appear to cooperate in vivo, with a dominant effect of DNA methylation toward histone acetylation, and repress expression of tumor suppressor genes hypermethylated in cancers.

Methionine adenosyltransferase 1A knockout mice are predisposed to liver injury and exhibit increased expression of genes involved in proliferation

Liver-specific and nonliver-specific methionine adenosyltransferases (MATs) are products of two genes, MAT1A and MAT2A, respectively, that catalyze the formation of S-adenosylmethionine (AdoMet), the principal biological methyl donor. Mature liver expresses MAT1A, whereas MAT2A is expressed in extrahepatic tissues and is induced during liver growth and dedifferentiation. To examine the influence of MAT1A on hepatic growth, we studied the effects of a targeted disruption of the murine MAT1A gene. MAT1A mRNA and protein levels were absent in homozygous knockout mice. At 3 months, plasma methionine level increased 776% in knockouts. Hepatic AdoMet and glutathione levels were reduced by 74 and 40%, respectively, whereas S-adenosylhomocysteine, methylthioadenosine, and global DNA methylation were unchanged. The body weight of 3-month-old knockout mice was unchanged from wild-type littermates, but the liver weight was increased 40%. The Affymetrix genechip system and Northern and Western blot analyses were used to analyze differential expression of genes. The expression of many acute phase-response and inflammatory markers, including orosomucoid, amyloid, metallothionein, Fas antigen, and growth-related genes, including early growth response 1 and proliferating cell nuclear antigen, is increased in the knockout animal. At 3 months, knockout mice are more susceptible to choline-deficient diet-induced fatty liver. At 8 months, knockout mice developed spontaneous macrovesicular steatosis and predominantly periportal mononuclear cell infiltration. Thus, absence of MAT1A resulted in a liver that is more susceptible to injury, expresses markers of an acute phase response, and displays increased proliferation.

Protein-coding genes are epigenetically regulated in Arabidopsis polyploids

The fate of redundant genes resulting from genome duplication is poorly understood. Previous studies indicated that ribosomal RNA genes from one parental origin are epigenetically silenced during interspecific hybridization or polyploidization. Regulatory mechanisms for protein-coding genes in polyploid genomes are unknown, partly because of difficulty in studying expression patterns of homologous genes. Here we apply amplified fragment length polymorphism (AFLP)–cDNA display to perform a genome-wide screen for orthologous genes silenced in Arabidopsis suecica, an allotetraploid derived from Arabidopsis thaliana and Cardaminopsis arenosa. We identified ten genes that are silenced from either A. thaliana or C. arenosa origin in A. suecica and located in four of the five A. thaliana chromosomes. These genes represent a variety of RNA and predicted proteins including four transcription factors such as TCP3. The silenced genes in the vicinity of TCP3 are hypermethylated and reactivated by blocking DNA methylation, suggesting epigenetic regulation is involved in the expression of orthologous genes in polyploid genomes. Compared with classic genetic mutations, epigenetic regulation may be advantageous for selection and adaptation of polyploid species during evolution and development.

Involvement of small GTP-binding proteins in defense signal-transduction pathways of higher plants.

Small GTP-binding proteins play a critical role in the regulation of a range of cellular processes--including growth, differentiation, and intracellular transportation. Previously, we isolated a gene, rgp1, encoding a small GTP-binding protein, by differential screening of a rice cDNA library with probe DNAs from rice tissues treated with or without 5-azacytidine, a powerful inhibitor of DNA methylation. To determine the physiological role of rgp1, the coding region was introduced into tobacco plants. Transformants, with rgp1 in either sense or antisense orientations, showed distinct phenotypic changes with reduced apical dominance, dwarfism, and abnormal flower development. These abnormal phenotypes appeared to be associated with the higher levels of endogenous cytokinins that were 6-fold those of wild-type plants. In addition, the transgenic plants produced salicylic acid and salicylic acid-beta-glucoside in an unusual response to wounding, thus conferring increased resistance to tobacco mosaic virus infection. In normal plants, the wound- and pathogen-induced signal-transduction pathways are considered to function independently. However, the wound induction of salicylic acid in the transgenic plants suggests that expression of rgp1 somehow interfered with the normal signaling pathways and resulted in cross-signaling between these distinct transduction systems. The results imply that the defense signal-transduction system consists of a complicated and finely tuned network of several regulatory factors, including cytokinins, salicylic acid, and small GTP-binding proteins.