Overdispersion models for correlated multinomial data: Applications to blinding assessment.

Overdispersion models have been extensively studied for correlated normal and binomial data but much less so for correlated multinomial data. In this work, we describe a multinomial overdispersion model that leads to the specification of the first two moments of the outcome and allows the estimation of the global parameters using generalized estimating equations (GEE). We introduce a Global Blinding Index as a target parameter and illustrate the application of the GEE method to its estimation from (1) a clinical trial with clustering by practitioner and (2) a meta-analysis on psychiatric disorders. We examine the impact of a small number of clusters, high variability in cluster sizes, and the magnitude of the intraclass correlation on the performance of the GEE estimators of the Global Blinding Index using the data simulated from different models. We compare these estimators with the inverse-variance weighted estimators and a maximum-likelihood estimator, derived under the Dirichlet-multinomial model. Our results indicate that the performance of the GEE estimators was satisfactory even in situations with a small number of clusters, whereas the inverse-variance weighted estimators performed poorly, especially for larger values of the intraclass correlation coefficient. Our findings and illustrations may be instrumental for practitioners who analyze clustered multinomial data from clinical trials and/or meta-analysis.

TLX1/HOX11-induced hematopoietic differentiation blockade.

Aberrant expression of the human homeobox-containing proto-oncogene TLX1/HOX11 inhibits hematopoietic differentiation programs in a number of murine model systems. Here, we report the establishment of a murine erythroid progenitor cell line, iEBHX1S-4, developmentally arrested by regulatable TLX1 expression. Extinction of TLX1 expression released the iEBHX1S-4 differentiation block, allowing erythropoietin-dependent acquisition of erythroid markers and hemoglobin synthesis. Coordinated activation of erythroid transcriptional networks integrated by the acetyltransferase co-activator CREB-binding protein (CBP) was suggested by bioinformatic analysis of the upstream regulatory regions of several conditionally induced iEBHX1S-4 gene sets. In accord with this notion, CBP-associated acetylation of GATA-1, an essential regulator of erythroid differentiation, increased concomitantly with TLX1 downregulation. Coimmunoprecipitation experiments and glutathione-S-transferase pull-down assays revealed that TLX1 directly binds to CBP, and confocal laser microscopy demonstrated that the two proteins partially colocalize at intranuclear sites in iEBHX1S-4 cells. Notably, the distribution of CBP in conditionally blocked iEBHX1S-4 cells partially overlapped with chromatin marked by a repressive histone methylation pattern, and downregulation of TLX1 coincided with exit of CBP from these heterochromatic regions. Thus, we propose that TLX1-mediated differentiation arrest may be achieved in part through a mechanism that involves redirection of CBP and/or its sequestration in repressive chromatin domains.

Assessment of the transcriptional activation potential of the HMG chromosomal proteins.

Chromosomal proteins HMG-14, HMG-17, and HMG-1 are among the most abundant, ubiquitous, and evolutionarily conserved nonhistone proteins. Analysis of their structure reveals features which are similar to those of certain transcription factors. The distribution of charged amino acid residues along the polypeptide chains is asymmetric: positive charges are clustered toward the N-terminal region, while negative charges are clustered toward the C-terminal region. The residues in the C-terminal region have the potential to form alpha helices with negatively charged surfaces. The abilities of HMG-14, -17, and -1 to function as transcriptional activators were studied in Saccharomyces cerevisiae cells expressing LexA-HMG fusion proteins (human HMG-14 and -17 and rat HMG-1) which bind to reporter molecules containing the beta-galactosidase gene downstream from a lexA operator. Fusion constructs expressing deletion mutants of HMG-14, -17, and -1 were also tested. Analysis of binding to the lexA operator with in vitro-synthesized fusion proteins shows that there are more sites for HMG-14, -17, and -1 binding than for LexA binding and that only the fusion constructs which contain the C-terminal, acidic domains of HMG-17 bind the lexA operator specifically. None of the LexA-HMG fusion protein constructs elevate the level of beta-galactosidase activity in transfected yeast cells. Thus, although HMG-14, -17, and -1 are structurally similar to acidic transcriptional activators, these chromosomal proteins do not function as activators in this test system.

PROSPECT improves cis-acting regulatory element prediction by integrating expression profile data with consensus pattern searches.

Consensus pattern and matrix-based searches designed to predict cis-acting transcriptional regulatory sequences have historically been subject to large numbers of false positives. We sought to decrease false positives by incorporating expression profile data into a consensus pattern-based search method. We have systematically analyzed the expression phenotypes of over 6000 yeast genes, across 121 expression profile experiments, and correlated them with the distribution of 14 known regulatory elements over sequences upstream of the genes. Our method is based on a metric we term probabilistic element assessment (PEA), which is a ranking of potential sites based on sequence similarity in the upstream regions of genes with similar expression phenotypes. For eight of the 14 known elements that we examined, our method had a much higher selectivity than a naïve consensus pattern search. Based on our analysis, we have developed a web-based tool called PROSPECT, which allows consensus pattern-based searching of gene clusters obtained from microarray data.

Dataset of Arabidopsis plants that overexpress FT driven by a meristem-specific KNAT1 promoter.

In this dataset we integrated figures comparing leaf number and rosette diameter in three Arabidopsis FT overexpressor lines (AtFTOE) driven by KNAT1 promoter, "A member of the KNOTTED class of homeodomain proteins encoded by the STM gene of Arabidopsis" [5], vs Wild Type (WT) Arabidopsis plats. Also, presented in the tables are some transcriptomic data obtained by RNA-seq Illumina HiSeq from rosette leaves of Arabidopsis plants of AtFTOE 2.1 line vs WT with accession numbers SRR2094583 and SRR2094587 for AtFTOE replicates 1-3 and AtWT for control replicates 1-2 respectively. Raw data of paired-end sequences are located in the public repository of the National Center for Biotechnology Information of the National Library of Medicine, National Institutes of Health, United States of America, Bethesda, MD, USA as Sequence Read Archive (SRA). Performed analyses of differential expression genes are visualized by Mapman and presented in figures. "Transcriptomic analysis of Arabidopsis overexpressing flowering locus T driven by a meristem-specific promoter that induces early flowering" [2], described the interpretation and discussion of the obtained data.

Transcriptomic analysis of Arabidopsis overexpressing flowering locus T driven by a meristem-specific promoter that induces early flowering.

Here we analyzed in leaves the effect of FT overexpression driven by meristem-specific KNAT1 gene homolog of Arabidopsis thaliana (Lincoln et al., 1994; Long et al., 1996) on the transcriptomic response during plant development. Our results demonstrated that meristematic FT overexpression generates a phenotype with an early flowering independent of photoperiod when compared with wild type (WT) plants. Arabidopsis FT-overexpressor lines (AtFTOE) did not show significant differences compared with WT lines neither in leaf number nor in rosette diameter up to day 21, when AtFTOE flowered. After this period AtFTOE plants started flower production and no new rosette leaves were produced. Additionally, WT plants continued on vegetative stage up to day 40, producing 12-14 rosette leaves before flowering. Transcriptomic analysis of rosette leaves studied by sequencing Illumina RNA-seq allowed us to determine the differential expression in mature leaf rosette of 3652 genes, being 626 of them up-regulated and 3026 down-regulated. Overexpressed genes related with flowering showed up-regulated transcription factors such as MADS-box that are known as flowering markers in meristem and which overexpression has been related with meristem identity preservation and the transition from vegetative to floral stage. Genes related with sugar transport have shown a higher demand of monosaccharides derived from the hydrolysis of sucrose to glucose and probably fructose, which can also be influenced by reproductive stage of AtFTOE plants.

Transposable elements donate lineage-specific regulatory sequences to host genomes.

The evolutionary implications of transposable element (TE) influences on gene regulation are explored here. An historical perspective is presented to underscore the importance of TE influences on gene regulation with respect to both the discovery of TEs and the early conceptualization of their potential impact on host genome evolution. Evidence that points to a role for TEs in host gene regulation is reviewed, and comparisons between genome sequences are used to demonstrate the fact that TEs are particularly lineage-specific components of their host genomes. Consistent with these two properties of TEs, regulatory effects and evolutionary specificity, human-mouse genome wide sequence comparisons reveal that the regulatory sequences that are contributed by TEs are exceptionally lineage specific. This suggests a particular mechanism by which TEs may drive the diversification of gene regulation between evolutionary lineages.

Retropseudogenes for human chromosomal protein HMG-17.

The human genome contains multiple copies of sequences homologous to the cDNA coding for non-histone chromosomal protein HMG-17. To study the mechanism of generation and dispersion of the HMG-17 multigene family a human genomic library was screened and 70 clones isolated and studied by Southern transfer and restriction site analysis. The results suggest that most of the clones contain unique sequences. Sequence analysis of two genomic clones indicates that they contain elements typical of processed retropseudogenes. Even though both sequences contained open reading frames the sequences lacked introns, were flanked by short, direct repeats and lacked elements associated with functional genes. The sequences of the two pseudogenes were 85% homologous to each other and each was 90% homologous to the human cDNA. Based on the sequence difference in the open reading frame between the pseudogenes and the cDNA it can be estimated that the sequences arose approximately ten million years ago from a common precursor. The present paper, which is the first study on genes coding for this nucleosomal binding protein, indicates that the HMG-17 multigene family is the largest known human retropseudogene family.

A single copy gene for chicken chromosomal protein HMG-14b has evolutionarily conserved features, has lost one of its introns and codes for a rapidly evolving protein.

The evolutionary origins and common features of the genes coding for the HMG-14/-17 family of chromosomal proteins have been studied by isolating and sequencing the chicken HMG-14b gene, the true homolog of the human and calf HMG-14 gene. Comparison of the structure of this gene to that of the human HMG-14 gene and to the human and chicken HMG-17 genes indicates that the HMG-14 and HMG-17 genes evolved from a common ancestor. We postulate that the ancestral gene consisted of six exons. In all genes the first exon codes for the entire 5' untranslated region and for the first four amino acids, which are invariant among all the known members of the HMG-14/-17 protein family. The last exon codes for ten to 16 amino acids and for the entire 3' untranslated region, which, for each gene, constitutes over 70% of the transcript. The DNA-binding domain of the proteins is encoded by two distinct exons. The genes are characterized by 5' regions that are highly enriched in G + C residues and have features characteristic of "housekeeping" genes. The HMG-17 genes are distinct from the HMG-14 in that the 5' regulatory region of the former has two TATA boxes while the HMG-14 genes have no such regulatory element. The chicken HMG-14b gene is a single-copy gene and produces a unique transcript. In this gene, exons II and III are fused and intron 2 is missing. The fusion of the two exons produced a codon for valine in a position that, among all HMG-14/-17 proteins, is unique to HMG-14b. The possible consequences of a valine insertion at the N-terminal end of the DNA-binding domains are discussed. The HMG-14 proteins evolve significantly faster than HMG-17, suggesting that the proteins are subject to different evolutionary pressure. However, certain amino acids are conserved among all the known members of the HMG-14/-17 protein family, suggesting that they are part of the functional domain of this family of chromosomal proteins.

Analysis of the functional role of conserved residues in the protein subunit of ribonuclease P from Escherichia coli.

The processing of precursor tRNAs and some other small cellular RNAs by M1 RNA, the catalytic subunit of Escherichia coli ribonuclease P, is accelerated by C5 protein (the protein cofactor) both in vitro and in vivo. In an effort to understand the mechanism by which the protein cofactor promotes and stabilizes certain conformations of M1 RNA that are most efficient for RNase P catalysis, we have used site-directed mutagenesis to generate mutant derivatives of C5 protein and assessed their ability to promote RNase P catalysis in vivo and in vitro. Our results indicate that certain conserved hydrophobic and basic residues in C5 protein are important for its function and that single amino acid residue changes in C5 protein can alter the substrate specificity of the RNase P holoenzyme.

HMGN4, a newly discovered nucleosome-binding protein encoded by an intronless gene.

We describe a newly discovered nuclear protein, HMGN4, that is closely related to the canonical HMGN2 nucleosome-binding protein. The protein is encoded by an intronless gene, which, in humans, is located in the hereditary hemochromatosis [correction of hemachromatosis] region at position 6p21.3. A single approximately 2-kb HMGN4 mRNA was found to be expressed, in variable amounts, in all human tissues tested; however, the HMGN4 transcript was significantly less abundant than that of HMGN2. The HMGN4 protein could be detected in HeLa cells by Western analysis with an antibody elicited against a unique region of the protein. Transfection of HeLa cells with a plasmid expressing HMGN4-GFP indicated that the protein localizes to the nucleus. Our results expand the multiplicity of the HMGN protein family and increase the known cellular repertoire of nucleosome-binding proteins.

RNP-1, an RNA-binding motif is conserved in the DNA-binding cold shock domain.

Sequence analysis has shown that there is a short motif of 8 amino acids, corresponding to the RNP-1 motif found in canonical RNA-binding domains, which is common to two families of apparently unrelated proteins. Many RNA-binding proteins contain the RNP-1 and RNP-2 motifs in an RNA-binding domain. The cold shock domain (CSD) family of proteins, which includes several transcription factors which have been shown to bind to DNA, has now been identified to contain a motif similar to RNP-1. A non-redundant protein sequence database was searched with regular expressions and with a weight/residue position matrix of the RNP-1 motif resulting in the identification of numerous known members of the RNA-binding family of proteins. In addition, the search identified that the CSD-containing family of proteins includes a motif which is almost identical to the RNP-1 motif. A determination of the statistical significance of this analysis showed that the RNP-1 motifs from these two families of proteins are indeed similar.

Chromosomal proteins HMG-14 and HMG-17. Distinct multigene families coding for similar types of transcripts.

Human nonhistone chromosomal proteins HMG-14 and HMG-17 are encoded by genes which are part of multigene families. Southern analysis of human, mouse, and rat genomic restriction digests reveals that the two families are distinct. Although the cDNAs of HMG-14 and HMG-17 do not cross-hybridize, they have several similar structural features: the open reading frame comprises only 23% of the transcripts, the 5'-untranslated region is extremely GC rich whereas the 3'-untranslated region is unusually long and AT rich. The overall sequence homology between the two transcripts is highest (71%) in the 90 nucleotides coding for the DNA-binding domains of the proteins. The sequence of the human HMG-14 and HMG-17 proteins, deduced from the open reading frame, differs by more than 50%; the DNA-binding domains of the proteins show 74% sequence homology. However, even in this 30-residue long peptide there are significant differences between the proteins as the proline content of HMG-17 (8 residues) is twice that of HMG-14. The two proteins have different hydropathy index profiles and are serologically distinct. The multigene families may have evolved independently from similar genetic elements or from a shared ancestral gene in which the nucleotide sequence coding for the DNA-binding domain of the protein is the most conserved region. The structural differences between the molecules and the differences in their DNA-binding domains suggest that the proteins may be involved in distinguishable cellular functions.

AT-hook motifs identified in a wide variety of DNA-binding proteins.

The AT-hook is a small DNA-binding protein motif which was first described in the high mobility group non-histone chromosomal protein HMG-I(Y). Since its discovery, this motif has been observed in other DNA-binding proteins from a wide range of organisms. Using pattern searches and position-dependent matrices, we have extracted the AT-hook motifs present in a non-redundant protein sequence database. We have classified these motifs into three types according to their sequence similarity and have found that they are prevalent in many eukaryotic nuclear proteins in single or multiple copies. Furthermore, AT-hook motifs are frequently associated with known functional domains seen in chromatin proteins and in DNA-binding proteins (e.g. histone folds, homeodomains and zinc fingers). In general, it appears that the AT-hook motif is an auxiliary protein motif cooperating with other DNA-binding activities and facilitating changes in the structure of the DNA either as a polypeptide on its own [e.g. HMG-I(Y)] or as part of a multidomain protein [e.g. Swi2p in Saccharomyces cerevisiae or HRX (ALL-1) in Homo sapiens]. It is most interesting that this motif seems to be quite specific to known or predicted chromosomal/DNA-binding proteins, suggesting that it may act as a versatile minor groove tether.