Relationship between amino acid composition and gene expression in the mouse genome.

BACKGROUND: Codon bias is a phenomenon that refers to the differences in the frequencies of synonymous codons among different genes. In many organisms, natural selection is considered to be a cause of codon bias because codon usage in highly expressed genes is biased toward optimal codons. Methods have previously been developed to predict the expression level of genes from their nucleotide sequences, which is based on the observation that synonymous codon usage shows an overall bias toward a few codons called major codons. However, the relationship between codon bias and gene expression level, as proposed by the translation-selection model, is less evident in mammals. FINDINGS: We investigated the correlations between the expression levels of 1,182 mouse genes and amino acid composition, as well as between gene expression and codon preference. We found that a weak but significant correlation exists between gene expression levels and amino acid composition in mouse. In total, less than 10% of variation of expression levels is explained by amino acid components. We found the effect of codon preference on gene expression was weaker than the effect of amino acid composition, because no significant correlations were observed with respect to codon preference. CONCLUSION: These results suggest that it is difficult to predict expression level from amino acid components or from codon bias in mouse.

GeneWaltz--A new method for reducing the false positives of gene finding.

BACKGROUND: Identifying protein-coding regions in genomic sequences is an essential step in genome analysis. It is well known that the proportion of false positives among genes predicted by current methods is high, especially when the exons are short. These false positives are problematic because they waste time and resources of experimental studies. METHODS: We developed GeneWaltz, a new filtering method that reduces the risk of false positives in gene finding. GeneWaltz utilizes a codon-to-codon substitution matrix that was constructed by comparing protein-coding regions from orthologous gene pairs between mouse and human genomes. Using this matrix, a scoring scheme was developed; it assigned higher scores to coding regions and lower scores to non-coding regions. The regions with high scores were considered candidate coding regions. One-dimensional Karlin-Altschul statistics was used to test the significance of the coding regions identified by GeneWaltz. RESULTS: The proportion of false positives among genes predicted by GENSCAN and Twinscan were high, especially when the exons were short. GeneWaltz significantly reduced the ratio of false positives to all positives predicted by GENSCAN and Twinscan, especially when the exons were short. CONCLUSIONS: GeneWaltz will be helpful in experimental genomic studies. GeneWaltz binaries and the matrix are available online at http://en.sourceforge.jp/projects/genewaltz/.

The RNA-binding protein Mex3b has a fine-tuning system for mRNA regulation in early Xenopus development.

Post-transcriptional control by RNA-binding proteins is a precise way to assure appropriate levels of gene expression. Here, we identify a novel mRNA regulatory system involving Mex3b (RKHD3) and demonstrate its role in FGF signaling. mex3b mRNA has a 3' long conserved UTR, named 3'LCU, which contains multiple elements for both mRNA destabilization and translational enhancement. Notably, Mex3b promotes destabilization of its own mRNA by binding to the 3'LCU, thereby forming a negative autoregulatory loop. The combination of positive regulation and negative autoregulation constitutes a fine-tuning system for post-transcriptional control. In early embryogenesis, Mex3b is involved in anteroposterior patterning of the neural plate. Consistent with this, Mex3b can attenuate FGF signaling and destabilize mRNAs for the FGF signaling components Syndecan 2 and Ets1b through their 3' UTRs. These data suggest that the 3'LCU-mediated fine-tuning system determines the appropriate level of mex3b expression, which in turn contributes to neural patterning through regulating FGF signaling.

Evaluation of the effect of CpG hypermutability on human codon substitution.

Understanding the cause underlying the changes in amino acid composition of proteins is essential for understanding protein evolution and function. Accurate models of DNA and protein evolution are essential for studying molecular evolution. Although many models have been developed, most models assume that each site evolves independently and that substitutions are time reversible. In mammals and other organisms, CpG hypermutability is one of the major causes of nucleotide mutations because CpG dinucleotides are often methylated at C, and the methyl-C mutation spontaneously deaminates to yield T about 3 times more rapidly than other types of point mutations. In this study, we evaluate the effect of CpG hypermutability on codon substitution by comparing thousands of coding regions in the human and chimpanzee genomes and by inferring ancestral sequences by using mouse as the outgroup. We found that 14% of synonymous and nonsynonymous substitutions on human genes were caused by CpG hypermutability. Based on these results, we developed a model that incorporates CpG hypermutability as well as the transition/transversion ratio and changes in the chemical properties of amino acids.

The universal trend of amino acid gain-loss is caused by CpG hypermutability.

Understanding the cause of the changes in the amino acid composition of proteins is essential for understanding the evolution of protein functions. Since the early 1970s, it has been known that the frequency of some amino acids in protein sequences is increasing and that of others is decreasing. Recently, it was found that the trends of amino acid changes were similar in 15 taxa representing Bacteria, Archaea, and Eukaryota. However, the cause of this similarity in the trend of the gains and losses of amino acids continued to be debated. Here, we show that this trend of the gain and loss of amino acids can be simply explained by CpG hypermutability. We found that the frequency of amino acids coded by codons with TpG dinucleotides and those with CpA dinucleotides is increasing, while that of amino acids coded by codons with CpG dinucleotides is decreasing. We also found that organisms that lack DNA methyltransferase show different trends of the gain and loss of amino acids. DNA methyltransferase methylates CpG dinucleotides and induces CpG hypermutability. The incorporation of CpG hypermutability into models of protein evolution will improve studies on protein evolution in different organisms.

Exploration of human ORFeome: high-throughput preparation of ORF clones and efficient characterization of their protein products.

In this study, we established new systematic protocols for the preparation of cDNA clones, conventionally termed open reading frame (ORF) clones, suitable for characterization of their gene products by adopting a restriction-enzyme-assisted cloning method using the Flexi cloning system. The system has following advantages: (1) preparation of ORF clones and their transfer into other vectors can be achieved efficiently and at lower cost; (2) the system provides a seamless connection to the versatile HaloTag labeling system, in which a single fusion tag can be used for various proteomic analyses; and (3) the resultant ORF clones show higher expression levels both in vitro and in vivo. With this system, we prepared ORF clones encoding 1,929 human genes and characterized the HaloTag-fusion proteins of its subset that are expressed in vitro or in mammalian cells. Results thus obtained have demonstrated that our Flexi ORF clones are efficient for the production of HaloTag-fusion proteins that can provide a new versatile set for a variety of functional analyses of human genes.

Influence of the 3'-UTR-length of mKIAA cDNAs and their sequence features to the mRNA expression level in the brain.

We have previously described the sequence features of approximately 1500 mouse KIAA (mKIAA) genes in comparison with those of human KIAA genes (Okazaki, N., Kikuno, R., Inamoto, S., Hara, Y., Nagase, T., Ohara, O., and Koga, H. 2002, DNA Res., 9, 179-188; Okazaki, N., Kikuno, R., Ohara, R., Inamoto, S., Aizawa, H., Yuasa, S., Nakajima, D., Nagase, T., Ohara, O., and Koga, H. 2003, DNA Res., 10, 35-48; Okazaki, N., Kikuno, R., Ohara, R., Inamoto, S., Koseki, H., Hiraoka, S., Saga, Y., Nagase, T., Ohara, O., and Koga, H. 2003, DNA Res., 10, 167-180; and Okazaki, N., F-Kikuno, R., Ohara, R., Inamoto, S., Koseki, H., Hiraoka, S., Saga, Y., Seino, S., Nishimura, M., Kaisho, T., Hoshino, K., Kitamura, H., Nagase, T., Ohara, O., and Koga, H. 2004, DNA Res., 11, 205-218). To validate the orthologous relationship between mKIAA and KIAA genes in detail, we examined their chromosomal positions and evolutionary rate of synonymous substitutions and confirmed that >93% of the mKIAA/KIAA gene pairs are orthologous. During the sequence analysis of mKIAA genes, we found that 3'-untranslated region (3'-UTR) lengths of mKIAA and KIAA genes are extremely long. In the meanwhile, we have also examined the tissue-specific expression of approximately 1700 mKIAA genes using cDNA microarray and verified predominantly their expression in adult brain (Koga, H., Yuasa, S., Nagase, T., Shimada, K., Nagano, M., Imai, K., Ohara, R., Nakajima, D., Murakami, M., Kawai, M., Miki, F., Magae, J., Inamoto, S., Okazaki, N., Ohara, O. 2004, DNA Res., 11, 293-304). To connect these two evidences, we statistically analysed the relationship between them by using the mKIAA genes. Consequently, a positive correlation was observed between the 3'-UTR lengths and the relative expression intensities in adult brain. Furthermore, we searched sequence elements in the 3'-UTR possibly related with their expression and found some candidates regarding the brain-specific expression.

CDNA library construction from a small amount of RNA: adaptor-ligation approach for two-round cRNA amplification using T7 and SP6 RNA polymerases.

In this study, we developed a method that allows cDNA library construction from a small amount of RNA without causing serious size bias in the resulting cDNA population. For this purpose, we adopted two-round cRNA amplification by T7 and SP6 RNA polymerases. The first-round cDNAs, flanked by the promoter sequences of T7 and SP6 RNA polymerases, were synthesized from 1 microg total RNA and then subjected to two rounds of cRNA amplification. Comparison of the sizes of the first-round and the second-round cRNAs indicated that the size-bias effect of the second-round cRNA synthesis was not serious. The resultant double-stranded cDNAs were cloned into a plasmid by in vitro lambda phage recombination with an efficiency of 1.2 x 10(11) colony-forming unit/microgram of starting total RNA. Characterization of the resultant cDNA library in terms of the insert size, clone redundancy, and integrity of 3' ends of cDNAs indicated that the amplified library was comparable to a library constructed by a conventional method, although large cDNAs tend to be slightly truncated in the amplified library. This method enables the construction of a library from a small amount of RNA, and calculations suggest that the strategy would be efficient enough to use even a single cell as starting material.

Preparation of a set of expression-ready clones of mammalian long cDNAs encoding large proteins by the ORF trap cloning method.

Although we have so far identified and sequenced >2000 human long cDNAs, known as KIAA cDNAs, half of them have yet to be functionally annotated. Expression-ready cDNA clones derived from these genes, where the open reading frame (ORF) of the gene of interest is placed under the control of an appropriate promoter, are critical for functional characterization of these gene products. In this study, we attempted to systematically convert original cDNA clones to expression-ready forms for native and fusion proteins. For this purpose, we developed a new method for ORF cloning based on a homologous recombination in Escherichia coli to avoid laborious manipulations and artificial introduction of mutations in ORF. Using 1589 putative full-length ORFs (from 1002 KIAA genes, 119 human known genes and 468 mouse genes) with an average size of 2.8 kb, we successfully prepared expression plasmids for 1463 native proteins and for 1343 fusion proteins by this method. The resultant expression-ready clones were examined using an in vitro transcription/translation system followed by SDS-polyacrylamide gel electrophoresis and by transient expression of GFP-fusion proteins in human embryonic kidney (HEK) 293 cells. This set of expression-ready clones of long cDNAs encoding large proteins would open a new route to experimentally analyze their functions on a proteomic scale, since unavailability of expression-ready clones for mammalian large proteins has been a major obstacle to the functional analysis of these cDNAs.

Alternative splice variants encoding unstable protein domains exist in the human brain.

Alternative splicing has been recognized as a major mechanism by which protein diversity is increased without significantly increasing genome size in animals and has crucial medical implications, as many alternative splice variants are known to cause diseases. Despite the importance of knowing what structural changes alternative splicing introduces to the encoded proteins for the consideration of its significance, the problem has not been adequately explored. Therefore, we systematically examined the structures of the proteins encoded by the alternative splice variants in the HUGE protein database derived from long (>4 kb) human brain cDNAs. Limiting our analyses to reliable alternative splice junctions, we found alternative splice junctions to have a slight tendency to avoid the interior of SCOP domains and a strong statistically significant tendency to coincide with SCOP domain boundaries. These findings reflect the occurrence of some alternative splicing events that utilize protein structural units as a cassette. However, 50 cases were identified in which SCOP domains are disrupted in the middle by alternative splicing. In six of the cases, insertions are introduced at the molecular surface, presumably affecting protein functions, while in 11 of the cases alternatively spliced variants were found to encode pairs of stable and unstable proteins. The mRNAs encoding such unstable proteins are much less abundant than those encoding stable proteins and tend not to have corresponding mRNAs in non-primate species. We propose that most unstable proteins encoded by alternative splice variants lack normal functions and are an evolutionary dead-end.

Prediction of the coding sequences of mouse homologues of KIAA gene: IV. The complete nucleotide sequences of 500 mouse KIAA-homologous cDNAs identified by screening of terminal sequences of cDNA clones randomly sampled from size-fractionated libraries.

We have been conducting a mouse cDNA project to predict protein-coding sequences of mouse homologues of human KIAA and FLJ genes since 2001. As an extension of these projects, we herein present the entire sequences of 500 mKIAA cDNA clones and 4 novel cDNA clones that were incidentally identified during this project. We have isolated cDNA clones from the size-fractionated mouse cDNA libraries derived from 7 tissues and 3 types of cultured cells. The average size of the 504 cDNA sequences reached 4.3 kb and that of the deduced amino acid sequences from these cDNAs was 807 amino acid residues. We assigned the integrity of CDSs from the comparison with the corresponding human KIAA cDNA sequences. The comparison of mouse and human sequences revealed that two different human KIAA cDNAs are derived from single genes. Furthermore, 3 out of 4 proteins encoded in the novel cDNA clones showed moderate sequence similarity with human KIAA proteins, thus we could obtain new members of KIAA protein families through our mouse cDNA projects.

Prediction of the coding sequences of mouse homologues of FLJ genes: the complete nucleotide sequences of 110 mouse FLJ-homologous cDnas identified by screening of terminal sequences of cDNA clones randomly sampled from size-fractionated libraries.

We have been conducting a mouse cDNA project to predict protein-coding sequences of mouse KIAA-homologous genes since 2001. As an extension of this project, we also started to accumulate mouse cDNA clones homologous to the human FLJ cDNA clones which are another long cDNA resource produced in our institute. We have isolated the cDNA clones from size-fractionated cDNA libraries derived from five different mouse tissues and natural killer T-cells. Although the human FLJ cDNA clones were originally derived from human spleen libraries, one-third of their mouse homologues were obtained from the brain library. We designated these homologues "mFLJ" plus a 5-digit number and herein characterized 110 mFLJ cDNA clones. We assigned an integrity of the CDSs from the comparison of the 110 cDNA clones with the corresponding human FLJ cDNA clones. The average size of the 110 mouse cDNA sequences was 3.8 kb and that of the deduced amino acid sequences from their longest CDS in each cDNA was 663 amino acid residues. Homology and/or motif search against public databases revealed new domains and/or motifs in 26 mFLJ gene products which provide additional speculation regarding the function of FLJ genes.

The CAP-Gly domain of CYLD associates with the proline-rich sequence in NEMO/IKKgamma.

CYLD was originally identified as the human familial cylindromatosis tumor suppressor. Recently, it was reported that CYLD directly interacts with NEMO/IKKgamma and TRAF2 in the NF-kappaB signaling pathway. The two proteins bind to a region of CYLD that contains a Cys-box motif and the third cytoskeleton-associated protein-glycine conserved (CAP-Gly) domain. Here we report that the third CAP-Gly domain of CYLD specifically interacts with one of the two proline-rich sequences of NEMO/IKKgamma. The tertiary structure of the CAP-Gly domain shares the five-stranded beta sheet topology with the SH3 domain, which is well known as a proline-rich sequence-recognition domain. However, chemical shift mapping revealed that the peptide binding site of the CAP-Gly domain is formed without the long peptide binding loop characteristic of the SH3 domain. Therefore, CAP-Gly is likely to be a novel proline-rich sequence binding domain with a mechanism different from that of the SH3 domain.

Method for systematic targeted isolation of homologous cDNA fragments in a multiplex format.

In this study, a two-step method for systematic multiplex cloning of homologous cDNAs from related species was developed. The first step, called MUCH (multiplex cloning of homologous genes), is cloning of partial but authentic cDNA fragments of homologous cDNAs by hybridization to arrayed cRNA probes of specified genes on a nylon membrane, followed by PCR amplification of the hybridized fragments. The second step is PCR-based screening of a library that contains longer cDNA inserts based on the sequences obtained in the first step. To evaluate this method, we tried to isolate mouse counterparts of 53 human large cDNAs by MUCH and could successfully isolate 32 mouse counterpart cDNAs from a single library. Complete sequencing of two mouse cDNAs isolated by PCR-based screening further demonstrated that this method enabled us to isolate multiple homologous cDNAs in parallel. We thus expect that this method could be applied to high-throughput cloning of homologous cDNAs in related species.

HUGE: a database for human KIAA proteins, a 2004 update integrating HUGEppi and ROUGE.

We have been developing a Human Unidentified Gene-Encoded (HUGE) protein database (http://www.kazusa.or.jp/huge) to summarize results from sequence analysis of human novel large (>4 kb) cDNAs identified in the Kazusa cDNA sequencing project. At present, HUGE contains 2031 cDNA entries (KIAA cDNAs), for each of which a gene/protein characteristic table has been prepared. Since we have been shifting our research attention from the identification and cloning of novel cDNAs to the functional analysis of the proteins encoded by these cDNAs (KIAA proteins), we have not substantially increased the number of cDNA entries in HUGE for some time. Instead, we have manually curated 451 KIAA cDNAs in order to prepare a set of genetic resources to facilitate the functional analysis of KIAA proteins. In addition, we have updated the contents of the corresponding gene/protein characteristic tables in HUGE and have constructed two subsidiary databases, HUGEppi (http://www. kazusa.or.jp/huge/ppi) and ROUGE (http://www. kazusa.or.jp/rouge), to make available the results from our study of KIAA protein function. HUGEppi shows detailed information on protein-protein interactions detected between 84 pairs of KIAA proteins by yeast two-hybrid screening. ROUGE summarizes the results of computer-assisted analyses of approximately 1000 mouse homologues of human large cDNAs that we identified.

Prediction of the coding sequences of mouse homologues of KIAA gene: III. the complete nucleotide sequences of 500 mouse KIAA-homologous cDNAs identified by screening of terminal sequences of cDNA clones randomly sampled from size-fractionated libraries.

We have conducted a human cDNA project to predict protein-coding sequences (CDSs) in large cDNAs (> 4 kb) since 1994, and the number of newly identified genes, known as KIAA genes, already exceeds 2000. The ultimate goal of this project is to clarify the physiological functions of the proteins encoded by KIAA genes. To this end, the project has recently been expanded to include isolation and characterization of mouse KIAA-counterpart genes. We herein present the entire sequences and the chromosome loci of 500 mKIAA cDNA clones and 13 novel cDNA clones that were incidentally identified during this project. The average size of the 513 cDNA sequences reached 4.3 kb and that of the deduced amino acid sequences from these cDNAs was 816 amino acid residues. By comparison of the predicted CDSs between mouse and human KIAAs, 12 mKIAA cDNA clones were assumed to be differently spliced isoforms of the human cDNA clones. The comparison of mouse and human sequences also revealed that four pairs of human KIAA cDNAs are derived from single genes. Notably, a homology search against the public database indicated that 4 out of 13 novel cDNA clones were homologous to the disease-related genes.

A new heat shock gene, AgsA, which encodes a small chaperone involved in suppressing protein aggregation in Salmonella enterica serovar typhimurium.

We discovered a novel small heat shock protein (sHsp) named AgsA (aggregation-suppressing protein) in the thermally aggregated fraction from a Salmonella enterica serovar Typhimurium dnaK-null strain. The -10 and -35 regions upstream of the transcriptional start site of the agsA gene are characteristic of sigma(32)- and sigma(72)-dependent promoters. AgsA was strongly induced by high temperatures. The similarity between AgsA and the other two sHsps of Salmonella serovar Typhimurium, IbpA and IbpB, is rather low (around 30% amino acid sequence identity). Phylogenetic analysis suggested that AgsA arose from an ancient gene duplication or amplification at an early evolutionary stage of gram-negative bacteria. Here we show that overproduction of AgsA partially complements the DeltadnaK52 thermosensitive phenotype and reduces the amount of heat-aggregated proteins in both DeltadnaK52 and DeltarpoH mutants of Escherichia coli. These data suggest that AgsA is an effective chaperone capable of preventing aggregation of nonnative proteins and maintaining them in a state competent for refolding in Salmonella serovar Typhimurium at high temperatures.

Genome-wide expression analysis of mouse liver reveals CLOCK-regulated circadian output genes.

CLOCK is a positive component of a transcription/translation-based negative feedback loop of the central circadian oscillator in the suprachiasmatic nucleus in mammals. To examine CLOCK-regulated circadian transcription in peripheral tissues, we performed microarray analyses using liver RNA isolated from Clock mutant mice. We also compared expression profiles with those of Cryptochromes (Cry1 and Cry2) double knockout mice. We identified more than 100 genes that fluctuated from day to night and of which expression levels were decreased in Clock mutant mice. In Cry-deficient mice, the expression levels of most CLOCK-regulated genes were elevated to the upper range of normal oscillation. Most of the screened genes had a CLOCK/BMAL1 binding site (E box) in the 5'-flanking region. We found that CLOCK was absolutely concerned with the circadian transcription of one type of liver genes (such as DBP, TEF, and Usp2) and partially with another (such as mPer1, mPer2, mDec1, Nocturnin, P450 oxidoreductase, and FKBP51) because the latter were damped but remained rhythmic in the mutant mice. Our results showed that CLOCK and CRY proteins are involved in the transcriptional regulation of many circadian output genes in the mouse liver. In addition to being a core component of the negative feedback loop that drives the circadian oscillator, CLOCK also appears to be involved in various physiological functions such as cell cycle, lipid metabolism, immune functions, and proteolysis in peripheral tissues.

Prediction of the coding sequences of mouse homologues of KIAA gene: II. The complete nucleotide sequences of 400 mouse KIAA-homologous cDNAs identified by screening of terminal sequences of cDNA clones randomly sampled from size-fractionated libraries.

We have accumulated information of the coding sequences of uncharacterized human genes, which are known as KIAA genes, and the number of these genes exceeds 2000 at present. As an extension of this sequencing project, we recently have begun to accumulate mouse KIAA-homologous cDNAs, because it would be useful to prepare a set of human and mouse homologous cDNA pairs for further functional analysis of the KIAA genes. We herein present the entire sequences of 400 mouse KIAA cDNA clones and 4 novel cDNA clones which were incidentally identified during this project. Most of clones entirely sequenced in this study were selected by computer-assisted analysis of terminal sequences of the cDNAs. The average size of the 404 cDNA sequences reached 5.3 kb and that of the deduced amino acid sequences from these cDNAs was 868 amino acid residues. The results of sequence analyses of these clones showed that single mouse KIAA cDNAs bridged two different human KIAA cDNAs in some cases, which indicated that these two human KIAA cDNAs were derived from single genes although they had been supposed to originate from different genes. Furthermore, we successfully mapped all the mouse KIAA cDNAs along the genome using a recently published mouse genome draft sequence.

Characterization of long cDNA clones from human adult spleen. II. The complete sequences of 81 cDNA clones.

To accumulate information on the coding sequences (CDSs) of unidentified genes, we have conducted a sequencing project of human long cDNA clones. Both the end sequences of approximately 10,000 cDNA clones from two size-fractionated human spleen cDNA libraries (average sizes of 4.5 kb and 5.6 kb) were determined by single-pass sequencing to select cDNAs with unidentified sequences. We herein present the entire sequences of 81 cDNA clones, most of which were selected by two approaches based on their protein-coding potentialities in silico: Fifty-eight cDNA clones were selected as those having protein-coding potentialities at the 5'-end of single-pass sequences by applying the GeneMark analysis; and 20 cDNA clones were selected as those expected to encode proteins larger than 100 amino acid residues by analysis of the human genome sequences flanked by both the end sequences of cDNAs using the GENSCAN gene prediction program. In addition to these newly identified cDNAs, three cDNA clones were isolated by colony hybridization experiments using probes corresponding to known gene sequences since these cDNAs are likely to contain considerable amounts of new information regarding the genes already annotated. The sequence data indicated that the average sizes of the inserts and corresponding CDSs of cDNA clones analyzed here were 5.0 kb and 2.0 kb (670 amino acid residues), respectively. From the results of homology and motif searches against the public databases, functional categories of the 29 predicted gene products could be assigned; 86% of these predicted gene products (25 gene products) were classified into proteins relating to cell signaling/communication, nucleic acid management, and cell structure/motility.