Using amino acid patterns to accurately predict translation initiation sites

The translation initiation site (TIS) prediction problem is about how to correctly identify TIS in mRNA, cDNA, or other types of genomic sequences. High prediction accuracy can be helpful in a better understanding of protein coding from nucleotide sequences. This is an important step in genomic analysis to determine protein coding from nucleotide sequences. In this paper, we present an in silico method to predict translation initiation sites in vertebrate cDNA or mRNA sequences. This method consists of three sequential steps as follows. In the first step, candidate features are generated using k-gram amino acid patterns. In the second step, a small number of top-ranked features are selected by an entropy-based algorithm. In the third step, a classification model is built to recognize true TISs by applying support vector machines or ensembles of decision trees to the selected features. We have tested our method on several independent data sets, including two public ones and our own extracted sequences. The experimental results achieved are better than those reported previously using the same data sets. Our high accuracy not only demonstrates the feasibility of our method, but also indicates that there might be "amino acid" patterns around TIS in cDNA and mRNA sequences.     

Comparison of complexity and diversity of polyadenylated polysomal and informosomal messenger ribonucleic acid from Chinese hamster cells.

The sequence complexity and relative abundance of cytoplasmic polyadenylated polysomal (ribosome-bound) mRNA and cytoplasmic polyadenylated informosomal (ribosome-free) mRNA were analyzed in exponentially growing Chinese hamster cells (line CHO) using the technique of cDNA hybridization to excess poly(A)+ mRNA. Polysomal and informosomal mRNAs had similar complexities ( approximately 8300 mRNA species), but both the fraction of mRNA and the number of sequences comprising the mRNA abundance classes were different. Heterologous annealing reactions showed that all of the mRNA sequences detected were shared by the polysomal and informosomal mRNAs. However, the most abundant informosomal mRNA component was considerably different from the most abundant polysomal mRNA component. For a more detailed analysis, cDNA complementary to the most abundant informosomal and polysomal mRNAs was isolated. By use of the fractionated cDNA, it could be demonstrated that the most abundant informosomal mRNA sequences were distributed in the polysomal mRNA with an approximately fivefold reduction in relative frequency. These results are not compatible with models postulating translational control of gene expression by the complete sequestering of some mRNA sequences in an untranslatable form in the cytoplasm. The data are, however, consistent with models encompassing differential rates of initiation on the polysome and/or preferential affinity of some mRNAs for initiation factors.     

Abscisic acid induction of cloned cotton late embryogenesis-abundant (Lea) mRNAs

Summary Earlier studies found that cotton (Gossypium hirsutum L.) cotyledons contain several mRNAs which are more abundant during late embryogenesis than in mid-embryogenesis or early germination. They are here termed Late embryogenesis-abundant mRNAs, encoded by Lea loci. Complementary DNA clones for 18 such mRNA sequences, defined at a hybridization criterion of T_m-15°C, were identified in a mature embryo cDNA library by differential cDNA hybridization. At a lower hybridization criterion, some sequence homology was found within several of these cloned Lea mRNA sequences. Each Lea mRNA sequence comprises 0.04–1.3% of mature embryo poly(A)⁺ mRNA, a level ten-fold to several hundred-fold higher than in young embryo or 24 h seedling poly(A)⁺ mRNA. Of 18 Lea mRNA sequences examined in cultured young embryos, the level of at least 13 are specifically increased by exogenous abscisic acid (ABA), several to a level near that in normal mature embryos. However, the abundance of several of the sequences does not appear to be significantly modulated by ABA. The LEA polypeptides encoded by 10 Lea mRNA sequences were identified by hybrid-arrested translation. They include most of the late embryogenesis-abundant, ABA-inducible, polypeptides previously identified. Preliminary results suggest that many of the individual Lea mRNA sequences are transcribed from 1–3 genes in each of cotton's two subgenomes.     

Messenger ribonucleic acid metabolism in mammalian mitochondria: relationship between the decay of mitochondrial mRNA and their poly(A).

The mitochondrial poly(adenylic acid)-containing mRNAs in mouse ascites cells pulsed with radioactive precursors contain 35-55 nucleotide long poly(adenylic acid) sequences. These sequences are shortened with age in cells chased with cold medium. The possible relationship between the decay rates of mRNA and their poly(adenylic acid) sequences has been investigated by using mitochondrial RNA and protein synthesis inhibitors. The pattern of mRNA decay as determined by a new solid-phase-bound complementary DNA procedure indicates the presence of two classes of poly(adenylic acid) containing mRNA in mammalian mitochondria: one decaying with a t1/2 of 45 min and the other class with a t1/2 of 210 min. Inhibitors such as ethidium bromide and puromycin which accelerate the decay of mitochondrial mRNA also cause an enhanced decay of poly(adenylic acid) sequences. These results have been interpreted as evidence supporting the involvement of poly(adenylic acid) sequences in the mRNA stability.     

Cloned complementary deoxyribonucleic acid probes for untranslated messenger ribonucleic acid components of mouse sarcoma ascites cells

Mouse sarcoma ascites cells contain several abundant mRNA species that occur to a large extent in an untranslated state. RNA preparations enriched in these species were used as starting material to construct recombinant plasmids. Cloned plasmids bearing sequences homologous to four of the untranslated mRNA species were identified by translation of hybrid-selected material. These plasmids, as well as a recombinant plasmid derived from chick alpha-actin mRNA, were used as probes for the estimation of mRNA levels in polyribosomes and in small ribonucleoprotein (RNP) particles of the ascites cells. Considerable amounts of the mRNA molecules belonging to the untranslated species were present in polyribosomes as well as in mRNPs. The actin mRNA, on the other hand, was present almost exclusively in polyribosomes. The distributions obtained by the hybridization assay resembled those estimated by translation of the same RNA preparations in cell-free systems. This indicates that the mRNA molecules of a given species engaged in translation in the cells and those present as untranslated RNP particles are equally effective in cell-free translation systems.     

Effect of a short term fast on the distribution of cytoplasmic albumin messenger ribonucleic acid in rat liver. Evidence for formation of free albumin messenger ribonucleoprotein particles

Recently, using molecular hybridization techniques with albumin [3H]cDNA, we have determined that in normally fed rats 98% of total liver polyribosomal albumin mRNA sequences are found in membrane-bound polyribosomes (Yap, S. H., Strair, R. K., and Shafritz, D. A. (1977) Proc. Natl. Acad. Sci. U. S. A. 74, 5397-5401). We now observe that a 24- to 30-h withdrawal of food leads to major changes in the amount and subcellular distribution of albumin mRNA molecules. The total amount of cytoplasmic albumin mRNA per liver and concentration of albumin mRNA per unit of membrane-bound polyribosomal RNA are decreased. However, the proportion of albumin mRNA present in the postribosomal supernatant fraction increases dramatically in a short term fast, so that it now represent 60% of total cytoplasmic albumin mRNA sequences. Most of the albumin mRNA sequences in the postribosomal supernatant fraction sediment between 30 S and 50 S. These findings suggest that albumin mRNA is probably stored in the messenger ribonucleoprotein fraction during the fasting state.     

Two subfamilies of plant purple acid phosphatases

The cDNA sequences of two purple acid phosphatases (previously purified AcPase1 and putative AcPase2) from yellow lupin ( Lupinus luteus L) have been determined. AcPase1 mRNA is 1755 nt in length, including a single open reading frame of 1434 nt that encodes 477 amino acid residues (EMBL accession number AJ458943 ). The AcPase2 mRNA is a 170-nt molecule, including a single open reading frame of 1392 nt that encodes 463 amino acid residues (EMBL accession number AJ505579 ). Both sequences were compared with known purple acid phosphatases (PAPs). The analysis of all known PAP sequences showed that they are divided into two subfamilies. The characteristic feature of the first subfamily containing AcPase1 is the lack of cysteine residues in the C-terminal region of the polypeptide around position 350. AcPase2 belongs to the second subfamily which contains a disulphide bridge formed of cysteines present around position 350. A possible relationship between structure and substrate specificity in these two PAP subfamilies is presented. Expression of AcPase1 and AcPase2 shows that the first enzyme is produced in the early stage of germination (its mRNA in polyadenylated form is present already in dry seeds), whereas the second phosphatase is produced mainly in the roots of mature plants. The early expression of AcPase1 was confirmed by a comparative analysis of a cotranslational N-glycosylation of AcPase1 and diphosphonucleotide phosphatase/phosphodiesterase ( PPD1 ) known to be expressed in the final stage of germination.     

Control of ribonucleic acid function by oligonucleoside methylphosphonates

Oligodeoxyribonucleoside methylphosphonates contain nonionic 3'-5' linked methylphosphonate internucleotide bonds in place of the normal charged phosphodiester linkage of natural nucleic acids. These oligomers are resistant to nuclease hydrolysis, can pass through the membranes of mammalian cells in culture and can form stable hydrogen-bonded complexes with complementary nucleotide sequences of cellular RNAs such as mRNA. The oligomers are readily synthesized on insoluble polymer supports. Their chainlength and nucleotide sequence can be determined by chemical sequencing procedures. Oligonucleoside methylphosphonates which are complementary to the 5'-end, initiation codon region, or coding region of rabbit globin mRNA inhibit translation of the mRNA in rabbit reticulocyte lysates and globin synthesis in rabbit reticulocytes. This inhibition is due to the interaction of the oligomers with mRNA and the extent of inhibition is influenced by the secondary structure of the mRNA and the location of oligomer binding site on the mRNA. Oligomers complementary to the initiation codon regions of N, NS and G protein mRNAs of Vesicular stomatitis virus (VSV) inhibit virus protein synthesis in VSV-infected Mouse L-cells. These oligomers do not affect L-cell protein synthesis or growth. Virus protein synthesis and growth can also be selectively inhibited by oligonucleoside methylphosphonates which are complementary to the donor or acceptor splice junctions of virus pre mRNA. An oligomer complementary to the donor splice junction of SV40 large T antigen mRNA inhibits T-antigen synthesis in SV40-infected African green monkey kidney cells but does not inhibit overall cellular protein synthesis.(ABSTRACT TRUNCATED AT 250 WORDS)     

Detection of early changes in androgen-induced mouse renal beta-glucuronidase messenger ribonucleic acid using cloned complementary deoxyribonucleic acid

beta-Glucuronidase mRNA was purified from androgen-induced mouse kidney by immunoadsorption of polysomes to protein A-Sepharose. Cell-free translation of mRNA isolated from the protein A bound RNA followed by immunoprecipitation revealed that beta-glucuronidase mRNA represented approximately 2% of the purified mRNA fraction. This mRNA preparation was used to produce complementary DNA clones by recombination with pBR322. Clones containing sequences that were enriched during the purification procedure were selected by differential colony hybridization. These were further screened for homology with beta-glucuronidase mRNA by hybrid-selected translation. A beta-glucuronidase cDNA clone, designated pGUS7, was identified by these criteria. With this plasmid, the abundance of beta-glucuronidase mRNA in total poly(A) mRNA from androgen-induced mouse kidney was estimated to be less than 0.04%. The beta-glucuronidase cDNA plasmid hybridized to a mRNA of 2.6 kb in length, which was induced in an androgen receptor dependent fashion over a time course of 21 days. Treatment of female mice with a single dose of testosterone (10 mg) revealed that beta-glucuronidase mRNA concentration begins to increase between 12 and 24 h after hormone administration.     

The presence of polyriboadenylic acid sequences in calf lens messenger RNA

The presence of poly(rA) sequences in lens RNA has been demonstrated by the isolation of RNase A and T₁-resistant fragments of approximately 50 nucleotide residues. These poly(rA)-rich sequences, obtained from lenses incubated for six hours in organ culture with [³H]adenosine, are located at the 3′ termini of mRNA as determined by 3′ exoribonuclease digestion. Limited digestion of the [³H]adenosine-labeled mRNA with the enzyme led to the abolition of binding to poly(rU)-filters and a concomitant loss of template activity with avian myeloblastosis virus RNA-dependent DNA polymerase. Furthermore, after incubation of lenses in organ culture with 3′-deoxyadenosine, the isolated polysomal RNA was unable to function as a template in an avian myeloblastosis virus RNA-dependent DNA polymerase-catalyzed reaction system.     

The primary structure of rat ribosomal proteins: the amino acid sequences of L27a and L28 and corrections in the sequences of S4 and S12

The amino acid sequences of rat ribosomal proteins L27a and L28 were deduced from the sequences of nucleotides in recombinant cDNAs and confirmed from the NH2-terminal amino acid sequences of the proteins. L27a contains 147 amino acids (the NH2-terminal methionine is removed after translation of the mRNA) and has a molecular weight of 16 476. Hybridization of the cDNA to digests of nuclear DNA suggests that there are 18-22 copies of the L27a gene. The mRNA for the protein is about 600 nucleotides in length. L27a is homologous to mouse L27a (there are 3 amino acid changes) and to yeast L29. Rat ribosomal protein L28 has 136 amino acids (its NH2-terminal methionine is also processed after translation) and has a molecular weight of 15 707. Hybridization of the cDNA to digests of nuclear DNA suggests that there are 9 or 10 copies of the L28 gene. The mRNA for the protein is about 640 nucleotides in length. L28 contains a possible internal duplication of 9 residues. Corrections are recorded in the sequences reported before for rat ribosomal proteins S4 and S12.     

Estradiol stabilizes estrogen receptor messenger ribonucleic acid in sheep endometrium via discrete sequence elements in its 3'-untranslated region

The preovulatory surge of estrogen up-regulates estrogen receptor-alpha (ER) gene expression in the uterus during the estrous/menstrual cycles of female mammals. Previously, we demonstrated that the 5-fold increase in ER mRNA levels in endometrium of ovariectomized ewes treated with a physiological dose of estradiol (E2) is entirely due to an increase in ER mRNA stability. Our current work confirms that the E2 effect is specific to ER mRNA. The sequence of ER mRNA, cloned from sheep endometrium, shows a high degree of conservation with those of other species, even in the 5'- and the very long 3'-untranslated regions. In a cell-free assay, ER mRNA demonstrates greater stability with endometrial extracts from E2-treated ewes compared with those from untreated ovariectomized ewes. The E2-enhanced stability of ER mRNA was ablated by prior treatment of the extracts with proteinase K, 70 C heat, and oxidizing and alkylating reagents, indicating that a protein is responsible for stabilization of the message. The 3'-untranslated region of ER mRNA contains discrete sequences required for E2-enhanced stability, four of which were identified by extensive deletion mutant analyses. Transfer of two of the four minimal E2-modulated stability sequences conferred E2-enhanced stability to a heterologous RNA. These minimal E2-modulated stability sequences contain a common 10-base, uridine-rich sequence that is predicted to reside in a loop structure. Throughout our studies, estrogen stabilization of ER mRNA in sheep endometrium resembled that of vitellogenin mRNA in frog liver, indicating conservation of this ancient mechanism for enhancing gene expression in response to estrogen.     

Measurement of the complexity and diversity of poly(adenylic acid) containing messenger RNA from rat liver.

The complexity of rat liver poly (A)+ messenger RNA (mRNA) has been measured by analysis of the kinetics of hydridization with both complementary DNA (cDNA) and single copy DNA. The complementary DNA-poly(A)+ mRNA hybridization reaction demonstrates the existence of three abundance classes representing 18, 37, and 45% of the cDNA and 4, 290, and 24 000 different 1800-nucleotide sequences respectively. The poly(A)+ mRNA driven single copy DNA hybridization reaction reveals a single major transition accounting for 1.9% of the haploid rat genome. The kinetics of the poly(A)+ mRNA driven single copy DNA reaction suggest that approximately 45% of the mass of the mRNA population contains over 95% of the complexity. Although higher than previous estimates, the base sequence complexities of rat liver poly(A)+ mRNA measured in these two ways are in good agreement, suggesting that the technique of poly(A)+ mRNA-cDNA hybridization may be used in approximating the complexity as well as abundance of a messenger RNA population. DNA-driven cDNA reactions reveal that about 10% of rat liver poly(A)+ mRNA is transcribed from repetitive sequences in the rat genome.     

Molecular cloning of gene sequences for avian fatty acid synthase and evidence for nutritional regulation of fatty acid synthase mRNA concentration

A double-stranded cDNA library was constructed using total poly(A)+ RNA from the goose uropygial gland. Clones containing sequences complementary to fatty acid synthase mRNA were initially identified by colony hybridization with a 32P-labeled cDNA transcribed from RNA enriched for fatty acid synthase mRNA. Identity of the fatty acid synthase clones was confirmed by hybrid-selected translation. Mature fatty acid synthase mRNA is approximately 16 kilobases in length. When unfed neonatal goslings were fed for 24 hr, relative synthesis of hepatic fatty acid synthase increased more than 42-fold. Concomitantly, hepatic fatty acid synthase mRNA levels increased 70-fold. Thus, nutritional regulation of the synthesis of hepatic fatty acid synthase probably occurs at the pretranslational level. The availability of a specific probe for fatty acid synthase mRNA should allow us to analyze the regulation of expression of this gene during development, by nutrition and by hormones in both liver and uropygial gland.     

Reduction of prostatic binding protein-messenger ribonucleic acid sequences in rat prostate by castration

Messenger RNA coding for the three subunits of prostatic binding protein was isolated from polysomal RNA of rat ventral prostate by oligo (dT)-cellulose affinity chromatography and purified by repeated sedimentations through sucrose gradients under denaturing conditions. The purified mRNA migrated as a 9S peak in sucrose gradient centrifugation and hybridized with its cDNA within 2 log Rot units. In a cell-free reticulocyte lysate system, the mRNA directed the synthesis of three polypeptides of 12000, 9000, and 8000 daltons. These translation products were identified as the subunits of prostatic binding protein by immunoreaction with antibodies to this protein. Quantitation of prostatic binding protein-mRNA sequences in normal and castrated rats by hybridization with the cDNA probe showed that 3-day castration reduced the prostatic binding protein-mRNA sequences to less than 2% of the normal level. Similar hybridization was performed by using the cDNA to determine the level of prostatic binding protein coding sequences in polysomal poly(A) RNA following castration. The results showed a first-order rate constant of 3.92 X 10-2 h-1 for reduction of prostatic binding protein-mRNA sequences in polysomes. The period of castration required to reduce the level of these sequences to 50% of the normal level was calculated to be 17.6 h.     

The occurrence in amino acid sequences of extensive informational symmetries based on possible codon-codon complementarity in the encoding polynucleotides.

1. A procedure is described for the detection and assessment of informational complementarity in an amino acid sequence; it is based on possible autocomplementarity in the mRNA, and involves codon-to-codon matching. 2. This procedure was applied to myelin basic protein, a variety of protamines, histone IV, silk fibroin, rat skin collagen alpha1 chain and a sheep keratin. A multiplicity of extensive low-probability informational symmetries, based on codon-to-codon matching, were detected. 3. These low-probability orderings, which are independent of the actual mRNA codons, are rationalized in terms of the evolutionary ordering of the amino acid sequences concerned, in such a way that constraints on the secondary structure of the coding polynucleotides were satisfied. This possible interpretation is supported by a number of significant common properties of the protein sequences analysed.