NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Madame Curie Bioscience Database [Internet]. Austin (TX): Landes Bioscience; 2000-.

Cover of Madame Curie Bioscience Database

Madame Curie Bioscience Database [Internet].

Show details

Bioinformatics: Mystery, Astrology or Service Technology?

*.

* Corresponding Author: Research Institute of Molecular Pathology, Dr. Bohr-Gasse 7, A-1030 Vienna, Republic Austria. Email: Frank.Eisenhaber@imp.univie.ac.at

Summary

Mathematical interpretation and integration of experimental data for the goal of biological theory development has had little, if none, impact on previous progress in life sciences compared with the sophistication of experimental approaches themselves. The genesis of recent spectacular breakthroughs in molecular biology that led to the discovery of the enzymatic function of several nonmetabolic enzymes illustrates that this relationship is beginning to change.

The development of high-throughput technologies, for example of complete genome sequencing, leads to large amounts of quantified data on biological system without a direct link to biological function that require formalized and complex mathematical approaches for their interpretation. The research success in life sciences depends increasingly on the ability of researchers in experimental and theoretical biology to jointly focus on important questions. Currently, theoretical methods have the best chances to contribute to new biological insight independently of experiments in the area of genome text interpretation and especially for gene function prediction. Experimental studies can help progress in the development of theoretical methods by providing verified, sufficiently large, and variable sequence datasets for the exploration of sequence-function relationships.

To caricature, the typical research process in life sciences consists of periodic repetitions of weeks/months of bench work by a PhD or postdoctoral student, followed by an hour of looking at the results by the lab head, after which the coworker again disappears into the cold room or behind the microscope with new directives. Generations of life scientists have been educated that the most important goal consists in producing “hard”, quantitative experimental data describing biological structures and processes. Pure theoretical efforts directed at biological data analysis are believed to add little more than an intellectual speculation or a colorful illustration in the form of a graph or an alignment. The biological theory itself has remained logically simple and with little or no mathematics or formal structure. Typically, all creativity has been directed into sophistication and rationalization of experimental procedures and techniques for the wet lab. This type of life science has successfully produced breakthroughs and will, apparently, continue to stay the major source of new biological insight in the near future.

This situation is especially astonishing for people that come from more formalized sciences, such as physics, where a typical experiment is preceded by months of calculations and computer simulations. Such research is necessary in this area to derive the most interesting research targets and to check the consistency of new hypotheses with existing knowledge. There had been several waves of efforts to inject mathematics into life sciences, for example statistics (beginning with Mendel's ratios), kinetics (of enzymes and ligand binding, of transport systems, in population dynamics), or 3D biomolecular structure modeling (together with quantum chemistry, QSAR studies, and molecular dynamics; especially in context with the hypothesis of DNA double-helical structure). Although each of these waves has enriched life sciences in some aspects, neither one has become a continuous source of qualitatively new biological knowledge or has made biology a truly theoretical, quantitative, and predictive science.

Beginning with the 1960s, yet another stream of efforts focused on the esoteric topic of analysis of text strings representing the monomer sequences of proteins and nucleic acids and of the evolution of these strings after multiple single-point mutations.1,2 Thanks to these pioneering efforts, theoretical concepts, computational methods, and sequence databases have been established that allow the prediction of function for experimentally uncharacterized genes from their sequence, most importantly, together with the quantification of the prediction error (prediction reliability) in probabilistic terms.3 The impact of this development is perceived in different ways by various parts of the generally wet lab-focused life science community, depending on personal background and experience: as mystery, astrology, or service technology. None of these three ways is a really appropriate assessment for the recent step in the difficult development of life sciences toward a formalized theory of living systems, as the discussion below will attempt to show.

Mystery

Sometimes, success stories are sensed euphorically as a mystery by those scientists that receive a tremendous boost in their experimental work from a function prediction. At the background of the general weakness of theory in life sciences, it is indeed perceived as a bolt from the blue by the experimental life science research community that a number of recent scientific breakthroughs in biology has originated from theoretical studies for gene function prediction. Several instances of discoveries of enzyme activities for a number nonmetabolic proteins, typically without any previous hint or suspicion from experimental findings, are remarkable evidence for the growing predictive power of theoretical biology.

Important science-organizational and cognitive aspects of this process toward new biological knowledge can be illuminated by historically viewing some recent examples of enzymatic function assignment to nonmetabolic enzymes. Three stories with considerable biological impact, namely

  1. the discovery of the molecular function of Fringe in Notch signaling,
  2. the determination of the protease domain of separin triggering the transition from metaphase to anaphase during the cell cycle, and
  3. the understanding of heterochromatin formation as initiated by the histone methyltransferase activity of the Su(var)3-9 homologues, are described in brief in Boxes 1, 2, and 3.

Box Icon

Box 1

As early as January 1997, Peer Bork and three of his collegues reported that a sequence domain in the D. melanogaster Fringe and Brainiac proteins, as well as in their vertebrate homologues, might have glycosyltransferase activity. Relying on simplified (more...)

Box Icon

Box 2

Cellular and nuclear divisions (mitosis) have been known since the early days of studying biological tissue with microscopes. The arrangement of sister chromosomes in the metaphase plate is shown in every cytology textbook. Nevertheless, we have learned (more...)

Box Icon

Box 3

After more than 5 years of studying the function of mice and human suv39h1 (and of the later identified testis-specific suv39h2), the vertebrate homologues of the D. melanogaster Su(var)3-9 proteins, with genetic and cellular biological methods, it was (more...)

There are more of such nontrivial findings, and it is not possible to give a complete list here. For example, a C-terminal domain in yeast protein dot1p was assigned to the SAM sequence family with suggested methyltransferase activity. The loss-of-function phenotype of the dot1 gene (disruption of telomere silencing) implied a possible role in the posttranslational modification of histones.4 Indeed, a biochemical assay was able to show that dot1p does methylate histone H3 at Lys79.5 As the authors acknowledge, the previously published theoretical report was critical for their decision to launch the experimental test.

In another case, the yeast protein eco1p was found critical for the establishment of cohesion between sister chromatids,6 but the biological experiments did not give any hint with respect to a possible molecular function of eco1p. Sequence analysis studies pointed to an acetyl-coenzyme A binding site in the C-terminal domain of eco1p. It was this finding that changed the priorities in the wet lab, and the subsequent experiments showed that eco1p has indeed acetyltransferase activity and, apparently, is part of a yet unknown mitotic pathway that involves acetylation of some cohesion complex proteins.7

What can be learned from these stories?

  1. A hypothesis derived from an extensive theoretical analysis suggested a previously unknown direction of thought and creatively enriched biological theory, not only in details, but for the given field of research, in a principal aspect. This happens everyday in formalized sciences but it is new in biology.
  2. There is an increasing need for an, although time- and resource-consuming, effort of theoretical analysis of biological experimental data accompanying life science research projects from the start onward. If completed successfully, it can lead to nontrivially new, creative directions of experimentation or directly to new biological insight. The new role of theory represents a qualitative change resulting from quantitative progress in the experimental method development accumulated over decades. Large amounts of quantitative data without a direct link to known biological processes at different levels of organizational hierarchy are now produced with high-throughput techniques and require a nontrivial effort for interpretation.
  3. Not incidentally, it is the area of gene function prediction where, for the first time, theoretical data analysis has become a widely indispensable activity for planning experimental strategies affecting, increasingly, the research efficiency in evermore branches of life sciences. Sequencing of biomolecules has been the first high-throughput technology in life sciences. During the last decade of the 20th century, electronic sequence databases, coupled with scientific literature sources, have reached such a dimension and representation for different biological subfields that the theoretical meta-analysis of this data without any additional experimentation can lead to surprising new biological insight.
  4. This bioinformatics activity is the truly integrating factor for the life sciences as a whole. Different subfields interact with each other via their entries in databases, which are analyzed as one body of data in bioinformatics research efforts. For example, mammalian signaling was connected with bacterial biochemistry in the Fringe story, and the functional characterization of the SET domain became possible with information on plant enzymes.
  5. In many instances, the theoretical analysis ended up in weak hypotheses that they could not be published directly due to lack of intrinsic logical rigor in their derivation. Therefore, many such hints will never appear in computer-generated public sequence database annotations, since it is considerable theoretical and/or experimental work to judge significance of a below-threshold hit. At the same time, pure experimental approaches would have hardly uncovered the catalytically active domains in the near future. The true partnership between both approaches (and people from different groups and complementing backgrounds) was the major precondition of success. However, strong hierarchies and star-centered organizations, typical for many life science research units, will have difficulties to accommodate such an interdisciplinary research style.
  6. Efforts in applied mathematical and biophysical research for biological sequence analysis, which have been smiled at in the biological community, have matured, and oftentimes reliable predictions or, at least useful hints for further experimental studies, can be made. The most important hallmark (but not the only one) is represented by the technique of sequence alignment and the evaluation of mutual substitution rates of amino acid types during evolution in the form of substitution matrices. This advance allowed the quantification of sequence similarity in rigorous probabilistic terms, essentially a distance measure between sequences.3 The concept of protein homology with the postulate of a common evolutionary ancestor for a family of sequentially similar gene segments is the basis for structure and function prediction by annotation transfer to uncharacterized proteins from experimentally analyzed homologues.8,9

The feeling of a mysterious appearance of the new biological insight results from the cultural collision of the world of experimental life scientists not trained in thinking with statistical or physical categories because of the previous lack of necessity. The flipside is the expectation of repeated wonder predictions with new sequence targets, an expectation that can hardly be satisfied in all instances. There are a large number of full sequences and sequence segments (as a rule of thumb, one third of an eukaryote proteome), where currently computational biology has to lay down its arms. The situation is even grimmer if the whole genome, instead of the protein-coding part (only ca. 1.5% of the genome), is considered.

Self-critically, the bioinformatics community has to acknowledge that, especially in early work, the way that the theoretical conclusion was achieved was described superficially, and for those who are not in the know, it can be difficult to repeat the deduction and to evaluate the judgments. The incomplete description of procedures is a general problem in life sciences. Repetition of published experimental recipes typically ends in failure. Sadly, the traditional Methods section in scientific articles was moved from the place following the Introduction to the end of the text and typed in small font, signaling its relative insignificance.

Astrology

Multiple views of the same topic have a right for existence. Since biological sequence analysis is not a fundamental science in the same sense as, for example, physics, it is sometimes jokingly called “sequence astrology”. This has several reasons:

Although Anfinsen10 has shown with renaturation experiments that the information contained in the protein sequence is generally sufficient to determine the protein structure, attempts to derive native structures ab initio (with fundamental physical principles) from protein sequences alone without additional high-resolution experimental information (macromolecular X-ray crystallography and NMR) have practically failed due to the system's complexity. If the assumption basis is less restrictive and includes atomic force fields empirically fitted to observation from small molecules, the protein-modeling problem is more tractable, but the intramolecular energy criteria become too inaccurate to distinguish between native and nonnative structures. Thus, ab initio approaches trying to compute a protein's structure directly from its sequence with fundamental physical principles are of little help in structure and, more so, function predictions for real life examples.

Thus, computational sequence analyses in biological application studies rely on empirically established sequence-structure and sequence-function relationships:

  1. The possibility of extrapolation of such correlation to uncharacterized sequences cannot fully be proven in principle, but it is silently assumed in practical applications. Considerable published research is devoted to establish statistical, physical, and/or biological criteria to assess the reliability and limits of such extrapolations (for example, Refs. 11-14).
  2. Further, dramatic simplifications are often necessary to treat the problems rigorously with mathematical means. For example, the sequence evolution models underlying current sequence comparison and alignment techniques do, as a rule, ignore the generally weak but possibly existing mutual dependence among sequence positions.
  3. At the beginning of working on a new prediction algorithm, the researcher compiles a so-called learning set of sequences with experimentally verified structural/functional properties. Usually, he encounters the first problem here: Since only the first discovery is granted with a high-impact publication, the experimental community pays little attention to increasing the list of verified examples, and the resulting learning set is (a) small and, probably (b) does not represent the full variety of naturally occurring sequences with the same feature. 15-19 To conclude, even if the researcher honestly tries not to fool himself and his colleagues in the field with prediction rate cosmetics, the estimates of accuracy have a tendency to be too good looking, both with respect to false-positive and false-negative predictions.

This causes a perception problem of sequence analysis-based predictions: Experimental researchers seeing all of the advertised ‘well-working’ prediction techniques around get disappointed since, for their specific target, the straightforwardly applied methods (especially if used with a restrictive parameter setting) appear not applicable, or in the other extreme, seem to produce obviously false output; thus, the seed of interest for nonexperimental approaches is nipped in the bud.

The typical experimental researcher also has difficulties to distinguish valuable suggestions from false hints. Here, we compare two examples of false predictions to make a frequently important point: The protein kinase activity suggested for scc2p20 might be argued with the incomplete conservation of a short sequence motif typical for kinases but can be ruled out by structural considerations showing scc2p being a HEAT-repeat protein.21,22 Generally, conserved short motifs involving only a handful of residues, especially with incomplete conservation of functionally important residues, are not indicative for evolutionarily retained function if there is no additional argument. Thus, a kinase assay for scc2p appears not indicated.

In contrast, the suspected ras-binding activity of myr-5 is based on the significant sequence similarity over a specific ca. 100 AA-long domain to proteins known from experimental reports to interact with ras in that region.23 Indeed, the structural similarity was further substantiated, but the binding activity could not.24 Nevertheless, the extrapolation of function from family members leads to a reasonable, productive hypothesis since no contrary arguments, such as nonconservation of known functional motifs or additional experimental data, could be found.

For some people having a strong background in mathematical and physical sciences, all this combination of formally not proven sequence-function correlation, of function extrapolation, and of intuition for still-correct borderline cases in sequence studies seems intellectually not pleasing. Nevertheless, relatively simple sequence-pattern recognition techniques produce surprisingly many reliable predictions. In this context, it appears necessary to emphasize the importance of evolutionary history in biology. As a result, the search space of real sequence alternatives appears dramatically shrinked compared to the theoretically thinkable set of variants compatible with fundamental laws.

It should also be noted here that experiments cannot be considered the final word in every instance either.25 Iyer et al26 list half a dozen of cases where the molecular function predicted using well tested, statistically sound computational means is in straight contradiction with experimental results, arguing for another enzymatic function (typically originally predicted from short motif occurrences or with a threading approach). Whereas execution of an enzyme activity by nonhomologous proteins implies an (possibly unlikely) unusual chemistry, over-interpretation of experimental data or lack of additional controls might have lead to the contradictory conclusion.

Service Technology

This is the third point of view with respect to bioinformatics. It is typically followed by computer science-driven researchers and by experienced leaders of large experimental units (especially in pharmaceutical industry) who want to extract value from complete genome sequencing and other high-throughput activities for research in their fields of life sciences. Bioinformatics is considered a service effort to store and retrieve biological information, to create integrated software solutions, and to apply existing suites of programs in a routine manner and to supply the necessary output immediately after a request from an experimentalist is issued. This view received additional backing with the availability of WWW-servers for sequence analyses (especially BLAST servers) that appear useable, similarly to TV sets, without necessarily understanding the algorithms applied.

Organizationally, such attitude results in a part-time bioinformatics person equipped with a PC linked to the internet in academia or a small service group with a high budget for ever larger and complete database and software license purchases in industry. Often, this personnel becomes unsatisfied after being overloaded with a large number of scientifically and methodically disconnected requests when, at the same time, they have to maintain their local working environment themselves. Of course, it is a serious limitation that not all necessary methods are available at WWW-servers, many reply slowly or not at all, and most predictors are disintegrated and have to be navigated individually, and is what causes extensive time losses. Not surprisingly, most nontrivial discoveries based on sequence analysis considerations have been made at other places. Specialized researchers who work at academic bioinformatics research centers with sufficient computing power and a local, well-maintained database and software environment, who have spent significant periods of their lives in analyzing sequence data and accumulated experience, who have taken part in methodical developments and interact with their colleagues in the field, have clearly greater chances of success.25

Already the unfortunate notion “bioinformatics” is derived from the superficial view of doing something in biology with computers, whereas the essence consists in research aimed at understanding the complex genotype-to-phenotype transformation relying on sequences, expression profiles, other types of high-throughput experimental data, and the electronically available scientific literature. Only the complexity of theoretical concepts and the large amount of data require the use of computers as a tool. It is ever more important to understand that sequence analysis and generally computational biology is a field of ongoing research with many problems still unsolved. Except for obvious cases of sequence homology and sequence domain matches, considerable creativity is required for nontrivial application of existing methods to squeeze out a hint on structure/function of a target considered out of its sequence. Since many functionally important sequence signatures still remain hidden, the development of new techniques recognizing more distantly related sequence family members or other biological features, such as many posttranslational modifications or localization signals, continue to have important academic and practical relevance.

Finally, who is servicing whom? Is it more important for new biological insight to generate a new hypothesis based on theoretical data analysis, or to create an experimental setup for its verification? Running simple BLAST searches or standard electrophoresis gels are equally routine (service?) activities. I think that such a discussion is counterproductive and distracts from the main issue, that of the scientific search coordinated among researchers with complementing professional backgrounds who strive for a common success.

The development of high-throughput experimental technologies and its first major breakthrough, the complete sequencing of the genomes of organisms ranging from viruses to bacteria, lower eukaryotes to human, has changed life science research qualitatively. For the first time in history, the biological object can be studied in its totality at the molecular level. The immediate task for the coming decade consists in assigning functions to all genes known by sequence. Apparently, sensible quantitative gene network studies will be possible only after most of the genes have been assigned a function qualitatively and all major players of processes and their interaction topology are known. Since the new data are so large and their biological interpretation requires complex approaches, theoretical science can and must contribute decisively to the research progress. Hints from theoretical studies may shorten experimental searches and save postdoc years.

Obviously, there is a new division of labour in the cognition process in life sciences. Both sides have to learn to translate prediction results into new experimental designs and to formulate new theoretical tasks based on the existing experimental data. There are now many examples of computational biology helping experimental science, and more decisive support can be expected in the future. However, the interaction cannot remain a one-way direction. The value of experimental work will be increasingly measured via its effect on the improvement of methods in computational biology and the thereby-increased power of extrapolation into the unknown part of the genome. For example, this includes the experimental generation of learning sets for sequence-function correlations with reliable methods to boost prediction method development, an aspect that has not received sufficient attention yet. Only a repeatedly iterative interaction promises to produce maximal research progress. The experimental community should not worry: The gaps in theory are still and will remain large.

The book “Discovering Biomolecular Mechanisms with Computational Biology” unites a collection of articles of researchers in theoretical biology showing areas of life science where theory has contributed with considerable impact. Martijn A. Huynen et al, Karin Schleinkofer, Thomas Dandekar, and I review methodical approaches for analyzing biomolecular sequences and structures in section I, accompanying the text with examples of research breakthroughs. Since the scientific literature is dramatically expanding and, at the same time, becomes increasingly electronically accessible, dedicated text analysis tools for biomedical reports can link them to genomic features and contribute decisively to prioritizing research targets. Hong Pan et al and Carolina Perez-Iratxeta et al summarize the state of the art in section II. Gene and protein network analysis is still in its infancy, since many network nodes remain unknown and the quantitative characterization of most gene/protein functions is missing. Nevertheless, studies of regulatory and metabolic networks already produce now-important insights with theoretical work alone, as Harmen Bussemaker and Stefan Schuster et al show in section III. Finally, Edward N. Trifonov, Christian Schlötterer, Saurabh Astana, and Shamil Sunyaev and Yurij I. Wolf et al present compelling evidence in section IV that the evolutionary viewpoint is indispensable for understanding function and interaction of today's genes and proteins. This book will fulfill its task if the examples described here encourage the readers to find new areas in life science research where theoretical research can qualitatively change the rate of progress in understanding biomolecular mechanisms and help to move toward a quantitative and predictive biology.

References

1.
Hagen JB. The origins of bioinformatics. Nat Rev Genet. 2000;1:231–236. [PubMed: 11252753]
2.
Ouzounis C. Bioinformatics and the theoretical foundations of molecular biology. Bioinformatics. 2002;18:377–378. [PubMed: 11934735]
3.
Altschul S, Boguski M, Gish W. et al. Issues in searching molecular sequence databases. Nature Genetics. 1994;6:119–129. [PubMed: 8162065]
4.
Dlakic M. Chromatin silencing protein and pachytene checkpoint regulator dot1p has a methyltransferase fold. Trends Biochem Sci. 2001;26:405–407. [PubMed: 11440840]
5.
van LeeuwenF, Gafken PR, Gottschling DE. Dot1p modulates silencing in yeast by methylation of the nucleosome core. Cell. 2002;109:745–756. [PubMed: 12086673]
6.
Toth A, Ciosk R, Uhlmann F. et al. Yeast cohesin complex requires a conserved protein, Eco1p(Ctf7), to establish cohesion between sister chromatids during DNA replication. Genes Dev. 1999;13:320–333. [PMC free article: PMC316435] [PubMed: 9990856]
7.
Ivanov D, Schleiffer A, Eisenhaber F. et al. Eco1 is a novel acetyltransferase that can acetylate proteins involved in cohesion. Curr Biol. 2002;12:323–328. [PubMed: 11864574]
8.
Bork P, Dandekar T, Diaz-Lazcoz Y. et al. Predicting function: From genes to genomes and back. J Mol Biol. 1998;283:707–725. [PubMed: 9790834]
9.
Ponting CP, Schultz J, Copley RR. et al. Evolution of domain families. Adv Protein Chem. 2000;54:185–244. [PubMed: 10829229]
10.
Anfinsen CB. Principles that govern the folding of protein chains. Science. 1973;181:223–230. [PubMed: 4124164]
11.
Sander C, Schneider R. Database of Homology-derived protein structures and the structural meaning of sequence alignment. Proteins. 1991;9:56–68. [PubMed: 2017436]
12.
Bork P, Gibson TJ. Applying motif and profile searches. Meth Enzymol. 1996;266:162–184. [PubMed: 8743684]
13.
Altschul SF, Koonin EV. Iterated profile searches with PSI-BLAST-a tool for discovery in protein databases. Trends Biochem Sci. 1998;23:444–447. [PubMed: 9852764]
14.
Devos D, Valencia A. Practical limits of function prediction. Proteins. 2000;41:98–107. [PubMed: 10944397]
15.
Nielsen H, Engelbrecht J, Brunak S. et al. Identification of procaryotic and eucaryotic signal peptides and prediction of their cleavage sites. Protein Eng. 1997;10:1–6. [PubMed: 9051728]
16.
Eisenhaber B, Bork P, Eisenhaber F. Sequence properties of GPI-anchored proteins near the .-Site: Constraints for the polypeptide binding site of the putative transamidase. Protein Eng. 1998;11:1155–1161. [PubMed: 9930665]
17.
Eisenhaber B, Bork P, Eisenhaber F. Prediction of potential GPI-modification sites in proprotein sequences. J Mol Biol. 1999;292:741–758. [PubMed: 10497036]
18.
Maurer-Stroh S, Eisenhaber B, Eisenhaber F. N-terminal N-myristoylation of proteins: Prediction of substrate proteins from amino acid sequence. J Mol Biol. 2002;317:541–557. [PubMed: 11955008]
19.
Eisenhaber F, Eisenhaber B, Maurer-Stroh S. Prediction of Post-translational modifications from amino acid sequence: Problems, pitfalls, methodological hintsIn: Andrade MM, ed.Bioinformatics and Genomes: Current PerspectivesWymondham: Horizon Scientific Press,2003. Chapter 581–105.
20.
Jones S, Sgouros J. The cohesin complex: Sequence homologies, interaction networks and shared motifs Genome Biol 2001. 2(RESEARCH0009) [PMC free article: PMC30708] [PubMed: 11276426]
21.
Neuwald AF, Hirano T. Heat repeats associated with condensins, cohesins, and other complexes involved in Chromosome-related functions. Genome Res. 2000;10:1445–1452. [PMC free article: PMC310966] [PubMed: 11042144]
22.
Panizza S, Tanaka T, Hochwagen A. et al. Pds5 cooperates with cohesin in maintaining sister chromatid cohesion. Curr Biol. 2000;10:1557–1564. [PubMed: 11137006]
23.
Ponting CP, Benjamin DR. A novel family of Ras-binding domains. Trends Biochem Sci. 1996;21:422–425. [PubMed: 8987396]
24.
Kalhammer G, Bahler M, Schmitz F. et al. Ras-binding domains: Predicting function versus folding. FEBS Lett. 1997;414:599–602. [PubMed: 9323044]
25.
Ouzounis C. Two or three myths about bioinformatics. Bioinformatics. 2000;16:187–189. [PubMed: 10869011]
26.
Iyer LM, Aravind L, Bork P. et al. Quoderat demonstrandum? The mystery of experimental validation of apparently erroneous computational analyses of protein sequences Genome Biol 2001. 2(RESEARCH0051) [PMC free article: PMC64836] [PubMed: 11790254]
27.
Yuan YP, Schultz J, Mlodzik M. et al. Secreted Fringe-like signaling molecules may be glycosyltransferases. Cell. 1997;88:9–11. [PubMed: 9019410]
28.
Tatusov RL, Altschul SF, Koonin EV. Detection of conserved segments in proteins: Iterative scanning of sequence databases with alignment blocks. Proc Nat Acad Sci USA. 1994;91:12091–12095. [PMC free article: PMC45382] [PubMed: 7991589]
29.
Altschul SF, Madden TL, Schaffer AA. et al. Gapped BLAST and PSI-BLAST: A new generation of protein database search programs. Nucleic Acids Res. 1997;25:3389–3402. [PMC free article: PMC146917] [PubMed: 9254694]
30.
Irvine KD, Wieschaus E. Fringe, a Boundary-specific signaling molecule, mediates interactions between dorsal and ventral cells during drosophila wing development. Cell. 1994;79:595–606. [PubMed: 7954826]
31.
Wu JY, Wen L, Zhang WJ. et al. The secreted product of xenopus gene lunatic fringe, a vertebrate signaling molecule. Science. 1996;273:355–358. [PMC free article: PMC2080353] [PubMed: 8662522]
32.
Panin VM, Papayannopoulos V, Wilson R. et al. Fringe modulates Notch-ligand interactions. Nature. 1997;387:908–912. [PubMed: 9202123]
33.
Fleming RJ, Gu Y, Hukriede NA. Serrate-mediated activation of notch is specifically blocked by the product of the gene fringe in the dorsal compartment of the drosophila wing imaginal disc. Development. 1997;124:2973–2981. [PubMed: 9247339]
34.
Klein T, Arias AM. Interactions among delta, serrate and fringe modulate notch activity during drosophila wing development. Development. 1998;125:2951–2962. [PubMed: 9655817]
35.
Moloney DJ, Panin VM, Johnston SH. et al. Fringe is a glycosyltransferase that modifies notch. Nature. 2000;406:369–375. [PubMed: 10935626]
36.
Bruckner K, Perez L, Clausen H. et al. Glycosyltransferase activity of fringe modulates Notch-delta interactions. Nature. 2000;406:411–415. [PubMed: 10935637]
37.
Nasmyth K. Separating sister chromatids. Trends Biochem Sci. 1999;24:98–104. [PubMed: 10203756]
38.
Ciosk R, Zachariae W, Michaelis C. et al. An ESP1/PDS1 complex regulates loss of sister chromatid cohesion at the metaphase to anaphase transition in yeast. Cell. 1998;93:1067–1076. [PubMed: 9635435]
39.
Uhlmann F, Lottspeich F, Nasmyth K. Sister-chromatid separation at anaphase onset is promoted by cleavage of the cohesin subunit Scc1. Nature. 1999;400:37–42. [PubMed: 10403247]
40.
Chen JM, Rawlings ND, Stevens RA. et al. Identification of the active site of legumain links it to caspases, clostripain and gingipains in a new clan of cysteine endopeptidases. FEBS Lett. 1998;441:361–365. [PubMed: 9891971]
41.
Eichinger A, Beisel HG, Jacob U. et al. Crystal structure of gingipain R: An Arg-specific bacterial cysteine proteinase with a Caspase-like fold. EMBO J. 1999;18:5453–5462. [PMC free article: PMC1171614] [PubMed: 10523290]
42.
Uhlmann F, Wernic D, Poupart MA. et al. Cleavage of cohesin by the CD clan protease separin triggers anaphase in yeast. Cell. 2000;103:375–386. [PubMed: 11081625]
43.
Jenuwein T, Laible G, Dorn R. et al. SET domain proteins modulate chromatin domains in Euand heterochromatin. Cell Mol Life Sci. 1998;54:80–93. [PubMed: 9487389]
44.
Aagaard L, Laible G, Selenko P. et al. Functional mammalian homologues of the drosophila PEV-modifier Su(Var)3- 9 encode Centromereassociated proteins which complex with the heterochromatin component M31. EMBO J. 1999;18:1923–1938. [PMC free article: PMC1171278] [PubMed: 10202156]
45.
O'Carroll D, Scherthan H, Peters AH. et al. Isolation and characterization of Suv39h2, a second histone H3 methyltransferase gene that displays Testis-specific expression. Mol Cell Biol. 2000;20:9423–9433. [PMC free article: PMC102198] [PubMed: 11094092]
46.
Klein RR, Houtz RL. Cloning and developmental expression of pea ribulose-1,5-bisphosphate carboxylase/oxygenase large subunit N-methyltransferase. Plant Mol Biol. 1995;27:249–261. [PubMed: 7888616]
47.
Rea S, Eisenhaber F, O'Carroll D. et al. Regulation of chromatin structure by Site-specific histone H3 methyltransferases. Nature. 2000;406:593–599. [PubMed: 10949293]
48.
Lachner M, O'Carroll D, Rea S. et al. Methylation of histone H3 lysine 9 creates a binding site for HP1 proteins. Nature. 2001;410:116–120. [PubMed: 11242053]
49.
Bannister AJ, Zegerman P, Partridge JF. et al. Selective recognition of methylated lysine 9 on histone H3 by the HP1 chromo domain. Nature. 2001;410:120–124. [PubMed: 11242054]
Copyright © 2000-2013, Landes Bioscience.
Bookshelf ID: NBK6396
PubReader format: click here to try

Views

  • PubReader
  • Print View
  • Cite this Page

Related information

  • PMC
    PubMed Central citations
  • PubMed
    Links to pubmed

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...