This Article Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via CrossRef Citing Articles via Web of Science (1) Citing Articles via Google Scholar Google Scholar Articles by Briggs, S. P. Articles by Singer, T. Search for Related Content PubMed PubMed Citation Articles by Briggs, S. P. Articles by Singer, T. Agricola Articles by Briggs, S. P. Articles by Singer, T.

VISION STATEMENT FOR PLANT PHYSIOLOGY

Genetic Networks

Section of Cell and Developmental Biology, Division of Biological Sciences, University of California, San Diego, La Jolla, California 92093–0380

The discovery of genetic networks has been accelerated by the development of microarray technology. Microarrays and associated equipment and reagents can be manufactured to high standards of reproducibility and the data are digital. Thus, it's feasible to create reference datasets from unrelated experiments for comparison with new samples. For example, a comparison of control sample A to test sample B may show changes in expression of 500 genes. But comparison of control sample A with the reference dataset may reveal that only 10 of the 500 changes are unique to the comparison with test sample B; this has been referred to as a vertical search (Zhu and Wang, 2000), and it can dramatically enrich the list for genes of interest. An organized effort to curate a reference dataset and provide a vertical search capability to investigators who make A x B comparisons would be useful.

Recently, a second genetic network has been discovered that is composed of microRNAs (miRNAs). In humans, 1% of the genes are miRNAs, and they regulate protein levels of 10% of the genes. In Arabidopsis (Arabidopsis thaliana), up to 7% of the genes may be miRNA/silencing RNA (Gustafson et al., 2005); probably 10% or more of the genes are targets. Because target specificity is based upon nucleotide sequence homology, it is possible to predict miRNA/target pairs (Adai et al., 2005). Microarrays that contain unique flanking sequence from both transcripts of all such pairs could be useful in sorting out the regulation of the miRNA network. Regulation of protein function by direct binding of short, sequence-specific double-stranded RNA has been described in mammals but not yet in plants (Kuwabara et al., 2004).

Work has begun to identify the DNA targets of transcription factors. Chromatin immunoprecipitation reveals whether a particular protein is bound to a given gene in certain cells or tissues. Whole-genome tiling arrays or promoter arrays may enable all of the genes that are bound by a particular protein to be identified, although current methods are not robust. There has been tantalizing progress with whole-genome tiling arrays for Arabidopsis, but their cost remains prohibitive for a systematic study of DNA-binding proteins (Mockler and Ecker, 2005). Promoter arrays are limited by the quality of genome annotation, but they are smaller—and less expensive—than genome tiling arrays. An Arabidopsis promoter array could facilitate the construction of genetic networks downstream of a DNA-binding protein.

What about the network upstream of a gene? Investigators can determine whether a gene that is immunoprecipitated by one protein is also bound by another, indicating that both bind upstream of the same gene. But only candidate proteins can be tested; unsuspected candidates cannot be discovered. Considerable success has been achieved by searching for mutants that change the expression of promoter-reporter constructs (Chinnusamy et al., 2003). This does not distinguish direct from indirect effects, but an upstream network can be defined.

Another approach to mapping networks is to repress expression of one gene and then look to see which other genes respond. It is not possible to distinguish direct from indirect effects, but, nevertheless, gene disruption is a powerful approach. Maize (Zea mays) was among the first organisms to have a full set of gene disruption mutants that is addressable by DNA probes (Bensen et al., 1995), and then T-DNA collections in Arabidopsis were characterized according to their insertion sites (Sessions et al., 2002). But without some way to multiplex the plant mutant collections, it has not been feasible to build large transcription networks using gene disruptions.

By contrast, the use of DNA barcodes to mark deletion mutant strains has permitted genetic networks for fitness to be revealed in yeast (Saccharomyces cerevisiae). Each individual gene deletion mutant is combined in equal numbers and grown competitively in a population. The proportion of each strain remaining in the population is assayed using a microarray that detects the barcodes (Giaever et al., 2002). Mutants that disappear or become overly abundant define the fitness genes for growth under these conditions. This approach has been applied to the discovery of synthetic lethals: strains that contain two defective genes that individually are viable but together are lethal. So far, only a few hundred mutant genes have been used to query the full set for lethal combinations but already interesting patterns have emerged. On average, each gene has 30 lethal partners, and typically their function is related to that of the query gene. Often, many genes form synthetic lethals with the same set of 30 partners but not with each other. In these cases, the 30 genes form a pathway, and their interactors are part of a protein complex. It's estimated that in yeast there are 100,000 synthetic lethal interactions involving 5,000 viable gene deletion mutants compared with 25,000 protein-protein interactions (Tong et al., 2004).This method of analysis is so powerful yet simple that it clearly should be done with plants. The first step will be to create homozygous mutants from each member of the T-DNA collections. This will reveal the genes that individually are necessary for viability. They will be removed, and the remaining mutants will be intercrossed to find synthetic lethals. Lethality will appear in the progeny of the F1 plants either as aborted seed, dead seed, or plants that die after germination at a ratio of 1/16; pollen could be examined to detect gametophyte synthetic lethals.

While the use of synthetic lethality in whole plants will grow in importance, particularly when focused on small numbers of filtered genes, there are practical considerations that temper enthusiasm. To make all possible pairwise combinations will require approximately 350 million crosses, followed by recovery and growth of seed and then examination of the next generation. Even if the entire community was mobilized to handle 1 million crosses per year, it is likely that alternative approaches would supersede this effort. To mimic the yeast system in plants, transient expression of bar-coded, dominant gene suppressors could be applied to populations of dividing plant cells. A library that contains suppressors for every plant gene could be mobilized into cells and then monitored by amplifying the barcodes and hybridizing to a microarray. Barcodes that disappear are linked to genes that are required for viability. Then, a query gene could be added with the library and the microarray results compared. Barcodes (genes) that disappear when the query gene is added reveal synthetic lethals. Virus-induced gene silencing (Lu et al., 2003) or artificial transcription factors (Guan et al., 2002) could be used as dominant gene suppressors. The entire network could be revealed with only 27,000 queries rather than 350 million.

Strikingly absent from the literature are networks revealed by overexpressing genes. There is no obvious biological reason for this. Rather, it is generally much more difficult and expensive to activate genes than to disrupt them. Artificial transcription factors may help. They can be changed from repressors to activators simply by swapping domains. A library of artificial transcription factors could be made in two versions, one to repress and one to activate the target genes. It would be interesting to compare the two sets of synthetic lethals produced this way. Another advantage of artificial transcription factors is that, when fused to a protein transduction domain, they can be easily purified from Escherichia coli and added to cells as proteins, eliminating the cost and variability associated with transformation (Tachikawa et al., 2004).

As an alternative to ectopic expression, wild-type but variant allelic substitutions can be made at every locus and their effects on the mRNA levels of every gene measured. We've taken this approach by using recombinant inbred lines to multiplex the allelic substitutions and by using microarrays to measure mRNA levels. A remarkable degree of connectivity between genes was found, indicating that members of coordinately regulated protein complexes or biochemical pathways are themselves the regulators (T. Singer, Y. Fan, and S.P. Briggs, unpublished data). Arabidopsis transcriptional genetic networks are being drawn from these data. Studies in yeast lead to similar conclusions as ours: most regulators act in trans, and they are not transcription factors (Yvert et al., 2003).

Microarrays and chromatin immunoprecipitation can only predict changes in protein and metabolite levels. A large fraction of regulation is posttranscriptional, and this is undetected or counterindicated by changes in mRNA levels. For example, a combined microarray and proteomic survey showed that transfer of yeast from complete to minimal medium increased the levels of Tup1, a transcriptional repressor, by 8-fold without increasing mRNA levels (Washburn et al., 2003). A proteomic survey of rice (Oryza sativa) has shown that plants present no particular obstacle to this level of analysis (Koller et al., 2002). Like protein levels, the levels of metabolites in plants can be uncoupled from changes at the mRNA level (Laule et al., 2003). These observations underscore a key difference between biochemical networks, which are mechanistic but make unreliable predictions of relationships, and genetic networks, which reveal quantitative relationships but not mechanisms.

Combining multiple genetic networks to find genes at their intersection can be more effective than examining a single network. A rice genome microarray detected the induction of the suppressor of the G2 allele of Skp1 (SGT1) gene caused by rice blast. In barley (Hordeum vulgare) and Arabidopsis, SGT1 is required for disease resistance. The investigators found several other rice proteins that interacted with rice SGT1 in a yeast two-hybrid assay. The microarray revealed that one of these, elicitor-response protein (ERP), was induced by stress, and others reported its induction by an elicitor from the rice blast fungus. ERP itself, when used as bait, interacted still with other proteins, including one that was undefined based upon a lack of homology with defined proteins. For reasons given above, this undefined protein plus four that interacted directly with SGT1 (including ERP) were predicted to be required for disease resistance in rice. Orthologs of these rice genes were evaluated using Arabidopsis T-DNA mutants. Each mutation significantly disrupted disease resistance (Cooper et al., 2003).

	FOOTNOTES

 
www.plantphysiol.org/cgi/doi/10.1104/pp.104.900147.

^* Corresponding author; e-mail sbriggs{at}ucsd.edu; fax 858–534–5543.

	LITERATURE CITED

TOP LITERATURE CITED

 
Adai A, Johnson C, Mlotshwa S, Archer-Evans S, Manocha V, Vance V, Sundaresan V (2005) Computational prediction of miRNAs in Arabidopsis thaliana. Genome Res 15: 78–91[Abstract/Free Full Text]

Bensen RJ, Johal GS, Crane VC, Tossberg JT, Schnable PS, Meeley RB, Briggs SP (1995) Cloning and characterization of the maize An1 gene. Plant Cell 7: 75–84[Abstract]

Chinnusamy V, Ohta M, Kanrar S, Lee BH, Hong X, Agarwal M, Zhu JK (2003) ICE1: a regulator of cold-induced transcriptome and freezing tolerance in Arabidopsis. Genes Dev 17: 1043–1054[Abstract/Free Full Text]

Cooper B, Clarke JD, Budworth P, Kreps J, Hutchison D, Park S, Guimil S, Dunn M, Luginbuhl P, Ellero C, et al (2003) A network of rice genes associated with stress response and seed development. Proc Natl Acad Sci USA 100: 4945–4950[Abstract/Free Full Text]

Giaever G, Chu AM, Ni L, Connelly C, Riles L, Veronneau S, Dow S, Lucau-Danila A, Anderson K, Andre B, et al (2002) Functional profiling of the Saccharomyces cerevisiae genome. Nature 418: 387–391[CrossRef][Medline]

Guan X, Stege J, Kim M, Dahmani Z, Fan N, Heifetz P, Barbas CF III, Briggs SP (2002) Heritable endogenous gene regulation in plants with designed polydactyl zinc finger transcription factors. Proc Natl Acad Sci USA 99: 13296–13301[Abstract/Free Full Text]

Gustafson AM, Allen E, Givan S, Smith D, Carrington JC, Kasschau KD (2005) ASRP: the Arabidopsis Small RNA Project Database. Nucleic Acids Res (Database issue) 33: D637–D640[Abstract/Free Full Text]

Koller A, Washburn MP, Lange BM, Andon NL, Deciu C, Haynes PA, Hays L, Schieltz D, Ulaszek R, Wei J, et al (2002) Proteomic survey of metabolic pathways in rice. Proc Natl Acad Sci USA 99: 11969–11974[Abstract/Free Full Text]

Kuwabara T, Hsieh J, Nakashima K, Taira K, Gage FH (2004) A small modulatory dsRNA specifies the fate of adult neural stem cells. Cell 116: 779–793[CrossRef][Web of Science][Medline]

Laule O, Furholz A, Chang HS, Zhu T, Wang X, Heifetz PB, Gruissem W, Lange M (2003) Crosstalk between cytosolic and plastidial pathways of isoprenoid biosynthesis in Arabidopsis thaliana. Proc Natl Acad Sci USA 100: 6866–6871[Abstract/Free Full Text]

Lu R, Martin-Hernandez AM, Peart JR, Malcuit I, Baulcombe DC (2003) Virus-induced gene silencing in plants. Methods 30: 296–303[CrossRef][Web of Science][Medline]

Mockler TC, Ecker JR (2005) Applications of DNA tiling arrays for whole-genome analysis. Genomics 85: 1–15[CrossRef][Web of Science][Medline]

Sessions A, Burke E, Presting G, Aux G, McElver J, Patton D, Dietrich B, Ho P, Bacwaden J, Ko C, et al (2002) A high-throughput Arabidopsis reverse genetics system. Plant Cell 14: 2985–2994[Abstract/Free Full Text]

Tachikawa K, Schroder O, Frey G, Briggs SP, Sera T (2004) Regulation of the endogenous VEGF-A gene by exogenous designed regulatory proteins. Proc Natl Acad Sci USA 101: 15225–15230[Abstract/Free Full Text]

Tong AH, Lesage G, Bader GD, Ding H, Xu H, Xin X, Young J, Berriz GF, Brost RL, Chang M, et al (2004) Global mapping of the yeast genetic interaction network. Science 303: 808–813[Abstract/Free Full Text]

Washburn MP, Koller A, Oshiro G, Ulaszek RR, Plouffe D, Deciu C, Winzeler E, Yates JR III (2003) Protein pathway and complex clustering of correlated mRNA and protein expression analyses in Saccharomyces cerevisiae. Proc Natl Acad Sci USA 100: 3107–3112[Abstract/Free Full Text]

Yvert G, Brem RB, Whittle J, Akey JM, Foss E, Smith EN, Mackelprang R, Kruglyak L (2003) Trans-acting regulatory variation in Saccharomyces cerevisiae and the role of transcription factors. Nat Genet 35: 57–64[Web of Science][Medline]

Zhu T, Wang X (2000) Large-scale profiling of the Arabidopsis transcriptome. Plant Physiol 124: 1472–1476[Free Full Text]

This Article Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via CrossRef Citing Articles via Web of Science (1) Citing Articles via Google Scholar Google Scholar Articles by Briggs, S. P. Articles by Singer, T. Search for Related Content PubMed PubMed Citation Articles by Briggs, S. P. Articles by Singer, T. Agricola Articles by Briggs, S. P. Articles by Singer, T.