Transcript profiling of early lateral root initiation - Proceedings of the ...

Transcript profiling of early lateral root initiation Kristiina Himanen*†, Marnik Vuylsteke*, Steffen Vanneste*, Steven Vercruysse*, Elodie Boucheron‡, Philippe Alard*, Dominique Chriqui‡, Marc Van Montagu*, Dirk Inze´*§, and Tom Beeckman* *Department of Plant Systems Biology, Flanders Interuniversity Institute for Biotechnology, Ghent University, B-9052 Ghent, Belgium; and ‡Universite´ Pierre et Marie Curie, F-75252 Paris Cedex 05, France Contributed by Marc Van Montagu, January 5, 2004

At the onset of lateral root initiation in Arabidopsis thaliana, the phytohormone auxin activates xylem pole pericycle cells for asymmetric cell division. However, the molecular events leading from auxin to lateral root initiation are poorly understood, in part because the few responsive cells in the process are embedded in the root and are thus difficult to access. A lateral root induction system, in which most xylem pole pericycle cells were synchronously activated by auxin transport inhibition followed by auxin application, was used for microarray transcript profiling. Of 4,600 genes analyzed, 906 significantly differentially regulated genes were identified that could be grouped into six major clusters. Basically, three major patterns were discerned representing induced, repressed, and transiently expressed genes. Analysis of the coregulated genes, which were specific for each time point, provided new insight into the molecular regulation and signal transduction preceding lateral root initiation in Arabidopsis. The reproducible expression profiles during a time course allowed us to define four stages that precede the cell division in the pericycle. These early stages were characterized by G1 cell cycle block, auxin perception, and signal transduction, followed by progression over G1兾S transition and G2兾M transition. All these processes took place within 6 h after transfer from N-1-naphthylphthalamic acid to 1-naphthalene acetic acid. These results indicate that this lateral root induction system represents a unique synchronized system that allows the systematic study of the developmental program upstream of the cell cycle activation during lateral root initiation. auxin 兩 cell cycle 兩 microarray

D

uring the establishment of plant root architecture, signals from inside and outside the plant are transmitted to cells in the pericycle where the initiation of lateral roots takes place (1, 2). Genetic studies have shown that mutants deficient in auxin responses often have altered lateral root development (3). These studies have revealed a key role for transcriptional regulation as well as for proteolytic activities in mediating auxin responses (4–8). Furthermore, the regulation of auxin homeostasis by means of biosynthesis and transport, including active influx and efflux, also affects root growth and development (9–11). Taken together, auxin signaling is very complex, and various responses, such as cell division and cell expansion, appear to be mediated by different regulatory pathways (12, 13). Despite the large amount of data linking auxin and lateral root initiation, the precise signaling cascades involved in the process are still poorly understood (7, 14). Detailed molecular studies of lateral root initiation have been seriously hampered by the small number of cells involved in the initiation event as well as by the lack of synchrony between different initiation sites. Recently, we have characterized a lateral root induction system (LRIS) that is based on germination in the presence of the auxin transport inhibitor N-1naphthylphthalamic acid (NPA) followed by transfer to growth medium containing the auxin 1-naphthalene acetic acid (NAA) (15). In brief, by using several promoter-␤-glucuronidase reporter lines and兾or performing RT-PCR analysis for cell cycleand auxin-responsive genes, we could demonstrate a protoxylem pole-specific auxin response of pericycle cells in this system. The 5146 –5151 兩 PNAS 兩 April 6, 2004 兩 vol. 101 兩 no. 14

only tissues other than the pericycle responding to the auxin treatment were the root apical meristem and the hypocotyl. Therefore, these tissues were dissected and excluded from our analyses. The recorded auxin response in the pericycle resulted in asymmetric divisions, restricted to the protoxylem pole pericycle cells, that finally gave rise to two longitudinal rows of lateral root primordia covering the entire root. This pericycle reaction was so abrupt and nearly synchronous throughout the entire root that this feature opened the way to use it as a lateral rootinducible system for transcript profiling studies. We report on a genomic study based on this system to characterize the early molecular regulation induced by auxin, which activates the pericycle for cell division and lateral root formation. Changes in the root transcriptome were monitored during an early LRIS time course with microarrays that contained 4,575 unigene cDNA clones of Arabidopsis thaliana. An extensive statistical analysis to test for significant changes in transcript levels across time points provided a list of 906 genes showing reproducible differential expression over time. Cluster analysis and functional grouping of the coregulated genes revealed distinct, developmental stages preceding the cell division in the pericycle. Auxin signal perception and transduction was followed by induction of marker genes specific for G1兾S cell cycle transition and activation of protein translation machinery, after which xylem pole pericycle gained mitotic activity as indicated by the cytoplasmic changes and induction of G2兾M-specific genes. Materials and Methods One thousand two hundred seeds of the CYCB1;1::uidA line (16) of A. thaliana (L.) Heynh. were germinated on media containing 10 ␮M of the auxin transport inhibitor NPA. Three days after germination the seedlings were transferred on media with NAA (10 ␮M) but without NPA for 0 h, 2 h, 4 h, and 6 h as described (15). Root segments of 5 mm were collected by cutting away the root apical and adventitious root meristems, and ⬇300 root segments were pooled per sample and each sample was biologically replicated. The two replicated sample sets were labeled with different fluorescence dyes for microarray hybridization. The microarray experiment was set up as a reference design in which each sample was compared with one reference sample. As a reference sample, entire root systems from 2-week-old in vitro grown seedlings were used, in which most of the transcripts of the expressed genes of seedling roots were expected to be present. Total RNA was isolated with RNeasy (Qiagen, Hilden, Germany). Antisense RNA was amplified according to the protocol of in vitro transcription (17). The microarrays consisted of 4,608 cDNAs that were spotted in duplicate and distant from each other on Type V silane-coated slides (Amersham Biosciences). The clone set included 4,575 Arabidopsis genes from the unigene clone Incyte collection Abbreviations: funcat, functional category; IAA, indole-3-acetic acid; LRIS, lateral rootinducible system; NAA, 1-naphthalene acetic acid; NPA, N-1-naphthylphthalamic acid. †Present

address: Laboratory of Plant Molecular Biology, The Rockefeller University, 1230 York Avenue, New York, NY 10021.

§To

whom correspondence should be addressed. E-mail: [email protected].

© 2004 by The National Academy of Sciences of the USA

www.pnas.org兾cgi兾doi兾10.1073兾pnas.0308702101

Himanen et al.

Fig. 1. (A) Schematic presentation of the sampling for transmission electron microscopy (TEM). The scissors indicate the position where the sections were taken. The highlighted box on the scheme shows the positioning of the micrographs represented in B and C. (B and C) TEM micrographs of transverse sections of NPA- (B) and NAA-treated (C) Arabidopsis roots. Note the strongly vacuolated cells after NPA treatment (B) and the appearance of dense cytoplasm after transfer to NAA (C) in the pericycle at the protoxylem poles. Pe, xylem pole pericycle cells; PX, protoxylem (arrows).

nm) were cut with an OmU3 ultramicrotome (Reichert, Vienna) and collected on uncoated 200 hexagonal mesh copper grids. Sections were stained with a saturated solution of uranyl acetate in 50% EtOH and subsequently with lead citrate (23) for 15 min each. Grids were observed at 80 kV on a EM 201 transmission electron microscope (Philips, Einhoven, The Netherlands). At least seventeen sections from the axial half of four to six roots per time point were analyzed. Results and Discussion Xylem Pole Pericycle-Specific Cell Cycle Activation. Previously, we have demonstrated (15) that with LRIS, synchronous lateral root formation can be induced experimentally. To further characterize the LRIS, we first compared the noninduced (0 h) and induced (6 h) root samples at cytological level with transmission electron microscopy (Fig. 1A). The most striking difference between the two samples was found in the pericycle cells positioned at the protoxylem poles. In the noninduced samples, protoxylem pericycle cells contained large vacuoles (Fig. 1B) and appeared to be fully differentiated, whereas only these cells gained meristematic appearance, as indicated by development of dense cytoplasm after 6 h of auxin treatment (Fig. 1C). In the same sections, all other cell layers remained vacuolated after 6 h of treatment with NAA. Such distinct cellular changes are indicative of the specificity of the system that activates exclusively the xylem pole pericycle for lateral root initiation. Because PNAS 兩 April 6, 2004 兩 vol. 101 兩 no. 14 兩 5147

PLANT BIOLOGY

(Incyte Microarray Systems, Fremont, CA), which had been isolated from cDNA libraries representing all major organs and tissues of 4- and 6-week-old plants, and a set of control genes (see www.microarray.be and www.psb.ugent.be兾papers兾 lateralroot). For each clone spotted on the arrays, an Arabidopsis Genome Initiative code was assigned through BLAST (18) with the nucleic acid sequences of the Arabidopsis genome Databases (MAtDB) available at the Munich Information Center for Protein Sequences (MIPS; ftp:兾兾ftpmips.gsf.de兾cress兾arabidna). To identify the principal biological activities represented by each clone, they were classified according to the functional categories (funcats) of their ten best sequence homologs automatically derived from MAtDB at http:兾兾mips.gsf.de兾 (for detailed protocol, see www.psb.ugent.be兾papers兾lateralroot). Arrays were scanned with a Generation III scanner (Amersham Biosciences) at 532 nm and 635 nm wavelengths for the two fluorescence dyes Cy3 and Cy5 (Amersham Biosciences), respectively. Image analysis was performed with ArrayVision (Imaging Research, St. Catharines, Ontario, Canada). Spot intensities were measured as artifact-removed total intensities without correction for background. Variation in gene expression was assessed by using a two-step procedure essentially as outlined (19). In a first step, a linear normalization ANOVA model accounts for experiment-wide systematic effects (for instance, array effects) that could bias inferences made on the data from the individual genes. The residuals from this model represent normalized values and serve as input data for the gene models. In a second step, the gene models were fit separately to the normalized data from each gene, allowing inferences to be made by using separate estimates of variability. Here, estimates of primary interest are the time effects. Finally, as a measure of variability in expression levels among the time effects, Wald statistics were calculated and significance was assigned to this effect for each gene (for a detailed description of the statistical analysis, see www.psb. ugent.be兾papers兾lateralroot). For clustering analysis, two clustering algorithms were used, namely the hierarchical average linkage clustering (20) and the adaptive quality-based clustering (21). The hierarchical clusters were visualized with the TREEVIEW program (http:兾兾rana. lbl.gov兾EisenSoftware.htm), and the adaptive quality-based clustering program was used according to the instructions (www. esat.kuleuven.ac.be兾⬃thijs兾work兾clustering.html). Several settings for the minimum number of genes per cluster and for the probability of genes belonging to the cluster were tested to allow clustering of most of the significantly differentially expressed genes. Furthermore, statistically significant over- or underrepresentation of a particular funcat of the genes in each cluster was analyzed based on the categorization by MIPS and PEDANT (22). The thresholds used for the current calculation of the automatic funcat in PEDANT are a maximum of 10 hits and a maximum e-value of 0.001. We assigned a number of funcats to each of the 906 significantly differentially expressed genes, based on its best-hit homologies to genes for which a number of funcats was known by MIPS. The significance of the categories was estimated by a ␹2 goodness of fit test at the 5% level of significance (for details, see www.psb.ugent.be兾papers兾 lateralroot). For transmission electron microscopy, samples were collected 500 ␮m above the root apical meristems at 0 h and 6 h, treated according to the LRIS, and fixed with a mixture of 2% glutaraldehyde兾1.5% paraformaldehyde in 0.1 M sodium cacodylate buffer (pH 7.5) for 6 h at room temperature. After fixation, samples were washed four times at room temperature, rinsed, dehydrated in an ethanol-propylene oxide series, infiltrated with increasing concentrations of araldite-epon resin over a period of 40 h, and embedded in fresh resin. Ultra-thin sections (80–90

we are particularly interested in the signaling cascades that connect auxin with cell division during lateral root initiation, we chose this system to monitor the early transcriptional changes induced by auxin treatment, namely after 0 h, 2 h, 4 h, and 6 h of treatment with 10 ␮M NAA. Microarray Data. Changes in gene expression induced by the successive NPA and NAA treatments were monitored by microarray hybridizations, in which independent biological samples were labeled with the two fluorescence dyes and directly compared with a reference sample. Only a small fraction of the genes varied in expression across the two biological replications; consequently, the reproducibility of the experiment was high. The significances of the time and replicate effects, expressed as the negative 10-base logarithm of the P values assigned to each effect based on the Wald statistics, are plotted against each other (Fig. 2A). The accumulation of the dots on the left side of the plot, illustrating the negative correlation between the significance of both effects, indicates a high reproducibility of the replicated time courses. The number of genes significantly differentially expressed over the time course in this experiment was 906. The strongest effect on gene expression in the LRIS was an 18-fold change between two time points. Altogether, 123 genes had a differential expression ⬎4-fold, and 553 had a differential expression ⬎2-fold. As shown by Jin et al. (24), albeit for a small number of genes, the two dyes behaved differently at the gene level beyond that already accounted for at an overall level by the normalization model. However, the effect of changing dyes was ⬍1.7-fold, whereas the highest replication effect was a 2.1-fold change (see also www.psb.ugent.be兾papers兾lateralroot). Comparison of the identical expression profiles of the genes for which two independent cDNA clones had been spotted on the array confirmed the reproducibility of the hybridizations (Fig. 2 B–G). Furthermore, verification of the expression profiles of a set of genes by RT-PCR revealed profiles that were very consistent with the results obtained by the microarray hybridization (Fig. 2H). By combining the statistical analysis and control experiments performed, we can conclude that the selected genes from the microarray data reliably represent differential gene expression in the LRIS. Cluster Analysis. The expression profiles of all 906 significantly (P ⬍ 0.005) differentially expressed genes were visualized by using the hierarchical clustering algorithm (20) (Fig. 3A). Three main profiles could be identified from the clusters: up-regulated profiles with different timing characteristics (Fig. 3A, a–c), transiently induced profiles (Fig. 3A, d), and down-regulated profiles (Fig. 3A, e and f). The adaptive quality-based clustering program allowed us to cluster 869 of the 906 significantly differentially expressed genes by setting the minimum number of genes per cluster to 5 and a probability of 0.7. From the total of 11 clusters obtained, six had enough genes to allow statistical analysis of their content. Expression profiles of the six clusters corresponded to the up-regulated, transiently induced, and down-regulated expression profiles, similar to those obtained with the Eisen algorithm, but revealing more specific patterns of expression according to the time points of LRIS (Fig. 3B) (for the gene content of these clusters, see www.psb.ugent.be兾 papers兾lateralroot).

Fig. 2. (A) Blot of ⫺Log10 of the significance of the replicate effect versus the ⫺Log10 of the significance of the time effect, illustrating the negative correlation between the significance of both effects. (B–G) Validation of reproducibility of the hybridizations by comparing the hybridization results of six genes for which two independent cDNA clones had been spotted on the array. (B) Unknown protein (GenBank accession no. Atg49740). (C) Flavonol sulfotransferase (At1g74090). (D) Putative protein (At3g5700). (E) Sedoheptulosebisphosphatase precursor (At3g55800). (F) Unknown protein (At3g07390). (H) ATP-sulfurylase precursor (At5g4370). (G) Comparison of the expression patterns of five cell cycle genes in LRIS by RT-PCR and microarray. Expression profiles of G1兾S cell cycle marker genes Arath;E2Fa, histone H4, and Arath; KRP2, and the G2兾M-specific genes Arath;CDKB1;1 and Arath;CYCB2;1.

Functional Categorization of the Clustered Genes. The major biological activities represented by each cluster may be indicated by the over- or underrepresentation of funcats within the clusters (Table 1). In cluster 1, representing genes induced 2 h after auxin treatment, cellular communication and cellular environment categories were slightly overrepresented, indicating activation of signaling events. A gene implicated in auxin signaling that codes

for the heterotrimeric G protein ␣ subunit (GPA1) was strongly induced. This finding is in accordance with the hypothesis of its direct responsiveness to auxin and with its suggested role as an initiator of signaling cascades toward lateral root formation (25, 26). Furthermore, auxin response genes IAA2 and IAA11 were induced as a sign of activation of auxin response. Interestingly,

5148 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0308702101

Himanen et al.

Fig. 3. (A) Hierarchical average linkage clustering. Clusters a, b, c, d, e, and f were induced at 2 h, 4 h, 6 h, transiently, down-regulated at 4 h, and at 2 h, respectively. (B) The six major clusters obtained by adaptive quality-based algorithm, representing up-regulated (1–3), transient (4), and down-regulated (5, 6) clusters.

In cluster 2, representing genes induced 4 h after auxin treatment, several funcats were overrepresented, such as cell cycle, protein synthesis, and cellular fate, demonstrating that the cellular activity in the LRIS is early induced. Funcats related to metabolism and defense were underrepresented. At the 4-h time point, genes encoding ribosomal proteins, translation initiation factors, and regulators of ribosomal biogenesis constituted the largest funcat of the induced genes, ‘‘protein synthesis.’’ Some of the ribosomal proteins have been shown to be directly regulated by developmental signals, such as auxin (34, 35). Unlike transcriptional regulation during cell cycle, regulation of cell cyclespecific translation is poorly described (36). With our data, a large number of genes coregulated with ribosomal biogenesis factors are provided that allow the characterization of their role in cell division during lateral root development. Auxin may be involved in the cell cycle activation at G1兾S transition (37). Accordingly, G1兾S-specific genes, such as E2Fa and histone H4, were induced together with the DNA replication

Table 1. Overview of significantly over- and underrepresented functional categories in each cluster Profile

Cluster

Overrepresentation

Underrepresentation

Immediate up Intermediate up

1 2

Delayed up Transient

3 4

Communication and interaction with environment Cell cycle, cell fate, protein activity, protein with binding activity, and protein synthesis Protein synthesis Metabolism and transport

Immediate down

5

Metabolism

Delayed down

6

Cell differentiation, organ differentiation, and tissue differentiation

Protein fate Cellular environment, communication, energy, defense, and metabolism Energy Cell fate, development, protein with binding activity, protein fate, protein synthesis, and tissue differentiation Cell cycle, cell fate, protein with binding activity, and protein synthesis Protein synthesis

Himanen et al.

PNAS 兩 April 6, 2004 兩 vol. 101 兩 no. 14 兩 5149

PLANT BIOLOGY

two other Aux兾IAA genes, IAA7 and IAA8, did not respond in the LRIS. These differential responses indicate that the IAA proteins mediate different auxin signaling, perhaps in a tissuespecific or concentration-dependent manner (27, 28). In addition, and in concert to its suggested role as suppressor of lateral root development (29), the IAA28 gene was down-regulated by the auxin treatment (cluster 5). Furthermore, transcripts of two putative auxin influx carriers, LAX3 and amino acid permease 6 (30, 31), were induced, suggesting that active polar auxin transport was promoted in the system, which would further imply that auxin is capable of promoting its own uptake and兾or polar transport by the root cells. Interestingly, the putative auxin receptor, auxin-binding protein1 (ABP1), did not respond in the LRIS, although it was present on the microarray. ABP1 has been suggested to mediate high-affinity perception of auxin signal at low concentrations toward cell expansion, whereas high auxin concentrations, as is the case in our experiment, promote cell division (13, 32, 33).

genes DNA gyrase subunit A and a mismatch-binding protein at the 4-h time point. In cluster 3, representing genes induced 6 h after auxin, protein synthesis appeared as a major significantly overrepresented funcat, whereas energy metabolism was underrepresented. In addition to the genes related to protein translation, an interesting change could be observed among the genes from the funcat ‘‘cell cycle’’. Instead of G1兾S-specific genes, genes specific for the G2兾M cell cycle phase were induced, such as the cyclin-dependent kinase and cyclin genes Arath;CDKB1;1, Arath;CYCB1;1, and Arath;CYCB2;1. The xylem pole pericycle cells have been hypothesized to be arrested at G2 phase when they are susceptible for the auxin-mediated lateral root initiation (1, 37, 38). Further evidence for the importance of regulation of G2兾M transition during lateral root initiation has been provided by plants expressing the cdc25 gene of fission yeast (Schizosaccharomyces pombe), a CDK-activating phosphatase, which showed enhanced lateral root phenotypes (40). Although no cdc25 gene has been identified yet from the Arabidopsis genome, a similar phosphatase is likely to activate CDKAs at the G2兾M boundary (41). In addition, among the late-induced genes, c-myc and Myb transcription factors, which are implicated in G2兾M phase-specific cell cycle regulation, were induced (42, 43). In cluster 4, representing transient expression profiles, metabolism and regulation of transcription were overrepresented. In the immediate down-regulated cluster (cluster 5), funcats of cell cycle, protein synthesis, and cell fate were underrepresented, in accordance to their overrepresentation in induced clusters. However, two negative regulators of cell cycle activity, namely Arath;KRP2 (15, 44) and Arath;CKS1 (45), were strongly expressed in NPA-treated samples, indicating that they could be involved in mediating a G1 cell cycle block. In the delayed down-regulated cluster (cluster 6), three funcats related to differentiation were overrepresented. For the complete data set of the funcat analysis, see www.psb.ugent.be兾 papers兾lateralroot.

We thank Paul Van Hummelen for the microarray hybridizations, Annie Sauvanet for electron microscopy, and Pierre Hilson for critical reading of the manuscript. This work was supported by a grant from the Interuniversity Poles of Attraction Programme-Belgian Science Policy (P5兾13). K.H. is grateful to the Finnish Cultural Foundation and the Academy of Finland for fellowships. S. Vanneste and S. Vercruysse are indebted to the Flemish ‘‘Instituut voor de aanmoediging van Innovatie door Wetenschap en Technologie in Vlaanderen’’ for a predoctoral fellowship.

1. Bhalerao, R., Eklo ¨f, J., Ljung, K., Marchant, A., Bennett, M. & Sandberg, G. (2002) Plant J. 29, 325–332. 2. Malamy, J. E. & Ryan, K. S. (2001) Plant Physiol. 127, 899–909. 3. Casimiro, I., Beeckman, T., Graham, N., Bhalerao, R., Zhang, H., Casero, P., Sandberg, G. & Bennett, M. J. (2003) Trends Plant Sci. 8, 165–171. 4. Lincoln, C., Britton, J. H. & Estelle, M. (1990) Plant Cell 2, 1071–1080. 5. Fukaki, H., Tameda, S., Masuda, H. & Tasaka, M. (2002) Plant J. 29, 153–168. 6. Estelle, M. (2001) Curr. Opin. Plant Biol. 4, 254–260. 7. Xie, Q., Frugis, G., Colgan, D. & Chua, N.-H. (2000) Genes Dev. 14, 3024–3036. 8. Xie, Q., Guo, H.-S., Dallman, G., Fang, S., Weissman, A. M. & Chua, N.-H. (2002) Nature 419, 167–170. 9. Boerjan, W., Cervera, M.-T., Delarue, M., Beeckman, T., Dewitte, W., Bellini, C., Caboche, M., Van Onckelen, H., Van Montagu, M. & Inze´, D. (1995) Plant Cell 7, 1405–1419. 10. Sabatini, S., Beis, D., Wolkenfelt, H., Murfett, J., Guilfoyle, T., Malamy, J., Benfey, P., Leyser, O., Bechtold, N., Weisbeek, P. & Scheres, B. (1999) Cell 99, 463–472. 11. Marchant, A., Bhalerao, R., Casimiro, I., Eklo ¨f, J., Casero, P. J., Bennett, M. & Sandberg, G. (2002) Plant Cell 14, 589–597. 12. Celenza, J. L., Jr., Grisafi, P. L. & Fink, G. R. (1995) Genes Dev. 9, 2131–2142. 13. Chen, J.-G. (2001) J. Plant Growth Regul. 20, 255–264. 14. Scheres, B., Benfey, P. & Dolan, L. (September 30, 2002) In The Arabidopsis Book, eds. Somerville, C. R. & Meyerowitz, E. M. (American Society of Plant Biologists, Rockville) (doi兾10.1199兾tab.0101; www.aspb.org兾publications兾 arabidopsis) 15. Himanen, K., Boucheron, E., Vanneste, S., de Almeida Engler, J., Inze´, D. & Beeckman, T. (2002) Plant Cell 14, 2339–2351. 16. Ferreira, P. C. G., Hemerly, A. S., de Almeida Engler, J., Van Montagu, M., Engler, G. & Inze´, D. (1994) Plant Cell 6, 1763–1774. 17. Puska´s, L. G., Zvara, A., Hackler, L., Jr. & Van Hummelen, P. (2002) Biotechniques 32, 1330–1341. 18. Altschul, S. F., Gish, W., Miller, W., Myers, E. W. & Lipman, D. J. (1990) J. Mol. Biol. 215, 403–410.

19. Wolfinger, R. D., Gibson, G., Wolfinger, E. D., Bennett, L., Hamadeh, H., Bushel, P., Afshari, C. & Paules, R. S. (2001) J. Comput. Biol. 8, 625– 637. 20. Eisen, M. B., Spellman, P. T., Brown, P. O. & Botstein, D. (1998) Proc. Natl. Acad. Sci. USA 95, 14863–14868. 21. De Smet, F., Mathys, J., Marchal, K., Thijs, G., De Moor, B. & Moreau, Y. (2002) Bioinformatics 18, 735–746. 22. Frishman, D., Mokrejs, M., Kosykh, D., Kastenmu ¨ller, G., Kolesov, G., Zubrzycki, I., Gruber, C., Geier, B., Kaps, A., Albermann, K., et al. (2003) Nucleic Acids Res. 31, 207–211. 23. Reynolds, E. S. (1963) J. Cell Biol. 17, 208–212. 24. Jin, W., Riley, R. M., Wolfinger, R. D., White, K. P., Passador-Gurgel, G. & Gibson, G. (2001) Nat. Genet. 29, 389–395. 25. Ullah, H., Chen, J.-G., Young, J. C., Im, K.-H., Sussman, M. R. & Jones, A. M. (2001) Science 292, 2066–2069. 26. Ullah, H., Chen, J.-G., Temple, B., Boyes, D. C., Alonso, J. M., Davis, K. R., Ecker, J. R. & Jones, A. M. (2003) Plant Cell 15, 393–409. 27. Abel, S., Nguyen, M. D. & Theologis, A. (1995) J. Mol. Biol. 251, 533– 549. 28. Che, P., Gingerich, D. J., Lall, S. & Howell, S. H. (2002) Plant Cell 14, 2771–2785. 29. Rogg, L. E., Lasswell, J. & Bartel, B. (2001) Plant Cell 13, 465–480. 30. Bennett, M. J., Marchant, A., Green, H. G., May, S. T., Ward, S. P., Millner, P. A., Walker, A. R., Schulz, B. & Feldmann, K. A. (1996) Science 273, 948 –950. 31. Parry, G., Marchant, A., May, S., Swarup, R., Swarup, K., James, N., Graham, N., Allen, T., Martucci, T., Yemm, A., et al. (2001) J. Plant Growth Regul. 20, 217–225. 32. Chen, J.-G., Ullah, H., Young, J. C., Sussman, M. R. & Jones, A. M. (2001) Genes Dev. 15, 902–911. 33. Chen, J.-G., Shimomura, S., Sitbon, F., Sandberg, G. & Jones, A. M. (2001) Plant J. 28, 607–617. 34. Pe´rez, L., Aguilar, R., Pe´rez Me´ndez, A. & Sa´nchez de Jime´nez, E. (1990) Plant Physiol. 94, 1270–1275.

5150 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0308702101

Conclusions and Perspectives Characterization of genes that are activated during the early phases of lateral root development is an initial step toward understanding this developmental process. To our knowledge, this is the first report that comprehensively shows the transcriptional changes during the onset of root branching. We highlight the link between the auxin signal and the cell cycle reactivation. In summary, from the analysis of these microarray data, distinct developmental stages could be recognized during the time course. The first stage represents a cell cycle block as indicated by the low initial expression of several cell cycle genes (Arath;E2Fa, histone H4, B-type cyclins, and Arath;CDKB1;1) and their induction on auxin application. This cell cycle block has previously been shown to be a G1 arrest (15) and is also confirmed by the high expression level of Arath;KRP2 before auxin treatment. The next stage is characterized by auxin perception and signal transduction (cluster 1). Hereafter, the cell cycle machinery becomes activated, resulting in progression over G1兾S (cluster 2) coinciding with a down-regulation of genes involved in differentiation (cluster 6). Finally, pericycle cells start to divide, as revealed by induction of G2兾M-specific genes (cluster 3) and the gain of meristematic identity of the xylem pole pericycle cells (transmission electron microscopy analyses). Last but not least, the microarray data set contained a large number of unknown genes, and functional analysis of these genes will provide deeper insight into the regulation and processes involved in lateral root initiation.

Himanen et al.

41. Zhang, H. & Forde, B. G. (1998) Science 279, 407–409. 42. Ito, M., Araki, S., Matsunaga, S., Itoh, T., Nishihama, R., Machida, Y., Doonan, J. H. & Watanabe, A. (2001) Plant Cell 13, 1891–1905. 43. Planchais, S., Glab, N., Inze´, D. & Bergounioux, C. (2000) FEBS Lett. 476, 78 – 83. 44. De Veylder, L., Beeckman, T., Beemster, G. T. S., Krols, L., Terras, F., Landrieu, I., Van Der Schueren, E., Maes, S., Naudts, M. & Inze´, D. (2001) Plant Cell 13, 1653–1667. 45. De Veylder, L., Beemster, G. T. S., Beeckman, T. & Inze´, D. (2001) Plant J. 25, 617–626.

PLANT BIOLOGY

35. Dai, Z., Gao, J., An, K., Lee, J. M., Edwards, G. E. & An, G. (1996) Plant Mol. Biol. 32, 1055–1065. 36. Pyronnet, S. & Sonenberg, N. (2001) Curr. Opin. Genet. Dev. 11, 13–18. 37. Chriqui, D. (1985) Bull. Soc. Bot. Fr. 132, 127–141. 38. Beeckman, T., Burssens, S. & Inze´, D. (2001) J. Exp. Bot. 52, 403–411. 39. Casimiro, I., Marchant, A., Bhalerao, R. P., Beeckman, T., Dhooge, S., Swarup, R., Graham, N., Inze´, D., Sandberg, G., Casero, P. J. & Bennett, M. (2001) Plant Cell 13, 843–852. 40. Bell, M. H., Halford, N. G., Ormrod, J. C. & Francis, D. (1993) Plant Mol. Biol. 23, 445–451.

Himanen et al.

PNAS 兩 April 6, 2004 兩 vol. 101 兩 no. 14 兩 5151