none 10.7554/eLife.01567 eLife Sciences Publications, Ltd eLife Sciences Publications, Ltd. 4374 67124617 189365 20180823133646 10.7554 2018-08-23T13:41:49Z 2014-02-11T16:29:04Z 13 eLife 2050-084X 02 11 2014 3 Martial Sankar Department of Plant Molecular Biology, University of Lausanne, Lausanne, Switzerland Kaisa Nieminen Department of Plant Molecular Biology, University of Lausanne, Lausanne, Switzerland Laura Ragni Department of Plant Molecular Biology, University of Lausanne, Lausanne, Switzerland Ioannis Xenarios Vital-IT, Swiss Institute of Bioinformatics, Lausanne, Switzerland Christian S Hardtke Department of Plant Molecular Biology, University of Lausanne, Lausanne, Switzerland Among various advantages, their small size makes model organisms preferred subjects of investigation. Yet, even in model systems detailed analysis of numerous developmental processes at cellular level is severely hampered by their scale. For instance, secondary growth of Arabidopsis hypocotyls creates a radial pattern of highly specialized tissues that comprises several thousand cells starting from a few dozen. This dynamic process is difficult to follow because of its scale and because it can only be investigated invasively, precluding comprehensive understanding of the cell proliferation, differentiation, and patterning events involved. To overcome such limitation, we established an automated quantitative histology approach. We acquired hypocotyl cross-sections from tiled high-resolution images and extracted their information content using custom high-throughput image processing and segmentation. Coupled with automated cell type recognition through machine learning, we could establish a cellular resolution atlas that reveals vascular morphodynamics during secondary growth, for example equidistant phloem pole formation. Our understanding of the living world has been advanced greatly by studies of ‘model organisms’, such as mice, zebrafish, and fruit flies. Studying these creatures has been crucial to uncovering the genes that control how our bodies develop and grow, and also to discover the genetic basis of diseases such as cancer. Thale cress—or Arabidopsis thaliana to give its formal name—is the model organism of choice for many plant biologists. This tiny weed has been widely studied because it can complete its lifecycle, from seed to seed, in about 6 weeks, and because its relatively small genome simplifies the search for genes that control specific traits. However, as with other much-studied model systems, understanding the changes that underpin the development of some of the more complex tissues in Arabidopsis has been severely hampered by the shear number of cells involved. After it has emerged from the seed, the plant’s first stem will develop from a few dozen cells in width to several thousand cells with highly specialized tissues arranged in a complex pattern of concentric circles. Although this stem thickening process represents a major developmental change in many plants—from Arabidopsis to oak trees—it has been under-researched. This is partly because it involves so many different cells, and also because it can only be observed in thin sections cut out of the plant’s stem. Now Sankar, Nieminen, Ragni et al. have developed a novel approach, termed ‘automated quantitative histology’, to overcome these problems. This strategy involves ‘teaching’ a computer to automatically recognize different plant cells and to measure their important features in high-resolution images of tissue sections. The resulting ‘map’ of the developing stem—which required over 800 hr of computing time to complete—reveals the changes to cells and tissues as they develop that allow the transport of water, sugars and nutrients between the above- and below-ground organs. Sankar, Nieminen, Ragni et al. suggest that their novel approach could, in the future, also be applied to study the development of other tissues and organisms, including animals. 02 11 2014 e01567 10.7554/eLife.01567 SystemsX EMBO longterm post-doctoral fellowships Marie Heim-Voegtlin University of Lausanne 501100006390 http://creativecommons.org/licenses/by/3.0/ http://creativecommons.org/licenses/by/3.0/ http://creativecommons.org/licenses/by/3.0/ 1 eLifesciences www.elifesciences.org false 2013-09-20 2013-12-24 2014-02-11 SystemsX EMBO http://dx.doi.org/10.13039/501100003043 Swiss National Science Foundation http://dx.doi.org/10.13039/501100001711 University of Lausanne http://dx.doi.org/10.13039/501100006390 http://creativecommons.org/licenses/by/3.0/ http://creativecommons.org/licenses/by/3.0/ http://creativecommons.org/licenses/by/3.0/ Data from: Automated quantitative histology reveals vascular morphodynamics during Arabidopsis hypocotyl secondary growth 10.5061/dryad.b835k 10.7554/eLife.01567 https://elifesciences.org/articles/01567 https://cdn.elifesciences.org/articles/01567/elife-01567-v1.pdf https://cdn.elifesciences.org/articles/01567/elife-01567-v1.xml Nature Bonke 426 181 2003 APL regulates vascular tissue identity in Arabidopsis 10.1038/nature02100 Genetics Brenner 182 413 2009 In the beginning was the worm 10.1534/genetics.109.104976 Physiologia Plantarum Chaffey 114 594 2002 Secondary xylem development in Arabidopsis: a model for wood formation 10.1034/j.1399-3054.2002.1140413.x Neural computation Chang 13 2119 2001 Training nu-support vector classifiers: theory and algorithms 10.1162/089976601750399335 Machine Learning Cortes 20 273 1995 10.1007/BF00994018 Support-vector Networks Development Dolan 119 71 1993 Cellular organisation of the Arabidopsis thaliana root Seminars in Cell & Developmental Biology Elo 20 1097 2009 Stem cell function during plant vascular development 10.1016/j.semcdb.2009.09.009 Development Etchells 140 2224 2013 WOX4 and WOX14 act downstream of the PXY receptor kinase to regulate plant vascular proliferation independently of any role in vascular organisation 10.1242/dev.091314 PLOS Genetics Etchells 8 e1002997 2012 Plant vascular cell division is maintained by an interaction between PXY and ethylene signalling 10.1371/journal.pgen.1002997 Molecular Systems Biology Fuchs 6 370 2010 Clustering phenotype populations by genome-wide RNAi and multiparametric imaging 10.1038/msb.2010.25 Bio Systems Granqvist 110 60 2012 BaSAR-A tool in R for frequency detection 10.1016/j.biosystems.2012.07.004 Current Opinion in Plant Biology Groover 9 55 2006 Developmental mechanisms regulating secondary growth in woody plants 10.1016/j.pbi.2005.11.013 Plant Cell Hirakawa 22 2618 2010 TDIF peptide signaling regulates vascular stem cell proliferation via the WOX4 homeobox gene in Arabidopsis 10.1105/tpc.110.076083 Proceedings of the National Academy of Sciences of the United States of America Hirakawa 105 15208 2008 Non-cell-autonomous control of vascular stem cell fate by a CLE peptide/receptor system 10.1073/pnas.0808444105 Cell Meyerowitz 56 263 1989 Arabidopsis, a useful weed 10.1016/0092-8674(89)90900-8 Science Meyerowitz 295 1482 2002 Plants compared to animals: the broadest comparative study of development 10.1126/science.1066609 Plant Physiol Nieminen 135 653 2004 A weed for wood? Arabidopsis as a genetic model for xylem development 10.1104/pp.104.040212 Nature Biotechnology Noble 24 1565 2006 What is a support vector machine? 10.1038/nbt1206-1565 Proceedings of the National Academy of Sciences of the United States of America Olson 77 1516 1980 Classification of cultured mammalian cells by shape analysis and pattern recognition 10.1073/pnas.77.3.1516 Bioinformatics Pau 26 979 2010 EBImage–an R package for image processing with applications to cellular phenotypes 10.1093/bioinformatics/btq046 Plant Cell Ragni 23 1322 2011 Mobile gibberellin directly stimulates Arabidopsis hypocotyl xylem expansion 10.1105/tpc.111.084020 Dryad Digital Repository Sankar 2014 Data from: Automated quantitative histology reveals vascular morphodynamics during Arabidopsis hypocotyl secondary growth 10.5061/dryad.b835k Current Biology Sibout 18 458 2008 Flowering as a condition for xylem expansion in Arabidopsis hypocotyl and root 10.1016/j.cub.2008.02.070 The New Phytologist Spicer 186 577 2010 Evolution of development of vascular cambia and secondary growth 10.1111/j.1469-8137.2010.03236.x Machine Vision and Applications Theriault 23 659 2012 Cell morphology classification and clutter mitigation in phase-contrast microscopy images using machine learning 10.1007/s00138-011-0345-9 Cell Uyttewaal 149 439 2012 Mechanical stress acts via katanin to amplify differences in growth rate between adjacent cells in Arabidopsis 10.1016/j.cell.2012.02.048 Nature Cell Biology Yin 15 860 2013 A screen for morphological complexity identifies regulators of switch-like transitions between discrete cell shapes 10.1038/ncb2764 10.7554/eLife.01567.001 https://elifesciences.org/articles/01567#abstract 10.7554/eLife.01567.002 https://elifesciences.org/articles/01567#digest (A) Light microscopy of cross sections obtained from Arabidopsis hypocotyls (organ position illustrated for a 9-day-old seedling, lower left) at 9 dag (upper left) and 35 dag (right). Size bars are 100 μm. Blue GUS staining due to the presence of an APL::GUS reporter gene in this Col-0 background line marks phloem bundles. (B) Overview of the developmental series (time points and distinct samples per genotype) analyzed in this study. (C) Example of a high-resolution hypocotyl section image assembled from 11 × 11 tiles. (D) The same image after pre-processing and binarization, and (E) subsequent segmentation using a watershed algorithm. (F) Number of mis-segmented cells as determined by careful visual inspection in 12 sections, plotted against the total number of cells per section (log scale). 10.7554/eLife.01567.003 https://elifesciences.org/articles/01567#fig1 (A) Overview of the computational pipeline from image acquisition to analysis. (B) ‘Phenoprints’ for the different genotypes and developmental stages. 10.7554/eLife.01567.004 https://elifesciences.org/articles/01567#fig2 The green arrow points out the selected feature combination according to the criteria of minimum number of features with the highest performance and the lowest variation (the radiusV feature was excluded due to its putative variation in tissue location). 10.7554/eLife.01567.005 https://elifesciences.org/articles/01567/figures#fig2s1 (A) Principal component analysis (PCA) of the phenoprints shown in Figure 2B, performed with normalized values (Supplementary file 4). The inlay screeplot displays the proportion of total variation explained by each principal component. (B–E) Comparative plots of parameter progression in the two genotypes. In (D), xylem represents combined vessel, parenchyma, and fiber cells, phloem represents combined phloem parenchyma and bundle cells. Error bars indicate standard error. 10.7554/eLife.01567.006 https://elifesciences.org/articles/01567#fig3 (A and B) Spatial distribution of cell incline angle illustrates the vascular organization in Ler (B) as compared to Col-0 (A) at later stages of development, for example 30 dag. The size of the disc increases with the area of the cell. Blue color indicates radial cell orientation, red orthoradial. (C and D) Violin plots of incline angle distribution, illustrating increasingly bimodal distribution coincident with refined vascular organization and different dynamics of the process in the two genotypes. 10.7554/eLife.01567.007 https://elifesciences.org/articles/01567#fig4 The incline is the angle between the section radius through the center of an ellipse fit to a cell and the major axis of that ellipse extended towards the x axis. 10.7554/eLife.01567.008 https://elifesciences.org/articles/01567/figures#fig4s1 (A–J) Density plots of cell incline angle vs radial position for the two genotypes at the indicated developmental stages, representing all cells across all sections for a given time point. The red lines represent the fit of these cloud distributions with locally weighted linear regression (i.e., lowess), revealing the essential data trends. All sections were normalized from 0.0 (the manually defined center) to 1.0 (the average radius in a set of sections as determined by the average distance of the outermost cells from the center for individual sections). Box plots indicate the quartiles of the radian distribution for each cell-type class and are placed at the average position of the cell type with respect to the y axis. Outliers are shown as circles. 10.7554/eLife.01567.009 https://elifesciences.org/articles/01567#fig5 Cell number in a circle of 200–500 pixels around the section centers for Col-0. Cell count in a constant area of xylem over time across all averaged across all sections. 10.7554/eLife.01567.010 https://elifesciences.org/articles/01567/figures#fig5s1 (A) Example of Gaussian kernel density estimate of the location of predicted phloem bundles cells in a 30 dag Col-0 section. High density represents phloem poles. (B) Example of an analysis of emerging phloem pole position in a 30 dag Col-0 section. The plot represents a pixel intensity map after noise reduction along a circular region of interest across the emerging phloem poles. Intensity peaks are due to GUS staining conferred to phloem bundles by an APL::GUS reporter construct. (C) Probability density function of the data shown in (B) obtained from an automated Bayesian model. The dominant single peak indicates a constant arc distance of ca. 62 pixel between the phloem poles. 10.7554/eLife.01567.011 https://elifesciences.org/articles/01567#fig6 (A) An explanation of the extracted parameters that describe the cellular features. (B) Summary information of the hand-labeled training set for supervised machine learning. (C) Definition of the classifiers selected for analysis. (D) Summary of the classifier parameters for supervised machine learning. (E) Overview of the cell type classes recognized by the supervised machine learning approach and their assignment codes used in Data Files 3 and 4. 10.7554/eLife.01567.012 https://elifesciences.org/articles/01567/figures#SD1-data Quality control files for the Col-0 sections. 10.7554/eLife.01567.013 https://elifesciences.org/articles/01567/figures#SD2-data Quality control files for the Ler sections. 10.7554/eLife.01567.014 https://elifesciences.org/articles/01567/figures#SD3-data The normalized values of the phenoprints (Figure 2B) used for PCA. 10.7554/eLife.01567.015 https://elifesciences.org/articles/01567/figures#SD4-data 10.7554/eLife.01567.016 https://elifesciences.org/articles/01567#SA1 10.7554/eLife.01567.017 https://elifesciences.org/articles/01567#SA2