Developmentally equivalent tissue sampling ... - Wiley Online Library

4 downloads 345277 Views 1MB Size Report
(Creative Suite 1, Adobe). This script also increased the size of the images to provide additional precision in pinpointing intersection points of tag and plant ...
Research

Methods Developmentally equivalent tissue sampling based on growth kinematic profiling of Arabidopsis inflorescence stems Hardy Hall and Brian Ellis Department of Botany and the Michael Smith Laboratories, University of British Columbia, 2185 East Mall, Vancouver, BC, Canada, V6T 1Z4

Summary Author for correspondence: Brian Ellis Tel: +1 604 822 3451 Email: [email protected] Received: 18 October 2011 Accepted: 9 December 2011

New Phytologist (2012) 194: 287–296 doi: 10.1111/j.1469-8137.2012.04060.x

Key words: Arabidopsis thaliana, developmental equivalence, feature tracking, growth kinematics, primary stem development.

• Directional growth in Arabidopsis thaliana during bolting of the inflorescence stem makes this an attractive system for study of the underlying processes of tissue elongation and cell wall extension. Analysis of local molecular events accompanying Arabidopsis inflorescence stem elongation is hampered by difficulties in isolating developmentally matched tissue samples from different plants. • Here, we present a novel sampling approach in which specific developmental stages along the developing stem are defined nonintrusively in terms of their relative elemental growth rate by use of time-lapse imagery and subsequent derivation of growth kinematic profiles for individual plants. • Growth kinematic profiling reveals that key developmental transitions such as the point of maximum elongation rate and the point of cessation of elongation occur over broad and overlapping ranges across individuals within a population of the Columbia (Col-0) ecotype. The position of these transitions is only weakly correlated with overall plant height, which undermines the common assumption that physically similar plants have closely matched growth profiles. • This kinematic profiling approach provides high-resolution growth phenotyping of the developing stem and thereby enables the harvest, pooling and analysis of developmentally matched tissue samples from multiple Arabidopsis plants.

Introduction The Arabidopsis thaliana inflorescence stem undergoes a radical change in form and composition as the compressed internodes of newly formed primordial axillary stems elongate rapidly to their fully expanded state in the mature inflorescence. As in roots, this primary stem growth is accomplished through an iterative process of division, elongation, and maturation of successive cohorts of cells occurring within a developmental cassette at the plant apex. It has long been established that the anisotropic tissue expansion accompanying stem elongation requires a massive increase in the length of axial cell walls (reviewed in Erickson & Sax, 1956). Unlike roots, lateral branching of stems is established before elongation and, under normal long-day growing conditions, much of the aerial architecture of the primary stem, including transition from cauline leaf zone to true inflorescence, is already formed before primary stem bolting (Hempel & Feldman, 1994; Pouteau & Albertini, 2009). The dramatic bolting of the Arabidopsis inflorescence presents a simple and powerful system to study cell wall expansion phenomena. Furthermore, since  2012 The Authors New Phytologist  2012 New Phytologist Trust

extensive secondary cell wall formation and rigidification occur within a subset of cell types below the point of cessation of general cell wall expansion (Ye et al., 2002), this system also shows great potential for discovery of key processes and regulatory mechanisms associated with secondary cell wall formation (Ehlting et al., 2005; Zhong et al., 2008). Examination of developmental processes through an integrated systems biology approach is primarily challenged by accurate identification of tissues at distinct developmental stages (Boyes et al., 2001) and by the requirement for adequate amounts of those tissues for assays such as proteomics, metabolomics and transcriptomics (Pu & Brady, 2009). The challenge of pooling developmentally matched tissues across sufficient replicate plants is particularly acute in study of cell wall expansion processes, given the narrow spatiotemporal window at either root or shoot apices. A high degree of spatiotemporal resolution has been demonstrated in expression profiling studies of Arabidopsis root development that use fluorescence-based sorting of protoplasted Arabidopsis root cells from fluorescent marker-bearing lines (Birnbaum et al., 2003, 2005). However, uniform protoplasting New Phytologist (2012) 194: 287–296 287 www.newphytologist.com

New Phytologist

288 Research

of often-lignified primary stem tissues remains an intractable problem. It would be desirable to be able to physically isolate tissues of a known developmental state en masse from macroscopically discernible regions of the plant. Some studies of stem elongation have attempted to accomplish this by harvesting the base of the inflorescence stem at standardized principal growth stages as defined by flower status (Boyes et al., 2001; Minic et al., 2009) or inflorescence maturity in a subset of plants (Ko & Han, 2004). Others have compared different developmental stages cooccurring along the primary stem (Ehlting et al., 2005; Imoto et al., 2005), using microscopic examination of a subset of plants to establish guidelines for harvesting and pooling samples from a larger set of plants. Methods that group tissue samples selected on morphometric measures along the stem rely upon the assumption that plants that appear morphologically similar also share similar developmental proportioning. However, genetic and environmental variables, as well as the innately stochastic nature of developmental programming, make it unclear that such an assumption is valid. Inflorescence elongation is a dynamic process that lends itself to kinematic analysis (study of the motion of bodies) as a means of establishing how the process of cell elongation is distributed throughout the inflorescence stem. Manual measurement of the rates of separation of ink marks positioned 5 mm apart along the Arabidopsis inflorescence stem, relative to the base of the stem, did establish that no elongative growth occurred at distances > 7 cm below the apex of the stem in Columbia (Col-0) plants over a 24 h period, which made it possible to contrast pre- vs post-cessation states (Suh et al., 2005). These samples were presumed to broadly represent growth stages involving predominantly primary and secondary cell wall formation, respectively, but this approach of tracking ink mark movement cannot isolate regions of maximum tissue expansion rate. Kinematic analysis of root growth was first realized through the development of time-lapse imagery of corn roots marked by carbon particles (Erickson & Sax, 1956). Measurement of the rates of displacement of carbon particles from the root tip allowed derivation of the relative elemental growth rate (REGR) for points along the root, essentially providing a growth kinematic profile (GKP). More modern approaches have utilized root tip displacement rates, also referred to as the velocity field (Gandar, 1983), as the basis for direct calculation of REGR as a percentage change in length for elements (segments) over time (Schmundt et al., 1998; Basu et al., 2007). REGR data have also been generated indirectly for a number of species (not including Arabidopsis) through a modeling approach that uses a logistic function conditioned by estimated parameters in order to substitute for the missing velocity field data (Morris & Silk, 1992). This logistic function was used to model strain rates (cell wall expansion), essentially equivalent to REGR, from cell length data (destructive sampling) and overall root growth rate (Baskin et al., 1995). Unlike the apex-based growth analyses of roots, derivation of REGRs for inflorescence stems requires a much different approach, because the inflorescence stem apices are obscured by New Phytologist (2012) 194: 287–296 www.newphytologist.com

floral clusters, and lateral branching and flowers in the region of stem expansion prevent direct observation of all regions of the elongating primary stem. These impediments also preclude use of the recently developed ‘structure tensor’ methods for automated determination of velocity fields (optical flow) (Schmundt et al., 1998; Walter et al., 2002; Palaniappan et al., 2004; French et al., 2009), which depend upon a high density of natural features along a continuously visible, expanding surface. In Zea mays, manual measurement of internode lengths enabled REGR (mm d)1) values to be derived for individual, numbered internodes between 30 and 65 d postgermination (Collings et al., 1998) but the internode density of Arabidopsis is too low to allow detailed growth kinematic profiling. We reasoned that regularly spaced synthetic optical markers (plant tags) that extend beyond visual obstructions (branching) would provide a finer scale of measurement with which to compute localized REGR, largely irrespective of the degree of primary stem concealment. Physical tagging of plant stems would also permit post-observation isolation of specific, numbered segments that correspond to regions of interest along the cell wall expansion continuum. The finer spatial and temporal resolution provided by this strategy should, when applied to the Arabidopsis inflorescence stem, result in a powerful platform on which to build analyses of the developmental program of cell wall expansion in this organ. Here, we outline a rapid, noninvasive means of establishing fine-scale GKPs of individual elongating Arabidopsis primary stems. This system facilitates both growth-rate characterization and isolation of tissue from specific stages of elongative development. To validate our approach, we compared cell length between GKP-identified developmental states for each of two diffusely growing cell types. We also use the technique to examine the position of specific developmental transitions relative to the shoot apex, and to the overall height of the plant, as a test of the common assumption that individual plants of similar stature have similar GKPs.

Materials and Methods Plant material and growth conditions Cold-treated Arabidopsis thaliana (L.) Heynh Columbia (Col-0) seeds were sown in 32-plug tray inserts with soil-less potting mix (Sunshine Mix #5, Sun Gro Horticulture Canada Ltd, Seba Beach, Alberta, Canada) supplemented with liquid fertilizer 20N-20P-20K (Plant-prod soluble fertilizer, Plant Products Co. Ltd, Brampton, Ontario, Canada), then grown on short-day conditions (day 8 h, 21C : night 16 h, 19C) for 6 wk. To induce bolting, plants were transferred to long-day conditions (day 16 h, 21C : night 8 h, 21C) until the inflorescence reached a height of 10–15 cm. Plants were then removed from the growth chamber for image analysis and harvest. As detailed in Supporting Information, Video S1, 11 mm · 0.5 mm tags cut from 28 lb bright inkjet paper were glued to  2012 The Authors New Phytologist  2012 New Phytologist Trust

New Phytologist inflorescence stems with the application of a small drop of white glue to the center of the tag, bringing the tag into contact with the stem using forceps to allow the glue to bind tag to stem. Tags were placed c. 5 mm apart, a suitable separation for downstream analysis, and positioned so as to be perpendicular to the stem, for ease of tracking. While every attempt was taken to avoid fastening tags to developing siliques at the apex, this occurred c. 20% of the time for the uppermost tag. As an alternate stategy, two plants were tagged with 1 mm-diameter polyvinylidene fluoride (PVDF) disks, attached by latent electrostatic force at a similar density.

Research 289

image scale fluctuations resulting from digital camera autofocusing. A proxy for the stem base (not visible to the camera) was recorded and the offset distance to the stem was later measured after removal of the plant from the chamber (c. 2 cm). From the bottom upwards, left- then right-side tag–plant intersection points (Fig. 2, inset; Fig. 3, inset) were recorded throughout the image sequence. Finally, the position of the apex of the stem was recorded. Feature tracking data (image set ID, X coordinate, Y coordinate, image #) were exported in tab-delimited form for import into the R programming environment (http://www.Rproject.org).

Growth imaging and image processing Tagged plants were immediately loaded into single-plant subchambers ⁄ restraints (Fig. S4). For plants grown without nylon strings constraint, front and back plates were constructed of plexiglass, spaced 3 cm apart to allow the full range of lateral movement (circumnutation) of primary stems during elongation (growth profiles in Fig. S4, column 1). Otherwise, plants were placed within subchambers such that axillary branches and siliques protruded through the vertical nylon strings of the front plate. Nylon-strung back plates were placed behind the plants to constrain the primary stems, with rearward branches and siliques similarly protruding. To ensure uniform soil moisture during the imaging period, plants were watered before loading into the imaging chamber (Fig. S5). Once all plants were tagged and loaded, the growth chamber (Fig. S5) was closed and a time-lapse series (1 min interval) of all six plants was captured using a 10 megapixel compact digital camera (CoolPix p5000, Nikon) fitted with a polarizing lens (UR-E20, Nikon) to reduce Plexiglas reflectance. A neutral density gradient filter was added to the reflected light path in order to normalize image contrast from top to bottom. Exhaust fans ensured that ambient room temperature (21C) was maintained within the imaging chamber. Plants were subsequently removed at 30 min intervals, between image capture events, for harvesting and preservation. The overall harvesting period for the six plants was centered around 14:00 h, the midpoint of the day on a long-day cycle beginning at 06:00 h. Upon completion of the harvesting of segments, six-plant image sets (3648 · 2736 pixel) were subdivided into six separate, single-plant image sets using an action script in ImageReady (Creative Suite 1, Adobe). This script also increased the size of the images to provide additional precision in pinpointing intersection points of tag and plant within ImageJ. Final single-plant image dimensions were 1412 · 5472 pixel (JPEG), restricted by RAM memory limits within ImageJ (c. 300 MB for the time series).

Growth kinematic profiling All calculations and plotting of growth kinematic data were carried out in a script provided in its general form as Notes S1, and here outlined in the order in which these process are carried out in R. XY coordinates (pixels) were first normalized relative to reference marks before conversion to a centimeter scale based upon the 5 cm span between left and right reference marks. Plots such as Fig. 3 (excluding inset) were generated for each time point for quality assurance, while calculation of stem width, mean tag locations, and segment length were scripted to perform as described in the Results section. Segment lengths were treated with a locally weighted linear regression (LOWESS) smoothing function (‘loess’ function, base package, R) to dampen the effect of spurious measures on estimates of length in the final intervals before harvest. Relative (elemental) elongation rates were calculated using formula ‘1’ (results), then co-plotted with LOWESS regressions curves generated via the ‘loess’ function as in Fig. 4 (right-hand plots). Identification of developmental zones is addressed in the Results section. Segments corresponding to maximum growth rate and cessation stages were identified, and distances from the apex derived from the y-axis of the GKP plots. Histograms, scatterplots, Shapiro–Wilks tests for normality, Q–Q plotting, and Pearson correlations were carried out using R base package functions. Tissue harvesting for microscopic analysis Stem segments bounded by paper tags were harvested from individual plants in sequence from top to bottom over the course of c. 10 min (Video S3). Upon excision, segments were immersed in 150 ll fixation buffer (stock 2X PME; 50 mM 2-[4-(2-sulfoethyl)piperazin-1-yl]ethanesulfonic acid (PIPES), 2 mM MgSO4, 2 mM EGTA), within 0.2 ml dome-cap thermal cycler tubes. Segments were then subjected to three consecutive cycles of 5 min vacuum infiltration at 20 inches Hg and washed three times in 1X PME before long-term storage at 4C in 1X PME.

Time-lapse series analysis

Morphometric analysis

As detailed in Video S2, the single-plant JPEG image series was imported into ImageJ as an 8-bit RGB stack, and the ‘point tool’ used to record XY coordinates for single points of interest that could then be tracked through the series sequentially. Reference marks 5 cm apart were first recorded to allow for calibration of

Segments were trimmed to a uniform 3 mm length and longitudinally sectioned to bisect the interfascicular region along the length of the segment. To demarcate cell boundaries, each segment was stained with the xylan ⁄ cellulosic stain Congo Red (Cat#:60910, Fluka, Buchs, Switzerland) and imaged with a Zeiss Meta Inverted

 2012 The Authors New Phytologist  2012 New Phytologist Trust

New Phytologist (2012) 194: 287–296 www.newphytologist.com

290 Research

New Phytologist

confocal microscope. Z-stacks (225 lm · 225 lm · c. 35 lm) were collected through an LD C-Apochromatic ·40 ⁄ 1.1 W M27 objective (pinhole 88 lm) using a 488 nm laser, excluding intrinsic fluorescence with a BP 575-615 IR filter. Confocal Z-stacks were stitched together using the ‘3D stitching’ function of the Fiji version of ImageJ (http://pacific.mpi-cbg.de/) in order to provide a wider field of view for morphometric analysis. For each segment, 20 cortical and 10 endodermal cells from each segment were traced with the freehand line tool of ImageJ and entered as regions of interest (ROIs) into the ROI editor (Video S4). Cell length was extracted from ‘major axis’ metric (maximum dimension in microns) of the ROI summaries and used to compute mean cell lengths for cell types of a specific segment. These means were then used in statistical tests of consistency among cell lengths of the six replicate plants for each cell type, using base package statistical tools in R (Fig. 6).

Results Synthetic optical markers (tags) allow precise tracking of inflorescence elongation To record the progress of cohorts of cells relative to other cohorts, we fixed a series of synthetic optical markers (paper tags of fixed dimensions) along the stem surface at a density that allowed precise tracking of those epidermal cell cohorts (Fig. 1). While manual placement of either paper tags or polyvinylidene fluoride (PVDF) disks (Fig. S1) at discrete intervals requires care, their use offers distinct advantages over other marking systems. They allow for fine-scale characterization of primary stem segment REGRs, whereas natural features on the stem surface are typically not present in sufficiently high density, and do not present precise proxy to cohorts of cells. Similarly, the arbitrary placement of synthetic marks such as graphite particles on roots (Erickson & Sax, 1956; Mullen et al., 1998; Sugimoto et al., 2000) or silicone beads on trichomes (Schwab et al., 2003) substantially complicates the tracking and harvesting of the marked segments. Although the optical marker cannot always be positioned exactly horizontally, it is possible to routinely identify within the images the point at which the top edge of each paper tag intersects the left and right edge of the plant stem, and to use this information to precisely quantify the length of individual between-tag stem segments as they change over time (Fig. 2). Assuming that a tag is attached at a single point on the stem (tangential point of attachment), the location of the point of contact can be inferred by averaging the XY coordinates of the left and right tag–plant intersection points (Fig. 2, inset). Restriction of the primary stem to a two-dimensional plane allowed us to digitize the spatial arrangement of left and right tag–plant intersection points on an XY plane (Fig. 3), and to calculate the segment length as the Euclidean distance between the means of tag–plant intersection points for adjacent tags with a high degree of precision. We could also calculate the width of the stem at the tagged point, which we found not to measurably increase over the duration of the observation period. Further precision in establishing these intersection points was achieved by doubling the number of pixels in the image. New Phytologist (2012) 194: 287–296 www.newphytologist.com

Fig. 1 Images of an Arabidopsis thaliana plant within a growth chamber, just after tagging (start of the observation period) and then just before harvesting (end of the observation period). Primary stem growth was constrained within front and back plates of vertical nylon lines. The pair of images, selected from a 2 h 50 min time-series, are shown here to demonstrate a clear change in stem height. Normally, images were captured every min and final calculation of the growth kinematic profile was made on the final 10 min interval before harvest.

While the top of the plant was also tracked in this time-lapse imagery, flowers typically obscured the shoot apex. The flowers themselves did not provide a reliable indicator of apex movement because of floral organ movements obscuring the tip of the primary stem. The top segment (#1 in Fig. 3, inset) was therefore not incorporated into the kinematic analysis of primary stem elongation. Calculation of relative growth rates from tag movements To calculate the elongation rates of tag-defined segments over time, we chose an approach that closely parallels previous methods of deriving REGR from dimensional change in composite  2012 The Authors New Phytologist  2012 New Phytologist Trust

New Phytologist

Research 291

**

*

Before

After

Fig. 2 Illustration of segment length determination in tagged plants. Tag position is defined as the intersection point between the midline of the primary stem and the top edge of the tag (large arrows). Segment length change is computed as the change in the start (*) and end (**) distance between two tags over the time interval. Inset: calculation of tag position was formally computed as the mean Euclidean distance in XY space between the left and right tag–plant intersection points (arrowheads), where either the left or right edge of the primary stem was considered. An alternate tracking strategy for round disks is similarly portrayed in Supporting Information Fig. S1.

segments of roots over time (Sugimoto et al., 2000; Wenzel et al., 2000). Accounting for differences in absolute lengths of segments, and specifying the unit of time in h, we established the following formula: REGR ¼ ð½ðxn;f  xn;i Þ=xn;i =tint Þ  100

Eqn1

where n is the segment number; xf is the final length of segment (cm); xi is the initial length of segment (cm); and tint is the time interval (h). Relative elemental growth rate values, expressed as the rate of percentage change in length per h, were calculated for all segments spanned by tags using Eqn 1. These REGR values were then plotted as a function of distance from the base of the stem to generate GKPs that reflect the positioning of elongative growth along the bolting inflorescences through the course of observation (surface plots, Fig. 4). Variations in segment length over time resulting from measurement error were addressed with a smoothing function (linear regression) to provide predicted segment lengths for all time intervals (Fig. S2), and thus more robust estimation of the REGRs during the final interval before harvest. Since the REGR values generally fell within 65% confidence intervals for the best-fit LOWESS regression curve, we chose to use the LOWESS curves to provide a standardized approach to best identify specific stages of elongative development. In this manuscript, we present GKPs for all plants used for multi-plant analyses of growth stage transitions and cell length measurements (Figs 5, 6). We identified the points of maximal  2012 The Authors New Phytologist  2012 New Phytologist Trust

Fig. 3 XY coordinates of tag–plant intersection points derived from feature tracking dataset. Two marks for the left and right side of the stem enable calculation of stem width and mean tag placement on the plant. The XY coordinates have been normalized such that the base of the stem (*) is the origin, while the top of the plant is indicated as a triangle. Arrows labeled R1 and R2 depict calibration reference marks that remain a fixed distance apart (5 cm) throughout the time course to correct for the effect of digital focusing on scale. Inset (enlarged): dotted vertical lines bisect the span between the left and right stem–tag intersection points for the segments numbered from the top downwards.

growth rate and cessation in 24 individual Col-0 plants using either paper tag (Fig. S3; A1–E5) or PVDF disk (Fig. S3; A5, B5) optical markers. Initial time-lapse imagery of unrestrained plants indicated that circumnutation was pronounced in some individual plants (video not shown) and may contribute to apparent increases or decreases in the size of marker-spanned segments as a result of stem movement toward or away from the camera. The resulting inaccuracy in REGR determination may have led to increased noise in GKPs observed for plants grown under those conditions (Fig. S3; A1–F1). It is worth noting that accurate recording of dimensional change along the primary stem requires restriction of this expansion to a single dimension for high-resolution imagery, and precise tracking of the changes over short time intervals. Primarily for this reason, assessment of REGRs from two-dimensional (2D) observation of growing roots (single-point perspective) has largely been conducted on restrictive 2D surfaces such as glass plates (Erickson & Sax, 1956; Basu et al., 2007) or Petri dishes (Hummel et al., 2007; Mullen et al., 1998; Nagel et al., 2006). These data are recognized as being informative despite ongoing New Phytologist (2012) 194: 287–296 www.newphytologist.com

New Phytologist

292 Research

4 3 2 1

15 14 13

Sample quantiles

5

10

Frequency (number of plants)

6

(b)

12

Shapiro−Wilks Score, W= 0.9751 P−value= 0.8086 n= 23

7

11

(a)

0

10 11 12 13 14 15 16

−2

Plant height (cm)

2

2

15 14 13 10

0

r= 0.3301

12

4

15

1.5 2.0 2.5 3.0 3.5 4.0 4.5

6 4 2

(f)

r= 0.2403

10

0

15

Shapiro−Wilks Score, W= 0.8942 P−value= 0.0193 n= 23

8

Plant height (cm)

(e)

Max-growth distance from apex (cm)

14

10

13

5

12

0

11

Frequency (number of plants)

6

Max-growth distance from apex (cm)

Frequency (number of plants)

1

(d) Plant height (cm)

Shapiro−Wilks Score, W= 0.8508 P−value= 0.0028 n= 23

8

0

11

(c)

−1

Theoretical quantiles

0

5

10

15

4

Cessation distance from apex (cm)

(g)

5

6

7

Cessation distance from apex (cm)

(h)

4 2 0

2.0

Fig. 4 Three representative growth profiles (a–c) of Arabidopsis thaliana depicting growth rates (% change h–1) relative to plant position (segment number) for specific segments over time. Surface plots (left): growth rates (% change h–1, vertical axis) are plotted against the number of segments (defined by optical marker tags) from the apex downwards, over the duration of the imaging period in 10 min intervals. The dark-shaded, nearest profile denotes the last 10 min interval before harvest, depicted in the greater detail in the right-hand scatterplot. Scatterplots (right): growth rates (% change in length per h) are plotted against distance from the stem base for specific segments. Segments are numbered from the top of the plant downwards on the right-hand axis. The LOWESS regression curve follows the best fit through the growth rate data for this plant over a given 10 min interval. Dashed lines represented 65% confidence intervals for the LOWESS regression curve. Closed-box ⁄ arrow indicates the stem position that matches the maximum growth rate of the regression curve (segment 5), plotted as the rightmost vertical dotted line, while the open-box ⁄ arrow indicates the first position below the top of the stem where the growth rate falls to zero (segment 10).

New Phytologist (2012) 194: 287–296 www.newphytologist.com

15 14 13

r= −0.0565

12

6

11

8

Plant height (cm)

Shapiro−Wilks Score, W= 0.904 P−value= 0.0306 n= 23

10

Frequency (number of plants)

10

2.5

3.0

3.5

4.0

4.5

Max-growth to cessation distance (cm)

2.0

2.5

3.0

3.5

4.0

4.5

Max-growth to cessation distance (cm)

Fig. 5 Relationships between growth kinematic profile (GKP)-identified stem growth stages, the stem apex, and overall stem height in Arabidopsis thaliana. (a) Distribution of total heights (stem base to apex) for all 24 profiled plants. (b) Q–Q normality plots of plant height, showing correlation of sample quantiles to theoretical quantiles. (c) Distribution of distances between the plant apex and growth kinematic-identified maximum growth-rate points. (d) Scatterplot of distance from maximum growth rate points to apex vs plant heights with associated Pearson correlation coefficient (r). (e) Distribution of distances between the plant apex and GKPidentified cessation growth points. (f) Scatterplot of distance from cessation points to apex vs plant heights, with associated Pearson correlation coefficient (r). (g) Distribution of distances between maximum growth-rate and cessation points. (h) Scatterplot of distance from maximum growth rate points to cessation points vs plant heights, with associated Pearson correlation coefficient (r). Shapiro–Wilks ‘W’ scores, associated P-values for normality, and sample number (biological replication) are noted in (a), (c), (e) and (g).

 2012 The Authors New Phytologist  2012 New Phytologist Trust

New Phytologist debate of the effect of such confinement on actual growth rate (reviewed in Baskin et al., 1995). In our system, stem growth was essentially constrained to a single linear trajectory through use of fine vertical nylon guide lines that permit uninterrupted branching and floral development while constraining circumnutatory movement, without impeding vertical extension (Fig. S1). Maximum REGRs for GKPs of these plants (Fig. S3, columns 2–6) do not appear negatively impacted (3–4% change in length per h in the nylon-restrained environment compared with 2–3% change in length per h for the unrestrained condition). Such confinement appears unlikely to have an impact on expansive growth, in comparison to more invasive means of confining growth to two dimensions, such as were employed in a study of leaf expansion where tensile force was being exerted by leaf tethers (Hsiao et al., 1970; Schmundt et al., 1998). The latter process has documented effects on normal tensile forces involved in leaf expansion (Walter et al., 2002). Growth kinematic profiles derived in this fashion generally presented identifiable zones where the elongation rate was increasing, where elongation was maximal, where the elongation rate was declining, and where elongation had effectively ceased (cessation zone) (Fig. 4). Thus, while it is clear that no sharp developmental boundaries exist along the inflorescence stem, these profiles allow us to locate points of maximal cell expansion and of growth cessation along individual stems, and to then harvest, pool and compare tissues with contrasting elongative growth behaviours. While the point of cessation of elongation was clearly evident in most of the kinematic profiles, a precise point of maximal elongation rate was often less clear as a result of extended regions of similar elongation rate and ⁄ or growth rate anomalies. In 11 out of 24 plants, enhanced elongation at the apex generated two maximal growth rate peaks. In seven of these cases, an abrupt spike in the relative growth rates of near-apical segments was observed at various times during the observation period. In other cases, specific segments within the zone of greatest elongation rate displayed distinctly suppressed growth relative to their neighbours (Fig. 4c), greater than the measurement variability observed in the growth rates for regions below the point of cessation, which likely reflected measurement error and ⁄ or length-smoothing artifacts. Since one goal for the present study was to enable highly actively elongating tissues to be compared with those that had ceased elongating, we selected as the de facto representative of the region of maximal elongation rate the segment within that region that was located nearest to the defined point of growth cessation. The position of growth stage transitions is widely variable among individual plants Our growth kinematic profiling system allows formal testing of the assumption that plants of equal height have similar developmental proportioning. For this purpose, we compared the distances from the shoot apex of the experimentally determined points of maximum growth rate and cessation for 24 individual Col-0 plants. These plants ranged in height from 10 to 15 cm (Figs 5, S3), but while the distribution in plant heights (Fig. 5a)  2012 The Authors New Phytologist  2012 New Phytologist Trust

Research 293

is shown to be normal through both a Shapiro–Wilks test and Q–Q plot for normality (Fig. 5b), we found considerable spread in the distributions of the values for maximum growth-rate (mean = 2.57, SD = 0.954, n = 24) and cessation (mean = 5.75, SD = 1.16, n = 24) distance from the apex. This points to a high degree of variability in the position of these growth phases within individual plants (Fig. 5c,e). To test the significance of the differences in these distributions, we performed a Welch two-sample t-test to reveal that t = )10.1532 (df = 42.46) for a P-value of 6.3 · 10)13. The maximum growth rate distribution had a mean of 2.58 and a 95% confidence interval (CI) of ± 3.8, while the cessation distribution had a mean of 5.75 and a 95% CI of ± 2.54. While these means are significantly different from each other, the populations do overlap. Scatterplot graphing also demonstrates that the relationship between overall plant height and the distances of cessation (r = 0.24) was even weaker than that of maximum growth rate (r = 0.33) from the apex (Fig. 5d,f). We also found that the distance required to transition from maximum growth rate to cessation along a given stem was quite variable, with a wide distribution in this metric (mean = 3.174, SD = 0.67, n = 24) and a poor correlation with overall plant height (r = )0.06) (Fig. 5g,h). This indicates that identification of either the point of maximum growth rate or the point of growth cessation does not allow reliable prediction of the position of the other. Growth kinematic profiles facilitate isolation of discrete stages of cell length To assess the precision with which developmentally matched cell cohorts could be isolated from stem segments of different plants based on positional information gleaned from their GKPs, we examined longitudinal sections of the relevant segments and measured the cell lengths within two stem tissues – the photosynthetic cortical cells located immediately below the epidermal layer, and a layer of more elongated cells that forms a barrier between the cortical cells and developing fibres in the interfascicular region. Cell length data were collected from young (subapical) tissue just beginning to elongate, from maximally elongating tissues, from tissue identified as having just ceased elongating, and from tissue located > 7 cm below the apex, a zone in which cells are no longer expanding (Fig. 6). This morphometric analysis of two cell types across a sample of six plants (average plant height = 12.3 cm, SD = 1.6) revealed that, for both cell types, significant differences in cell length existed between young tissue (0.5–1.0 mm below apex), maximum growth rate tissue (mean = 2.6 cm from apex, SD = 0.58) and growth cessation tissue (mean = 5.7 cm from apex, SD = 0.71). Importantly, there was no significant difference in lengths of either cell type between the tissues identified as growth cessation tissues and tissues from below the 7 cm point (mature, nonexpanding stem). The uniformity within each cell type population and the significant differences observed between the tissue populations are both consistent with our expectation that cells and tissues harvested as guided by GKP analysis represent developmentally discrete populations. New Phytologist (2012) 194: 287–296 www.newphytologist.com

New Phytologist

294 Research 80

Cell length (µm)

60

40

20

0

Young

Max−growth

Cessation

7 cm

Stages Fig. 6 Summary of Arabidopsis thaliana mean cell lengths (lm) for cortex (dark gray bars) and endodermal (light gray bars) cell types at young, maximum growth-rate, cessation, and post-cessation (7 cm) sampling points, where 20 cortical and 10 endodermal cells were measured. Error bars denote standard deviations (n = 6, a = 0.05).

Discussion The measure of success of any tissue sampling methodology is whether pooled and compared tissues consistently represent the biological context of interest, so that the results of molecular ⁄ cytometric ⁄ biochemical analyses can be legitimately correlated with differences in developmental state. In order to study molecular events that regulate directional tissue expansion in the bolting Arabidopsis inflorescence stem, we wished to isolate and pool tissue samples representing developmentally distinct growth stages across biological replicate tissues. Many recent studies have inferred the developmental status of tissues based upon the assumption that tissues harvested from a particular stem region, or at a specific time in the plant growth cycle, will exhibit similar developmental proportioning (Ko & Han, 2004; Ehlting et al., 2005; Imoto et al., 2005). However, the validity of this assumption depends upon a high degree of uniformity in the developmental process, both along the developmental continuum that the stem represents and from plant to plant. In contrast to this assumption, we demonstrate here that developmental proportioning of the bolting inflorescence stem is highly irregular among plants of the widely studied Col-0 ecotype (Figs 5, S3). It has not been established whether this degree of variability is also observed for other Arabidopsis genotypes that have been used in stem elongation studies, such as Landsberg erecta (Ehlting et al., 2005). The resulting weak correlation between plant height and location of the regions of maximal growth rate and growth cessation can be expected to interfere with accurate analysis of cellular and molecular events accompanying this major developmental transition because of the averaging effect of pooling developmentally mismatched tissues. Our optical marker tracking system, developed for examination of primary inflorescence stems, reports the REGR status of individual stem segments, each of which can then be processed for cytological and ⁄ or molecular assays. This relatively highNew Phytologist (2012) 194: 287–296 www.newphytologist.com

resolution view of stem expansion reveals surprisingly frequent irregularities associated with the more apical region of the GKPs. Pronounced dips in REGR were recorded for specific segments on individual plants (e.g. Fig. S1; A1:seg8, A2:seg7, C2:seg9, A5:seg8, B5:seg11). In addition, many of the 24 plants profiled showed elevated growth rate at the apex. While it is possible that elongation in these plants is actually greatest at the apex, we cannot exclude the possibility that tags bounding these upper segments were inadvertently attached to silique pedicels appressed to the primary stem. This can introduce a technical error as the rapid movement of these tags would reflect the combined elongation of primary stem and silique pedicel. Despite these anomalous patterns in the growth of the upper part of the stem, we are able to routinely identify and harvest segments representing points of maximal growth rate and cessation, and morphometric analysis of the resulting sample pools yielded morphologically distinct cell populations (Fig. 6). Fifty years after the establishment of REGR calculation of root systems (Erickson & Sax, 1956), our approach finally provides a practical means to overcome the challenges of time-lapse imaging of the Arabidopsis inflorescence stem architecture, and comprehensively and accurately characterizes the distribution of expansive growth across this important model organ. This methodology is amenable to a broader range of biological questions through modest technical refinements. Rather than present the full range of inflorescence development, the data presented here focus on a narrow developmental window that is of greatest relevance to those studying both cell wall expansion and the transition from primary to secondary cell wall formation (Ehlting et al., 2005; Imoto et al., 2005; Ko & Han, 2004). The current image chamber design (Fig. S5) was optimized to observe the top 10 cm of the primary stem of 10– 15 cm plants, although modifications to growth chamber dimensions, focal distance and camera resolution could conceivably address the entire inflorescence over a broader developmental window. While our approach to deriving GKPs has proven useful in isolating developmentally matched segments from different plants, the REGR data also represent one aspect of the plant phenotype, and as such they reflect both genetic and environmental influences. In Ricinus communis (castor oil plant), averaged GKPs of multiple plants for leaf midrib expansion have been shown to be remarkably different among experimental groups (genotype, treatment) through a diurnal cycle (Walter et al., 2002), and a similar pattern was observed for root expansion in Arabidopsis under differing light intensities (Nagel et al., 2006). We anticipate that quantifiable differences would also be evident among GKPs of inflorescence primary stems for experimental groups (ex. genotypes) with divergent aerial architectures. Our system may also meet the demands of systems biology, quantitative genetics, and mutant screening if the rate of image analysis can be increased. In addition to the 24 plants presented in this study, we have analyzed an additional 51 plants for immunohistochemical and transcriptome analysis (data to be published separately). Manual analysis of an image set for a single plant currently takes c. 2 h, but exciting new advances are being made  2012 The Authors New Phytologist  2012 New Phytologist Trust

New Phytologist in the establishment of velocity fields, based upon optical flow (Schmundt et al., 1998; Basu et al., 2007; Nagel et al., 2006). This could potentially facilitate observation of hundreds of plants within a single experiment if such automated feature tracking approaches that rely on natural features of the expanding surface can be adapted to deal with the challenges of imaging primary stems. Finally, the use of relatively large-scale synthetic optical markers for tagging stem segments provides several advantages over other synthetic markings such as small, randomly placed particles (ex. graphite) or ink marks. Use of larger tags with outer dimensions that extend beyond visual barriers such as cauline leaves not only allows segments to be individually monitored without interference, but avoids any changes in tag dimensions over time, in contrast to ink marks. These larger tags also allow convenient cataloging of segments for future use in downstream analyses (Video S3). In addition, since structure tensor methods are not restricted to tracking small particles (ex. graphite) or secondary leaf veins, our tagging system is, in principle, also amenable to automated generation of velocity fields for another approach to deriving REGRs. Larger tags may also support the future development of three-dimensional tracking of nastic movements from a single camera perspective, by taking advantage of the apparent changes in imaged tag dimensions associated with movement of the plant toward or away from the camera.

Acknowledgements We would like to acknowledge the financial support of the Natural Sciences and Engineering Research Council (NSERC) of Canada. We are grateful to Thomas Berleth, Minako Kaneda, Lacey Samuels, Ju¨rgen Ehlting, Edgar Spalding, Herman Ho¨fte and Miki Fujita for their perspectives on research approaches for Arabidopsis primary stem development. We are indebted to Randy Dean for image chamber design consultation, to Kevin Hodgeson for confocal expertise, and to Noriko Tanaka for assistance with feature tracking and cell measurements.

References Baskin TI, Cork A, Williamson RE, Gorst JR. 1995. STUNTED PLANT 1, a gene required for expansion in rapidly elongating but not in dividing cells and mediating root growth responses to applied cytokinin. Plant Physiology 107: 233. Basu P, Pal A, Lynch JP, Brown KM. 2007. A novel image-analysis technique for kinematic study of growth and curvature. Plant Physiology 145: 305–316. Birnbaum K, Jung JW, Wang JY, Lambert GM, Hirst JA, Galbraith DW, Benfey PN. 2005. Cell type-specific expression profiling in plants via cell sorting of protoplasts from fluorescent reporter lines. Nature methods 2: 615–619. Birnbaum K, Shasha DE, Wang JY, Jung JW, Lambert GM, Galbraith DW, Benfey PN. 2003. A gene expression map of the Arabidopsis root. Science 302: 1956–1960. Boyes DC, Zayed AM, Ascenzi R, McCaskill AJ, Hoffman NE, Davis KR, Go¨rlach J. 2001. Growth stage-based phenotypic analysis of Arabidopsis: a model for high throughput functional genomics in plants. Plant Cell 13: 1499–1510. Collings DA, Winter H, Wyatt SE, Allen NS. 1998. Growth dynamics and cytoskeleton organization during stem maturation and gravity-induced stem bending in Zea mays L. Planta 207: 246–258.  2012 The Authors New Phytologist  2012 New Phytologist Trust

Research 295 Ehlting J, Mattheus N, Aeschliman DS, Li E, Hamberger B, Cullis IF, Zhuang J, Kaneda M, Mansfield SD, Samuels L et al. 2005. Global transcript profiling of primary stems from Arabidopsis thaliana identifies candidate genes for missing links in lignin biosynthesis and transcriptional regulators of fiber differentiation. The Plant Journal 42: 618–640. Erickson RO, Sax KB. 1956. Elemental growth rate of the primary root of Zea mays. Proceedings of the American Philosophical Society 100: 487–498. French A, Ubeda-Toma´s S, Holman TJ, Bennett MJ, Pridmore T. 2009. Highthroughput quantification of root growth using a novel image-analysis tool. Plant Physiology 150: 1784–1795. Gandar PW. 1983. Growth in root apices. I. The kinematic description of growth. Botanical Gazette 144: 1–10. Hempel FD, Feldman LJ. 1994. Bi-directional inflorescence development in Arabidopsis thaliana: acropetal initiation of flowers and basipetal initiation of paraclades. Planta 192: 276–286. Hsiao TC, Acevedo E, Henderson DW. 1970. Maize leaf elongation: continuous measurements and close dependence on plant water status. Science 168: 590–591. Hummel GM, Naumann M, Schurr U, Walter A. 2007. Root growth dynamics of Nicotiana attenuata seedlings are affected by simulated herbivore attack. Plant, Cell & Environment 30: 1326–1336. Imoto K, Yokoyama R, Nishitani K. 2005. Comprehensive approach to genes involved in cell wall modifications in Arabidopsis thaliana. Plant Molecular Biology 58: 177–192. Ko J, Han KH. 2004. Arabidopsis whole-transcriptome profiling defines the features of coordinated regulations that occur during secondary growth. Plant Molecular Biology 55: 433–453. Minic Z, Jamet E, San-Clemente H, Pelletier S, Renou JP, Rihouey C, Okinyo DP, Proux C, Lerouge P, Jouanin L. 2009. Transcriptomic analysis of Arabidopsis developing stems: a close-up on cell wall genes. BMC Plant Biology 9: 6. Morris A, Silk W. 1992. Use of a flexible logistic function to describe axial growth of plants. Bulletin of Mathematical Biology 54: 1069–1081. Mullen JL, Ishikawa H, Evans ML. 1998. Analysis of changes in relative elemental growth rate patterns in the elongation zone of Arabidopsis roots upon gravistimulation. Planta 206: 598–603. Nagel KA, Schurr U, Walter A. 2006. Dynamics of root growth stimulation in Nicotiana tabacum in increasing light intensity. Plant, Cell & Environment 29: 1936–1945. Palaniappan K, Jiang HS, Baskin TI. 2004. Non-rigid motion estimation using the robust tensor method. Computer Vision and Pattern Recognition Workshop 1: 25. Pouteau S, Albertini C. 2009. The significance of bolting and floral transitions as indicators of reproductive phase change in Arabidopsis. Journal of Experimental Botany 60: 3367–3377. Pu L, Brady S. 2009. Systems Biology Update: cell type-specific transcriptional regulatory networks. Plant Physiology 152: 411–419. Schmundt D, Stitt M, Ja¨hne B, Schurr U. 1998. Quantitative analysis of the local rates of growth of dicot leaves at a high temporal and spatial resolution, using image sequence analysis. The Plant Journal 16: 505–514. Schwab B, Mathur J, Saedler R, Schwarz H, Frey B, Scheidegger C, Hu¨lskamp M. 2003. Regulation of cell expansion by the DISTORTED genes in Arabidopsis thaliana: actin controls the spatial organization of microtubules. Molecular Genetics and Genomics: MGG 269: 350–360. Sugimoto K, Williamson RE, Wasteneys GO. 2000. New techniques enable comparative analysis of microtubule orientation, wall texture, and growth rate in intact roots of Arabidopsis. Plant Physiology 124: 1493. Suh MC, Samuels AL, Jetter R, Kunst L, Pollard M, Ohlrogge J, Beisson F. 2005. Cuticular lipid composition, surface structure, and gene expression in Arabidopsis stem epidermis. Plant Physiology 139: 1649–1665. Walter A, Feil R, Schurr U. 2002. Restriction of nyctinastic movements and application of tensile forces to leaves affects diurnal patterns of expansion growth. Functional Plant Biology 29: 1247. Wenzel CL, Williamson RE, Wasteneys GO. 2000. Gibberellin-induced changes in growth anisotropy precede gibberellin-dependent changes in cortical microtubule orientation in developing epidermal cells of barley leaves. Kinematic and cytological studies on a gibberellin-responsive dwarf mutant, M489. Plant Physiology 124: 813–822. New Phytologist (2012) 194: 287–296 www.newphytologist.com

New Phytologist

296 Research Ye Z, Freshour G, Hahn MG, Burk DH, Zhong R. 2002. Vascular development in Arabidopsis. In: Jeon KW, ed. A survey of cell biology. Maryland Heights, MO, USA: Elsevier, 225. Zhong R, Lee C, Zhou J, McCarthy RL, Ye Z. 2008. A battery of transcription factors involved in the regulation of secondary cell wall biosynthesis in Arabidopsis. Plant Cell 20: 2763–2782.

Supporting Information

Fig. S5 Illustrations of growth imaging chamber. Notes S1 Script for all calculations and plotting of growth kinematic data. Video S1 Tagging of one complete plant with c. 15 paper rectangles for growth analysis within the imaging chamber.

Additional supporting information may be found in the online version of this article.

Video S2 Video depicting the process of feature tracking plants in ImageJ.

Fig. S1 Electrostatically attached PVDF disks as an alternative optical marker system.

Video S3 Video depicting the harvesting of segments bounded by tags.

Fig. S2 Effect of smoothing length measurements on growth kinematic profiling.

Video S4 Video rendering of an image series of cell outlines used to calculate cell length.

Fig. S3 Surface plots of REGRs and LOWESS-predicted growth kinematic profiles for 24 plants.

Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office.

Fig. S4 Tagged plants placed within plant-specific growth restraints and chamber.

New Phytologist is an electronic (online-only) journal owned by the New Phytologist Trust, a not-for-profit organization dedicated to the promotion of plant science, facilitating projects from symposia to free access for our Tansley reviews. Regular papers, Letters, Research reviews, Rapid reports and both Modelling/Theory and Methods papers are encouraged. We are committed to rapid processing, from online submission through to publication ‘as ready’ via Early View – our average time to decision is