Académique Documents
Professionnel Documents
Culture Documents
Summary
Ethylene is structurally the simplest phytohormone and regulates multiple
aspects of plant growth and development. Its effects are mediated by a signal
transduction cascade involving receptors, MAP kinases and transcription
factors. Many morphological effects of ethylene in plant development, including
root size, have been previously described. In this article a combined geometric
and algebraic approach has been used to analyse the shape and the curvature
in the root apex of Arabidopsis seedlings. The process requires the fitting of
Bzier curves that reproduce the root apex shape and the calculation of the
corresponding curvatures. The application of the method has allowed to identify
significant differences in the root curvatures of ethylene insensitive mutants
(ein2-1 and etr1-1) with respect to wild-type Columbia.
Key words: Arabidopsis, Bzier curve, curvature, development, ethylene, root
apex.
Introduction
The Arabidopsis root has been used in recent years as a unique model system
to study cell differentiation during development. In the embryonic root a group of
four central cells in the quiescent center is sourrounded by distinct initial cells.
This group is the origin of the different cell types including the columella rootcap, cortical and endodermal cell files, epidermis, and the steele (pericycle and
vascular tissue) (Schiefelbein and Benfey, 1994; Schiefelbein et al., 1997).
Thus, in a reduced number of cell layers, differentiation goes from the nondividing and non-differentiated status in the quiescent center to cells
differentiated in distinct cell types, passing through dividing non-differentiated
meristematic cells. The molecular basis of cell differentiation involves the
coordinated expression of genes encoding key regulatory proteins such as
scarecrow, a member of the GRAS family of DNA binding proteins (Sabatini et
al., 2003). Hormonal regulatory networks control the expression of these genes
(Fu and Harberd, 2003).
During seed imbibition, embryonic root cells take water and increase their
size leading to germination. After radicle protrusion some cells begin to divide.
Once the developmental program is initiated, root shape will be established by
the coordinated action of their cells in response to hormones and environmental
conditions (Beemster et al., 2003). As it is well known, variables measuring
characteristics (e.g., root length, width or root hair number) which depend on
the additive effect of many low intensity factors follow normal distributions.
Ethylene, the simplest structurally known plant hormone, is produced in
plants in two steps from the amino acid methionine via SAM (S-adenosyl
methionine) and ACC (1-aminocyclopropane-1-carboxylic acid) by the
consecutive action of two enzymes: ACC synthase and ACC oxidase. Ethylene
affects diverse aspects of plant development ranging from germination and
seedling morphology to fruit ripening and senescence (Abeles et al., 1992).
Historically, the effects of ethylene on plant development were first described in
etiolated pea seedlings (Neljubov, 1901). They consist in swelling of the
hypocotyl , growth inhibition in the root and in the hypocotyl and an exaggerated
hook curvature. These characteristics of seedlings grown in the dark in the
presence of ethylene were called the triple response and have been useful in
the identification of mutants in the ethylene biosynthesis and signal transduction
pathways (Guzmn and Ecker, 1990). Other effects of ethylene in development
and in the modulation of several aspects of plant form include the response to
physical stress, leaf abscission, flower development and sex determination (for
a review see Dolan, 1997).
In Arabidopsis, ethylene has a remarkable effect on root development: it
inhibits elongation and promotes radial expansion. Similar effects are obtained
with ACC, the ethylene precursor. Treatments with ethylene and ACC are
effective in inducing root hairs (Masucci and Schiefelbein, 1996; Pitts et al.,
1999; Tanimoto et al., 1995). The response is rapid: changes in cell elongation
and the induction of root hairs can be observed after incubation of minutes (Le
et al., 2001). With longer incubation times, ethylene affects the overall root
shape, resulting in decreased root length and increased width. Thus, ethylene
may strongly influence root shape at early developmental stages. Treatment of
wild-type seedlings with ethylene results in morphological changes that are
coordinates of every point marked were provided by AnalySIS . In this way, for
each image, a data series of the form (xi, yi) were obtained. These data points
were used to create a function representing the curve that approximates the
shape of the root outline.
Curve approximation
Since in this work we focus on the shape of the root, traditional methods forcing
the curve to pass throughout the points are inappropriate. With these fitting
methods, the approximated curve is close to the real shape (in L 2 or L
sense, for example), but oscillates around it (Yang, 2001). So, we have chosen
Bzier curves to approximate the shape of the roots, and curvature as a
measure of shape fidelity.
Bzier curves (Bzier, 1968; Gordon et al., 1974) do not pass throughout
data points (except the first and the last ones), but data points are used as
control points. The multiplicity of data points can be used to improve the fitting.
Bzier curves have been widely used in computer drawing (Rogers and Adams,
1989). They have suitable features for our purposes: 1) Bzier curves have no
great oscillations around the points; 2) they follow the polygonal lines joining the
points and conserve the convexity, i.e. spurious inflection points do not appear;
3) their equations are given by polynomials, and then, are easy to handle (with
a computer code).
From a series of arranged data pi = (xi, yi), i=0,...n, the parametric Bzier
curve is obtained (see e.g. Risler, 1992) as X(t)=(x(t),y(t)) with
n
x(t) = xi Bi ,n (t ) ,
where t [0,1] and
i =0
y(t) = y i Bi ,n (t )
i =0
n
Bi ,n (t ) = (1 t ) n i t i
i = 0,...n,
i
represent the n+1 Bernstein polynomials (Lorentz, 1986). So, x(t) and y(t) are
polynomials of degree n and the parameter t, which varies in the interval [0,1],
expresses the percentage of the total curve length traversed. Notice that X (0)
is the initial point (x0, y0) and X (1) is the final point (xn, yn).
Geometric analysis
The curvatures of the adjusted Bzier curves were calculated from their
parametric equations (x(t), y(t)) by means of formula (DoCarmo, 1976) :
x' (t ) y ' ' (t ) x' ' (t ) y ' (t )
,
t [0,1] .
(t) =
( x' (t ) 2 + y ' (t ) 2 ) 3 / 2
Curvature measures the rate at which the unit tangent vector is changing with
respect to arc length. So, all graphical representations of the curvatures of
Bzier curves approximating root outlines have similar bell-shaped form (see
Figure 2b). The values of the curvature near to the extremes of the interval are
low, since the curve is similar to a line in their proximities. When the parameter
goes up to the root tip, curvature increases. The maximum value of the
curvature is attained on the apex vertex. Bzier curves fitting pointed roots lead
to tall and narrow bells, whereas those corresponding to blunt roots give
flattened bells.
Once calculated the maximum value of the curvature function for each root, a
variable called APEXCUR containing these values was created. For each root
this variable measures curvature value in the root tip.
A Mathematica code which, from a set of input data points, gives the Bzier
curve, its graphical representation (figure 1a), the graphic of a curve
representing the curvatures (figure 1b), and the maximum value of the curvature
has been written out and is available upon request at the authors e-mail
addresses.
Statistical analysis
We are interested in the homogeneity of populations col, ein and etr with
respect to variable APEXCUR. Thus, a sample of six root images for each
population (24 hours after imbibition) was analysed and, following the process
described above, their APEXCUR values were obtained. Main statistics of
variable APEXCUR were calculated.
ANOVA procedure (Fisher, 1935; Bliss, 1967) was used to test the significance
of the difference between the three populations. The null hypothesis H0 was that
they represent samples from the same normal distributed population, i.e.
populations col, ein and etr are normally distributed and their parameters (mean
and variance) coincide. We tested H0 in a level of significance = 0.01 .
To study if data have significant difference in dispersion, i.e. to test the variance
homogeneity, a Levene test was applied (Levene, 1960). In this case, the null
hypothesis H1 was that the samples come from normal distributed populations
with the same variance (level of significance = 0.01 ).
If H0 is rejected after ANOVA test and H1 is accepted in Levene test, then we
must conclude that populations col, ein and etr differ in their means but not in
their variances with respect to variable APEXCUR. In this case, to find which
pair or pairs of populations (col-ein, col-etr, ein-etr) give this heterogeneity,
many multiple comparison tests are available. In our case, a Scheff multiple
comparison test (Scheff, 1953) was carried out.
The analysis was repeated with an equal size sample of seedlings taken at 48
hours after imbibition.
Results
For each image, a Bzier curve fitting the root outline and the corresponding
curvature function, were obtained. Figure 3, which contains one example of
each genotype, illustrates this process. Afterwards, APEXCUR values of this
sample were calculated. Main statistics of variable APEXCUR are summarized
in Table 1.
The significance of the difference between root tip curvatures of groups col, ein
and etr was tested by analysis of variance. From the F value obtained in Levene
test (Table 2), there is insufficient evidence to reject the hypothesis that the
variances are equal. So we accept that our three samples represent normal
populations with the same variance.
The obtained value F=17.586 in ANOVA test (Table 3) corresponds to a
probability lower than P=0.001. On this basis, it was concluded that the groups
col, ein and etr have different means with a degree of reliability higher than
99.9%.
Finally, Scheffes multiple comparison test gives that a pair of populations
have unequal means if their sample means i , j fulfill the inequality
i j 0.0193 F1 ; 2,12 .
With = 0.01 , the value of the term on the right is 0.366 and we conclude that,
with this level of significance, col population differs from ein and etr in their
curvatures. Obviously, there is no evidence to reject the hypothesis that the
populations ein and etr have the same mean.
Figure 4 contains the twelve curvature functions corresponding to col and etr
samples (green, col functions; red, etr). It shows the differences between the
curvature functions for both populations and, in particular, the lack of
overlapping between their APEXCUR values.
The analysis was repeated taking different sets of points with similar results.
Also similar conclusions were obtained when the analysis was repeated for a
sample consisting of six seedlings for each genotype at 48 hours after
imbibition.
On the other hand, seedlings of the mutant genotype eto1-1 were included in a
similar analysis. The mean curvature value for eto1-1 was 1.965. The values
were included between 1.65 and 2.47 and the standard deviation for this
genotype was 0.29. Scheffs test revealed that curvature values for eto1-1 are
not statistically different from the values corresponding to the other genotypes
under study.
Discussion
Plant shape results from the coordinated growth of many thousands of cells.
From fertilization and early embryonic development, very efficient comunication
mechanisms are established involving the different organs of a plant. During
plant development, a multiplicity of mechanisms exert their effects both in short
(i.e. between neighbour cells) and long distance (e.g. hormonal signals
transmitted throughout the plant). Recently, root length has been shown to be
partly controlled by shoot produced phytohormones and regulated by the
coordinated action of gibberellins, auxins and ethylene (Achard et al., 2003;
Cervantes, 2003; Fu and Harberd, 2003).
In a recent article, Rolland-Lagan et al. (2003) proposed that growth is
governed by three factors: growth rate, anisotropy and direction. In root growth
the main axis determines the direction and, after germination, growth is
predominantly driven by the gravitropic stimulus. Major variations may concern
rate of growth and anisotropy. Changes in anisotropy are related to variability in
radial expansion. Ethylene is known to inhibit growth and to promote radial
expansion in roots (Dolan and Davies, 2004). Thus, this phytohormone
regulates crucial aspects in root morphogenesis allowing the adaptation of root
growth to rapid responses in environmental changes (Le et al., 2001).
From a practical point of view it may be convenient to analyse
developmental aspects that: 1) may be observed at early developmental stages.
2) can respond to known signals or environmental factors, and 3) can be
expressed in mathematical terms. Growth characteristics observed at early
stages of development are the consequence of interactions among a lower
Acknowledgements
We thank Jos Javier Martn Gmez for his help in the elaboration of figures.
References
Abeles F, Morgan P, Saltveit M (1992). Ethylene in Plant Biology, San Diego:
Academic press.
Achard P, Vriezen WH, Van Der Straeten D, Harberd NP (2003). Ethylene
regulates Arabidopsis development via the modulation of DELLA protein growth
repressor function. Plant Cell 15 (12), 2816-25.
Beemster, GTS, Fiorani F, Inz D (2003). Cell cycle: the key to plant growth
control? Trends in Plant Science, 8(4), 154-158.
Bzier, PE (1968). How Renault uses numerical control for car body design and
tooling. SAE Congress, paper 680010, Detroit.
Bleecker, A B, Estelle, MA, Somerville C, Kende H (1988). Insensitivity to
ethylene conferred by a dominant mutation in Arabidopsis thaliana. Science
241, 1086-1089.
Bliss, CI (1967). Statistics in Biology (vol. I), McGraw Hill, New York.
10
11
12
13
Figures
Figure 1
1.5
0.5
0.2
a)
0.4
0.6
0.8
b)
Figure 2
14
16
15.5
15
14.5
14
13.5
13
b)
2
1.5
1
0.5
c)
0.2
0.4
0.6
0.8
a)
14
13
12
11
10
b)
10
11
12
13
2
1.5
1
0.5
0.2
c)
0.4
0.6
0.8
a)
9.5
9
8.5
8
7.5
7
6.5
6
6.5
b)
7.5
8.5
9.5
10
1.5
1.25
1
0.75
0.5
0.25
c)
0.2
0.4
0.6
0.8
a)
Figure 3
15
3
2.5
2
1.5
1
0.5
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Figure 4
16
Tables
TYPE
col
Mean
2.17450
Cases
6
St. D.
.225846
Var.
.051006
Min.
1.921
Max.
2.509
ein
1.66496
.129154
.016680
1.509
1.848
etr
1.67350
.139406
.019434
1.432
1.847
Total
1.83766
18
.329275
.108422
1.432
2.509
Levene
1.969
DF1
2
DF2
15
Sig.
.174
ANOVA Table
SS
APEXCUR
* TYPE
Between groups
Within groups
Total
DF
MS
1.021
.511
.436
15
.029
1.457
17
F
17.586
Sig.
.000
Table 3. Results of ANOVA test comparing the means of groups col, ein and etr for variable
APEXCUR.
17