Content uploaded by Sureshkumar Balasubramanian
Author content
All content in this area was uploaded by Sureshkumar Balasubramanian on Oct 05, 2015
Content may be subject to copyright.
* Corresponding Address: email:esskay@tuebingen.mpg.d
1 * Trait is a character one is interested in understanding, which can comprise several phenotypes. For example, “ flowering time”
will be a trait and “ late-flowering” is a phenotype.
Natural Variation and QTL Mapping in Understanding Plant
Development
Sureshkumar Balasubramanian
Department of Molecular Biology, Max-Planck Institute for Developmental Biology, Spemannstrasse 39,
Tuebingen, Germany.
(Received on 10 March 2004; Accepted after revision 29 April 2004)
Understanding the genetic and molecular basis of a developmental mechanism remains a challenge. Mende-
lian approaches in model organisms such as C. elegans, Drosophila, Zebrafish, mouse and Arabidopsis have
greatly enhanced our understanding of discrete developmental processes such as organ identity and pattern
formation. However, many traits like phase changes during development are quantitative in nature. The con-
tinuous nature of the quantitative phenotypes and the underlying genetic complexity greatly diminish the
power of Mendelian genetic approaches. In a parallel approach natural isolates of a species that often show
genetically controlled phenotypic variation and experimental populations derived from these isolates can be
used for Quantitative Trait Locus (QTL) mapping that can explain this phenotypic variation. Powerful compu-
tational and statistical methods developed in the last decade coupled with the advent of molecular markers
have revolutionized our ability to detect the QTLs. At least for some of the major effect QTLs cloning of the
Quantitative Trait Gene(s) (QTG) underlying the QTL has been possible. Understanding the genetic architec-
ture of naturally occurring phenotypic variation in developmental traits will provide clues not only to the
molecular basis of quantitative traits, but also to how environment interacts with the physiology of the plant
and presumably explain the adaptive significance of such genetic variation. In this article, I review how the
natural variation and QTL mapping has helped our understanding of plant development using examples mainly
from maize, rice, Arabidopsis and tomato. Specifically, I will address how exploiting natural variation has
advanced our knowledge of the genetic basis of plant architecture, helped us identify novel genetic pathways
and genes and even elucidate unknown functions of known genes. At the end, I discuss the potential of natural
variation and QTL mapping to understand complex environmental effects that influence plant development.
Key Words: Arabidopsis, Rice, Natural variation, QTL mapping, Quantiative and Qualitative
Proc. Indian Natl. Sci. Acad. B70 No.4 pp. 431-441 (2004)
Introduction
Unlike animals, plants exhibit extensive post-embry-
onic development of features. Germination results in a
seedling, which contains the shoot and root meristems
that subsequently give rise to all the other organs of the
plant post-embryonically. As a major part of plant`s
development is post-embryonic, its environment
heavily influences its development. Therefore, most
developmental processes in plants are under the con-
trol of both endogenous and exogenous factors. The
interaction of endogenous genetic and exogenous en-
vironmental factors results in complex responses that
are not easily amenable to classical genetic approaches
using induced mutants and poses an interesting prob-
lem for understanding plant development. Tradition-
ally plant breeders approached this problem by look-
ing into the naturally occurring variation that is present
in natural populations and using this variation in breed-
ing programmes to generate viable useful new culti-
vars for domesticated plants. However, understanding
the genetic and molecular basis of such complex re-
sponses remains a challenge even today. In this review,
I will address the nature of quantitative developmen-
tal traits, advantages of using natural variation to un-
derstand these traits and how natural variation can be
exploited to enhance our knowledge of plant develop-
ment using examples of successful application of these
approaches in a few plant species.
432 Sureshkumar Balasubramanian
Quantitative and Qualitative Developmental
Traits
Trait is a character one is interested in understanding,
which can be comprised of several phenotypes. For ex-
ample, `flowering time` will be a trait and `late-flower-
ing` is a phenotype. Whether a particular trait is quan-
titative or qualitative depends on both the trait itself as
well as how the trait is measured. Some traits are quali-
tative in nature. For example the floral organ identity
is a qualitative trait with the floral homeotic mutants
displaying qualitative phenotypes, which is a discrete
measurement. If a population of plants that segregate
for such a phenotype were analyzed, one would see two
classes of plants, ones that have the wild-type pheno-
type and the ones that show a mutant phenotype. How-
ever, other developmental traits are quantitative in na-
ture e.g., the height of a plant, the number of flowers,
the size of a fruit etc. When such a trait is measured in a
population, often one does not see discrete classes but
rather a multitude of classes that form continuous dis-
tribution. Figure 1 gives an example of a continuous
variation observed in an F2 population of a cross be-
tween two wild strains of Arabidopsis in their flower-
ing time. Often such phenotypes result from allelic in-
teractions and environmental effects. As plant devel-
opment is mostly post-embryonic, many developmen-
tal characters show a significant environmental effects
resulting in several quantitative developmental traits.
Analysis of quantitative traits poses a serious challenge
to Mendelian approaches.
Induced Mutations have Certain Limitations.
Traditional Mendelian genetic approaches have greatly
enhanced our understanding of molecular mechanisms
controlling plant development. Using mutants in
Arabidopsis, rice, maize, tomato and other plants we
have begun to understand developmental processes at
a molecular and cellular level. However, induced mu-
tations have certain limitations. For instance, the phe-
notype of an induced mutant very much depends on
the `wild type` used for generating the mutant, which
in itself is an arbitrarily chosen strain based on charac-
ters that makes it suitable for laboratory research.
While, a lot of information can be obtained from a
mutagenised population, the definition of gene func-
tion will be restricted by the choice of genetic back-
ground. A regular mutagenesis screen will not be able
to identify genes for which the `wild-type` carries a non-
functional allele or even lacks the gene. Epistatic inter-
actions ensure certain phenotypes occur only in certain
`wild types`. Loss-of-function of a gene may also mask
the effects of mutations at other loci that contribute to a
similar phenotype. For example, mutagenesis of a win-
ter annual accession of Arabidopsis has revealed the
role of FRIGIDA LIKE 1(FRL1), which was not de-
tected in mutant screens performed previously using
regular laboratory strains (Michaels et al. 2004). Very
often mutants display their phenotype to varying
strengths depending on the genetic background. Le-
sions in genes that are essential during gametophytic
development and subsequently play other developmen-
tal roles will be difficult to detect. Furthermore, often
the screen enables one to identify drastic phenotypes
that usually result from `loss-of-function` alleles. How-
ever, there are subtle differences in phenotypes that can
have strong impact on the fitness of the plant. Such phe-
notypes will also be difficult to detect through tradi-
tional mutagenesis. For traits that show a significant
GXE interaction, such limitations are of considerable
worry.
Figure 1.An example for a continuous distribution: The figure
shows the flowering time of an F2 population of plants from a cross
between 2 wild strains of Arabidopsis. Total leaf number provides
a physiological measure of flowering time and it is highly corre-
lated with the number of days required for flowering.
Figure 2 shows the morphological variation that can be seen among
the accessions of Arabidopsis thaliana. These accessions are planted
on the same day and grown under identical conditions. The inher-
ent genetic differences and the variation in the GxE interactions
contribute to their distinct phenotypes.
QTL Mapping in understanding Plant Development 433
Natural Variants provide an Attractive Alter-
native to Induced Mutants
One way to complement the Mendelian genetic ap-
proach is to use naturally occurring variation for un-
derstanding plant development (Alonso-Blanco &
Koornneef 2000, Koornneef et al. 2004). First I will
broadly define the use of the term natural variation in
this review. I use the term natural variation in this re-
view to describe the naturally occurring variation in
phenotypes within multiple cultivars of the same spe-
cies. Some times, two cultivars may not show a pheno-
typic difference between them for a particular trait but
one might see transgression of the phenotype in a ge-
netic cross between the cultivars that show no pheno-
typic difference by themselves. I have used natural
variation in this review to describe this variation as
well, since this transgression though man-made is a
result of the interaction of naturally occurring alleles.
There are many advantages in using natural variants.
First, there is substantial amount of genetically con-
trolled natural variation that exists even within species.
Figure 2 shows the morphology of Arabidopsis plants
collected in different parts of the world (referred to as
accessions), but grown at the same time and conditions
in our laboratory. It is rather obvious that these acces-
sions are visibly distinct from each other. Second, quite
a few plant species are inbreeding and many others can
be maintained via inbreeding. As a result often species
such as Arabidopsis and rice can be made to contain
almost no heterozygosity after 7-8 generations of
selfing. This allows for obtaining recombinant inbred
lines (RILs). From F2 or F3 or BC1, single seed descen-
dent plants are grown for few generations (7-8 usually)
as single seed descendents to have lines that contain
very little heterozygosity in their genome. Selfing in
plants makes the construction of RILs relatively easier
in plants compared to animals with the exception of
C.elegans. Once a population of RILs is generated and
a genetic map is made, it provides an immortal source
that can be used again and again for analyzing mul-
tiple traits. Third, using natural variation one can de-
tect `change of function` alleles rather than “loss of func-
tion” alleles. Fourth, analysis of natural variants often
reveals novel alleles thereby aiding in deciphering pre-
viously unknown functions of known genes.
Quantitative Trait Loci (QTL)
Quantitative Trait Locus (QTL) refers to a region of a
chromosome that contributes to and thus can explain a
certain amount of variation that exists in a population
from which the QTL is detected. A QTL may contain a
single gene or a multiple closely linked genes. QTLs can
be large effect QTLs that can explain a larger propor-
tion of variation or can be small effect QTLs explaining
only little variation. Usually one detects multiple QTLs
contributing to a particular trait in an experimental
cross. However, taken together these QTLs can typically
account for only 70 percent of the phenotypic variation
and the remaining 30% is due to smaller effect QTLs,
which are difficult to detect. Since a QTL detected in an
experimental cross explains the variation among the
parental lines used in the cross, multiple populations
need to be analyzed to get a complete picture.
Due to the quantitative nature, the traditional map-
ping methods of mapping mutations is not efficient in
identifying QTLs. Therefore, statistical approaches are
needed to identify the QTLs. Using genome wide link-
age statistics on phenotypic and genotypic values of a
population, the ratio of the likelihood of having a QTL
and not having a QTL contributing to a trait in a par-
ticular region is calculated. This ratio (logarithm of
odds, LOD) provides the LOD score and a high LOD
score suggests a linkage and thus the presence of a QTL.
Using permutations, one can calculate the maximum
LOD score that can be achieved by mere chance. This
can then be used to identify the significance of the de-
tected QTLs. Recent years have seen an upsurge in the
number of powerful computational and statistical meth-
ods available for detection of QTL. Some of the new
methods view the detection of QTL, as a problem of
model building, in which a search for a QTL model that
can explain the maximum amount of variation is sought
for. Since the methodology of QTL detection would by
itself form a topic for a review I will advice the readers
to look into the following publications (Broman 2001,
Doerge & Churchill 1996, Hoeschele et al. 1997, Ma et
al. 2002, Satagopan et al. 1996, Wu et al. 2004).
Once a QTL is detected, further fine mapping can be
done using Near Isogenic Lines (NIL). NILs differ only
at the QTL regions while all other parts of the genome
remain homozygous. These can be generated for ex-
ample by back-crossing an RIL repeatedly to either of
the parental lines and by selecting for the segregation
of the phenotype. One can then follow the segregation
of the QTL, which is `Mendelised` in the NILs for sub-
sequent analysis. While it is agreeable that dissecting
the molecular basis of natural variation and quantita-
tive traits is cumbersome compared to simple tradi-
tional mutagenic approaches, the increasingly accumu-
lating genomic resources shall greatly enhance our
power of QTL detection and cloning immensely
(Borevitz & Nordborg 2003). Furthermore, QTL ap-
proaches are also complemented by association map-
ping methods, linkage disequilibrium mapping and
functional mapping approaches that help us under-
2 RIL: A recombinant inbred line can be viewed as immortalized F2 populations. Usually from F2, single seed descendent plants are
grown several generations (7-8 usually) as single seed descendents to have lines that contain ver
y little heterozygosity.
434 Sureshkumar Balasubramanian
stand the molecular basis of naturally occurring genetic
variation (Flint-Garcia et al. 2003, Ma et al. 2004,
Nordborg & Tavare 2002, Wu et al. 2004).
Using Natural Variation to Understand Plant
Architecture
One of the important questions in developmental biol-
ogy is how an organism attains a particular architec-
ture. This sculpturing involves many loci functioning
at distinct times individually as well as co-operatively
for the organisms to finally achieve a particular size and
shape. Plants exhibit enormous variation in size and
form and this provides a huge genetic resource with
which to ask questions on the genetic basis of plant ar-
chitecture. I will discuss two examples from maize and
tomato to illustrate how natural variation and QTL
mapping have been used to advance our understand-
ing of the molecular basis of plant architecture.
Maize (Zea mays ssp.mays) is a domesticated crop
plant derived from the wild ancestor teosinte (Zea mays
ssp.parviglumis) (White and Doebley, 1998). These two
sub-species differ substantially in their morphology
(Doebley & Stec 1991). Teosinte plants have a bushy
appearance compared to maize. The lateral branches
do not elongate in maize, which shows a strong apical
dominance while the teosinte plants show extensive
elongation of the lateral branches at all the internodes.
There exists a spontaneous mutant in maize called te-
osinte branched 1 (tb1), which resembles the wild an-
cestor teosinte in its inflorescence architecture. By do-
ing a QTL mapping on F2 populations derived from a
cross between maize and wild teosinte strains,
Doebley’s group has shown that indeed TB1 is a ma-
jor QTL that controls the architecture of the maize plants
(Doebley & Stec 1991, Doebley & Stec 1993). Subsequent
cloning of TB1 revealed that it encodes a protein simi-
lar to that of a gene called CYCLOIDEA (CYC), first
described in snapdragon (Antirrhinum majus)
(Doebley et al. 1997, Luo et al. 1996). Antirrhinum
plants that are mutants for cyc show alterations in the
floral architecture. Further analysis of TB1 from maize
and teosinte suggests that the domestication of maize
involved not a loss of function of TB1, but rather an
increase in the expression level of TB1 in maize com-
pared to teosinte (Clark et al. 2004, Hubbard et al. 2002).
An increased expression of TB1 results in suppression
of axillary meristem growth leading to maize like ar-
chitecture. Apart from providing insights into how
plant architecture can be altered, this example also
shows how natural variation can be used to understand
morphological evolution in natural populations.
A nice second example for the use of natural varia-
tion and QTL mapping to understand size and shape
variation is the cloning of QTLs controlling fruit size
and shape in tomato. After a decade long work
Tanksley’s laboratory was successful in cloning fw2.2,
a QTL that controls fruit size in tomato (Frary et al.
2000). The wild progenitors of tomato have fruits of
very small size compared to modern tomatoes. This
variation has been mapped to fw2.2, a major QTL regu-
lating fruit weight. Fw2.2 can explain up to 30% of the
variation observed in tomato fruit weight (Frary et al.
2000). Cloning and molecular analysis of FW2.2 re-
vealed once again that the molecular basis for the phe-
notypic variation might not lie at the protein level but
instead be at the expression level due to variation at
the promoter region. Recently, Tanksley’s laboratory
has shown that the basis of this natural variation could
be heterochronic regulatory mutations*1 at the fw2.2
QTL (Cong et al. 2002). FW2.2 gene expression varied
in its timing of expression between the large and small
fruited alleles by a week and this difference was corre-
lated with mitotic activity in the early stages of fruit
development (Cong et al. 2002). Therefore, it is very
likely that this heterochronic allelic variation is the un-
derlying cause for the size differences observed in to-
matoes.
QTL mapping in tomato enhanced our understand-
ing of not only fruit size but also fruit shape. Domesti-
cation from the wild ancestral plants was accompanied
by a change in the shape of the fruits and one of the
common fruit shapes that are often observed in domes-
ticated varieties is the pear-shaped fruit (Liu et al. 2002).
Contrary to the previously described QTLs, cloning of
the OVATE loci revealed that the round shaped and
pear-shaped fruits differ at their OVATE loci by a par-
tial or complete loss of function. Complementation and
over expression studies confirmed that OVATE func-
tions as a plant growth suppressor. Molecular cloning
of OVATE also showed that it encodes a protein that
belongs to a class of regulatory proteins that have not
been described so far (Liu et al. 2002). This is one of the
examples where QTL mapping have helped in identi-
fying factors that have not been detected through the
series of traditional mutagenic experiments possibly
due to subtle effects or epistatic interactions.
Using natural variation to discover novel path-
ways of developmental regulation
Sometimes the laboratory strains may be selected based
on certain features such as shorter life cycle. However,
this might sometime lead to selection of certain types
of strains that may not be a good choice for studying all
the developmental genetic pathways. I will illustrate
this using an example in flowering time regulation in
Arabidopsis thaliana. The floral transition is one of
QTL Mapping in understanding Plant Development 435
the crucial developmental decisions a plant has to make
during its life cycle. The timing of the floral transition
has direct consequences on the fitness of a plant since
the plant is exposed to different environmental condi-
tions such as light, temperature, precipitation etc. For
example, it will be detrimental for a plant to make the
floral transition in the peak of a winter. Therefore plants
have evolved diverse mechanisms to control the tim-
ing of floral transition. Genetic analyses have impli-
cated more than 80 loci in flower development (Levy &
Dean 1998). Most of this information comes from clas-
sical analysis of induced mutants using the commonly
used laboratory strains Landsberg erecta (Ler) and
Columbia (Col) (Koornneef et al. 1998). Both these
strains have a life cycle lasting about 8-10 weeks, are
relatively easy to transform and a vast literature has
accumulated on these strains in which thousands of
genetic and molecular experiments have been carried
out. Hundreds of naturally occurring wild strains
of Arabidopsis, referred to, as “accessions” are avail-
able from almost the entire northern hemisphere.
Arabidopsis plants can exist either as a rapid-cycler (to
which Ler and Col belong) or as a winter annual. The
winter annuals require exposure to cold temperatures
for prolonged periods for the promotion of flowering.
This process known as vernalization is an important
floral promoting process, not only in Arabidopsis, but
also in some important crop plants. The commonly used
laboratory strains Ler or Col do not require vernaliza-
tion for floral promotion and therefore the genetic ba-
sis of vernalization requirement was not possible to
study in these strains. Using accessions that require
vernalization, QTL mapping studies have shown that
the vernalization requirement segregates as a mono-
genic trait (Clarke & Dean 1994, Clarke et al. 1995). Sub-
sequent cloning and molecular analysis have shown
that this vernalization requirement is caused by the
FRIGIDA locus, for which Ler and Col carry loss of
function alleles (Johanson et al. 2000) and J. Lempe, S.
Balasubramanian and D. Weigel manuscript in prepa-
ration). Subsequent analysis on other accessions has
shown that two deletions that are present at the FRI
locus in Ler and Col are very common among the natu-
ral populations of Arabidopsis (Johanson et al. 2000).
FRI activates the transcription of a MADS domain con-
taining transcription factor FLOWERING LOCUS C
(FLC), which is a potent floral repressor (Michaels &
Amasino 1999, Sheldon et al. 1999). Vernalization
causes epigenetic modifications at the FLC locus that
subsequently leads to reduction in the expression lev-
els of FLC and relieves the floral repression caused by
FLC. Recent studies have also demonstrated that hav-
ing a non-responsive FLC allele can lead to a rapid-
cycling behavior. Such attenuated alleles are also
present in natural populations (Gazzani et al. 2003,
Michaels et al. 2003) and J.L., S.B and D.W ., manu-
script in preparation) . These developmental mecha-
nisms would have been difficult to identify without the
using natural variation as a genetic resource. Currently
we are aware that vernalization and photoperiod are
the two major floral promoting pathways studied in
detail in Arabidopsis thaliana. Out of the two, our
knowledge on the vernalization pathway essentially
comes from the use of QTL mapping and natural varia-
tion. There is a considerable effort currently to under-
stand natural variation in flowering time in Arabidopsis
thaliana and also to explore this genetic diversity to
understand complex responses such as thermoregula-
tion.
Use of natural variation as a classical genetic resource
Naturally existing variation can also be exploited as a
classical genetic resource. As discussed earlier, the phe-
notypes of the induced mutations are often very much
dependent on the genetic background. This change in
the phenotype occurs because of modifiers of the phe-
notype that are present in one or other genetic back-
grounds. Taking advantage of this, one can use the natu-
ral genetic variation to identify genetic interactors of
known genes as well as to identify novel genes that may
not be detectable in the genetic background in which
the mutation is created. I will discuss one example in
Arabidopsis, where such a classical genetic approach
was used successfully on natural populations to iden-
tify novel genetic modifiers of an induced mutant phe-
notype. apetala 1 (ap1) mutants of Arabidopsis show
defects in conversion of the inflorescence meristem to
the floral meristem (Irish & Sussex 1990). Furthermore,
AP1 functions as an “A” class gene contributing to
the organ identity in whorl 1 and 2 of the Arabidopsis
flower (Coen & Meyerowitz, 1991). AP1 encodes a
MADS domain containing transcription factor (Mandel
et al. 1992). By crossing the ap1 mutant in the Ler ge-
netic background to Wassilewskija (Ws) accession of
Arabidopsis, Bowman et al. (1993) identified a strong
recessive enhancer of the ap1 phenotype (Bowman et
al. 1993). This enhancement of the phenotype was spe-
cific to the Ws accession. This enhancer is due to loss-
of-function changes at the CAULIFLOWER locus,
which encodes a MADS domain containing transcrip-
tion factor that shows extensive similarity to AP1 and
functions redundantly with AP1 in maintaining floral
meristem identity (Kempin et al. 1995). Plants that are
3 Near Isogenic Lines differ only at the QTL region and all other parts of the genome remain the same as one of the parental lines.
4 Heterochronic regulatory mutations are mutations that cause alterations in the temporal expression pattern of a gene.
436 Sureshkumar Balasubramanian
double mutants for both ap1 and cal show a “cauli-
flower” like phenotype, which is a result of the failure
to convert the inflorescence meristem to floral meristem.
Subsequently each one of the primordia that is initiated
behaves as an inflorescence meristem that results in a
cauliflower appearance. This is a classic example that
shows how natural variation can be exploited to iden-
tify genetic modifiers of known mutants and enhance
our understanding of plant development.
Novel genes can be identified using natural
variants
Apart from aiding in the identification of genetic modi-
fiers, natural variants can help us identify previously
unknown genes and contribute to our understanding
of developmental processes. I will illustrate this using
an example from root development in Arabidopsis
thaliana.
Arabidopsis thaliana accessions show differences in
their root morphology and growth. By surveying
Arabidopsis accessions Mouchel et al. (2004) recog-
nized an accession from Umkirch (Uk-1) had very dis-
tinct root morphology with a short primary root and
more adventitious roots. In order to determine the mo-
lecular basis of this phenotype, Mouchel et al analyzed
an F2 population from cross between Uk-1 and Sav-0,
an accession from Slavice. The F2 plants showed a
rough recessive monogenic segregation of the Uk-1 root
phenotype. Mouchel et al. further went ahead and gen-
erated a RIL population from this cross and performed
a QTL analysis and figured that a single QTL can ex-
plain up to 80% of phenotypic variation that exists in
this population. Mouchel et al. cloned this locus
BREVIS RADIX (BRX) and showed that a mutation
causing a stop codon at the first exon of BRX is respon-
sible for the short primary root phenotype in Uk-1. Se-
quence analysis suggested that BRX belongs to a pre-
viously unidentified plant specific family of proteins.
This is one recent example, where a QTL based ap-
proach actually resulted in identifying a novel gene that
contributes to naturally occurring phenotypic variation.
Natural variation helps identify novel functions of
known genes
Analysis of natural variation can give novel insights
into the functions of even known genes. Mendelian ge-
netics relies heavily on mutations that have drastic phe-
notypes that are easy to score. Smaller effect mutations
are harder to detect and often missed in a screen. How-
ever, such small effect mutations alter the functions of
the gene products in subtle ways that gives a selective
advantage to the organism in its natural environment.
Such novel altered functions can be detected by exploit-
ing natural variation.
Arabidopsis is a facultative long day plant. Long
days (16 hr light, 8 hr dark) promote flowering and
short days (8hr light, 16 hr dark) delay flowering. Cvi,
an accession of Arabidopsis collected in the Cape Verde
Islands (15 º Northern Latitude), shows a relatively
early flowering phenotype even under short-day con-
ditions (Alonso-Blanco et al. 1998). By performing a
QTL mapping analysis, a QTL was mapped in that pro-
vides an Early Flowering Day Length Insensitive (EDI)
phenotype (Alonso-Blanco et al. 1998). A NIL carrying
the EDI locus of Cvi shows almost no sensitivity to
photoperiod (Alonso-Blanco et al. 1998). This QTL was
subsequently cloned and the EDI phenotype was
caused by an unusual allele at the
CRYPTOCHROME2 (CRY2) locus (El-Assal et al.
2001). Plants that are mutant for cry2 do not show an
early flowering under short day phenotype. Then, how
does the Cvi allele causes early flowering under short
days? Using biochemical and molecular methods,
Koornneef`s laboratory demonstrated that a single
amino acid change in the Cvi accession, leads to a more
stable CRY2 protein that otherwise would get degraded
quickly under short day conditions (El-Assal et al.
2001). While the regular function of CRY2 was known
from traditional genetic and molecular analysis, such
altered specially evolved functions contributing to
major developmental decisions can be studied using
natural variation.
Novel approaches expedite the understanding
of naturally occurring novel alleles
Recent advances in genomic and computational meth-
ods expedite the process of identifying the QTL. I will
illustrate this using an example that provided some in-
sights into one of most complex interactions that con-
tribute to plant development.
Light is a major factor that regulates almost every
aspect of plant development (Chory 1997, Fankhauser
& Chory 1997). By screening Arabidopsis accessions
for the response to light as measured by the length of
their hypocotyls after germination, Maloof et al. iden-
tified extensive variation among Arabidopsis acces-
sions (Maloof et al. 2001). Using an original approach,
Maloof et al. were able to identify a natural variant that
controls the light sensitivity in the Lm-2 accession of
Arabidopsis. Using the phenotypic measurements of
the accessions and mutants that show a light sensitive
phenotype under multiple environments, Maloof et al.
used hierarchical clustering to see whether any of the
accessions clustered together with known mutants.
Clustering techniques are used regularly to group the
genes based on their expression levels in microarray
experiments. Here, Maloof et al. used the phenotypic
QTL Mapping in understanding Plant Development 437
values and tried to cluster the accessions on the basis
of the phenotypes under multiple environments. In-
terestingly, the Lm-2 accession clustered together with
the phyA mutant of Arabidopsis. Both Lm-2 and phyA
mutants have a longer hypocotyls compared to the ref-
erence strain Col. Using genetic, biochemical and mo-
lecular methods, Maloof et al. were able to demonstrate
that a single amino acid change in the Lm-2 PHYA in-
creased the stability of the light labile PHYA protein
(Maloof et al. 2001).
Natural variation can be used to understand
complex G X E interactions
Analysis of natural variation can also shed some light
on how the genotype interacts with its environment. It
is very well accepted that plants do exhibit a signifi-
cant GXE interaction. GXE interaction refers to the dif-
ferential phenotypic responses that occur as a result of
the environmental effects that differ from one genotype
to another genotype. Due to the complexity of the na-
ture of such interactions, it has been difficult to under-
stand these interactions even today. However, steady
progress is being made that will lead us to have a bet-
ter understanding of this interaction. I will discuss one
example here.
Since QTL mapping in itself is a genomic approach,
it takes into account of the interactions that occur
throughout the genome. Many genetic loci have been
isolated and mapped that regulate floral transition in
Arabidopsis thaliana. However, when a QTL mapping
exercise is carried out for identifying QTLs that regu-
late floral transition, several regions in the genome
where no known mutations have been mapped appear.
Interestingly enough, when the same QTL mapping
experiment is carried out under different environmen-
tal conditions different QTLs can be detected. Weinig
et al performed a QTL study for floral transition under
a natural environment (in the field), which showed that
novel loci regulate this reproductive timing under natu-
ral conditions. Furthermore, many of the QTLs detected
showed significant GXE interactions. (Ungerer et al.
2003, Weinig et al. 2002). These works demonstrated
that there are other loci involved and there are signifi-
cant GXE interactions that occur during plant develop-
ment.
In a screen for flowering time we have screened more
than hundreds of Arabidopsis accession under multiple
environments. By comparing the specific flowering
times of individual accessions among environments we
were able to show that the Arabidopsis accessions dif-
fer extensively in their flowering response to photope-
riod, vernalization and ambient growth temperature.
We are able to show that the accessions exhibit highly
significant GXE interactions for their flowering re-
sponse (S.B., J.L and D.W., manuscript in preparation).
By analyzing the responses to environment, we were
able to identify accessions that show a significant GXE
response, which can then be used to understand the
molecular basis for such phenotypic responses.
Natural Variation and QTL mapping in provid-
ing insights into evolution of developmental
mechanisms
Identification of genes that are the underlying cause for
a QTL can also provide insights into evolutionary ad-
aptation mechanisms. I will discuss this issue first by
Figure 3 A schematic representation of the salient features of using natural variation and QTL mapping for various studies: Examples for
each of those features is discussed in the review.
438 Sureshkumar Balasubramanian
comparing Arabidopsis with rice in their photoperiodic
flowering response and then by comparing Arabidopsis
and wheat in vernalization response.
Rice is a monocot and Arabidopsis is a dicot and the
two diverged from each other approximately 150 Myrs
ago. QTL mapping in rice enabled the identification of
the molecular players involved in processes that are
studied well in Arabidopsis. Comparison of these mol-
ecules provides insights into adaptive developmental
mechanisms.
As discussed earlier, timing of floral transition is an
important developmental decision in a plant’s life
cycle. Arabidopsis is a facultative long day plant where
as rice is a short day plant. In Arabidopsis long days
promote floral transition. Isolation and characterization
of mutants that show a late flowering phenotype spe-
cifically under long day conditions in Arabidopsis led
to the identification of a major molecular regulatory
factor CONSTANS, which promotes flowering in re-
sponse to light (Putterill et al. 1995). CO is regulated
both transcriptionally as well as post-transcriptionally
at the protein level by light (Imaizumi et al. 2003,
Suarez-Lopez et al. 2001, Valverde et al. 2004, Yanovsky
& Kay 2002). CO encodes a putative transcription fac-
tor (Putterill et al. 1995), which essentially mediates its
effect on flowering by promoting the transcription of a
major integrator of floral transitional signals FLOW-
ERING LOCUS T (FT) (Kardailsky et al. 1999,
Kobayashi et al. 1999). QTL mapping and subsequent
cloning of the major effect QTLs for flowering time in
rice have shown there are at least 14 QTLs that control
floral transition in rice. One of the major effect QTLs,
HEADING DATE 1 (Hd1), is an orthologue of the
Arabidopsis gene CO (Yano et al. 2000). In contrast to
Arabidopsis, Hd1 in rice delays flowering in response
to long days and promotes flowering in response to
short days. Another QTL Hd3a is an orthologue of the
FT gene of Arabidopsis (Kojima et al. 2002). Therefore
the major players seem to be conserved between rice
and Arabidopsis. However, they result in an opposite
response in rice in response to long day conditions.
Furthermore, over expression of the orthologue of
GIGANTEA, an upstream activator of CO in
Arabidopsis leads to delayed flowering in rice (Hayama
et al. 2003). While this work gives a clear answer to how
the flowering is inhibited in rice under long day condi-
tions, the promotion of molecular basis of the promo-
tion of flowering under short days by the CO
orthologue in rice remains enigmatic. These experi-
ments suggest that the major players are adapted in rice
to perform functions that are suitable for rice (Simpson
2003). The CO orthologue of rice confers repression of
FT under long day conditions, while CO of
Arabidopsis promotes the expression of FT. However,
the function of FT seems to be conserved in both rice
and Arabidopsis (Kojima et al. 2002). Thus QTL map-
ping in rice enabled us to understand not only the regu-
lation of floral transition in rice, but also the adapta-
tion of pathways in evolution in correspondence with
the natural environment of a plant.
Similar to Arabidopsis, accessions of wheat differ in
their requirement for vernalization. Winter cultivars of
wheat flower after vernalization compared to the spring
varieties. Genetic analysis using the accessions that dif-
fer in their vernalization response, suggested two loci
to be the major players modulating vernalization re-
quirement in the winter and spring cultivars of wheat.
These loci are referred to as the VRN1 and VRN2,
which are different from the genes with the same names
in Arabidopsis. The cloning of VRN1 revealed that
VRN1 encodes a MADS domain transcription factor
with similarities to AP1 of Arabidopsis (Yan et al.
2003). In Arabidopsis, AP1 is involved mainly in the
promotion and maintenance of the floral meristem iden-
tity. Recently, the molecular nature of VRN2 has also
been revealed through positional cloning. Surprisingly,
VRN2 encodes a protein that contains a domain that
shows similarities to CO, CO-like and the TIMING
OF CAB EXPRESSION (TOC1) genes of
Arabidopsis (Yan et al. 2004). A single amino acid sub-
stitution in VRN2 at a conserved region was detected
in the spring cultivar of wheat. Further screening of
more than 40 wheat accessions showed that amino acid
substitution or the deletions at VRN2 or at the VRN1
promoter was sufficient to explain the spring or winter
habit of most of the wheat cultivars (Yan et al. 2004).
While both Arabidopsis and wheat have strains that
require vernalization, the pathway seems to involve
very different players. It is interesting to note that tem-
perate cereals and Arabidopsis have evolved different
mechanisms to finally achieve a similar phenotypic re-
sponse. Thus, studying naturally existing variation can
provide interesting insights into developmental mecha-
nisms in an evolutionary scale.
Challenges for the future
A summary of the salient features of natural variation
and QTL mapping is given in Figure 3. The advent of
molecular markers, efficient statistical and computa-
tional methods have revolutionized our views on QTLs
and quantitative genetics. Today we can think about
identifying QTLs in a much more easy manner com-
pared to the last decade. Steps are being taken even to
further our understanding of the molecular basis of the
detected QTLs. Quantitative Trait Genes (QTG) and the
QTL Mapping in understanding Plant Development 439
Quantitative Trait Nucleotides (QTNs) are identified
at an increasing speed in a variety of organisms.
It will discuss in a couple of areas where one can see
increase in our understanding of plant development.
While there is a lot of knowledge accumulated on play-
ers involved in different developmental pathways,
there is only emerging information on how the envi-
ronment modulates the developmental pathways at a
genetic level. The GXE interactions have been difficult
to study and the advances in QTL mapping and
genomics should help enhance our knowledge on this
area of plant research.
Another developmental pathway that has not been
studied extensively is seed dormancy and seed germi-
nation (Koornneef et al. 2002). There are some initial
QTL mapping results that are beginning to emerge and
one would expect to see a huge progress on this area of
plant development in the near future. Already a few
QTLs that regulate seed dormancy have been identi-
fied and the molecular cloning of these QTLs is under
progress (Alonso-Blanco et al. 2003). Identification of
the molecular players will greatly improve our under-
standing on this life history trait.
A third area of research that is slowly emerging is
the study of thermoregulation of plant development.
Light and temperature are the two major factors that
affect almost all parts of a plant`s life cycle. However,
how the ambient growth temperature affects any of the
developmental processes has received very little atten-
tion since the responses to temperature are very com-
plex. Recent studies have shown how temperature or-
chestrates its responses with light in the regulation of
the circadian clock (Michael et al. 2003). There is a lot
of interest in understanding thermoregulation of flow-
ering in plants. QTL mapping using natural popula-
tions should really help our understanding on these
complex responses during plant development.
With the genomic approaches becoming increasingly
available, there is a lot of interest in integrating all the
responses in a systems biology approach. Scientists are
very much interested in linking the genotype at a ge-
nomic level to multiple phenotypes or the so-called
phenome-genome associations. QTL mapping, inher-
ently a genomic approach will aid in these kinds of
whole genome association studies.
In this review, I discussed only a selected few ex-
amples to emphasize what we can learn using natural
variation and QTL mapping in plant development. QTL
mapping and natural variation extend well beyond
plant development and play a vital role in several other
quantitative traits such as yield, disease resistance,
metal tolerance, stress responses etc. Compared to the
Mendelian genetic approaches, which are easily achiev-
able in model organisms, QTL mapping and natural
variation studies can be done on organisms that are not
amenable to mutational analysis including humans.
Acknowledgements
I thank Detlef Weigel, Phil Wigge and Yasushi
Kobayashi for comments on the manuscript. I thank the
anonymous reviewers for extremely useful comments
on the manuscript. I would like to specially thank Detlef
Weigel, in whose laboratory I have been doing my work
on natural variation in Arabidopsis thaliana. I thank
financial support from European Molecular Biology
Organization (EMBO-LTF 473/2002) and Max-Planck
Society (MPG Auslander Stipendium).
.
References
Alonso-Blanco C, Bentsink L, Hanhart C J, Blankestijn-de
Vries, H. and Koornneef M (2003) Analysis of natural
allelic variation at seed dormancy loci of Arabidopsis
thaliana; Genetics 164 711-729
------, El-Assal S E, Coupland G and Koornneef M (1998)
Analysis of natural allelic variation at flowering time
loci in the Landsberg erecta and Cape Verde Islands
ecotypes of Arabidopsis thaliana; Genetics 149 749-764
------ and Koornneef M (2000) Naturally occurring variation
in Arabidopsis: an underexploited resource for plant
genetics; Trends Plant Sci. 5 22-29
Borevitz J O and Nordborg M (2003) The impact of genomics
on the study of natural variation in Arabidopsis. Plant
Physiol. 132 718-725
Bowman J L, Alvarez J, Weigel D, Meyerowitz E M and
Smyth D R (1993) Control of flower development in
Arabidopsis thaliana by APETALA1 and interacting
genes. Development 119 721-743
Broman K W (2001) Review of statistical methods for QTL
mapping in experimental crosses; Lab Anim (NY) 30
44-52
Chory J (1997) Light modulation of vegetative development;
Plant Cell 9 1225-1234
Clark R M, Linton E, Messing J and Doebley J F (2004) Pat-
tern of diversity in the genomic region near the maize
domestication gene tb1; Proc. Natl. Acad. Sci. USA
101 700-707
Clarke J H and Dean C (1994) Mapping FRI, a locus control-
ling flowering time and vernalization response in
Arabidopsis thaliana; Mol. Gen. Genet. 242 81-89
440 Sureshkumar Balasubramanian
------, Mithen R, Brown J K and Dean C (1995) QTL analysis
of flowering time in Arabidopsis thaliana; Mol. Gen.
Genet. 248 278-286
Coen E S and Meyerowitz E M (1991) The war of the whorls:
genetic interactions controlling flower development;
Nature 353 31-37
Cong B, Liu J and Tanksley S D (2002) Natural alleles at a
tomato fruit size quantitative trait locus differ by
heterochronic regulatory mutations; Proc. Natl. Acad.
Sci. USA 99 13606-13611
Doebley J and Stec A (1991) Genetic analysis of the morpho-
logical differences between maize and teosinte; Genet-
ics 129 285-295
------ and ------ (1993) Inheritance of the morphological dif-
ferences between maize and teosinte: comparison of
results for two F2 populations; Genetics 134 559-570
------, ------ and Hubbard L (1997) The evolution of apical
dominance in maize; Nature 386 485-488
Doerge R W and Churchill G A (1996) Permutation tests for
multiple loci affecting a quantitative character; Genet-
ics 142 285-294
El-Din El-Assal S, Alonso-Blanco C, Peeters A J, Raz V and
Koornneef M (2001) A QTL for flowering time in
Arabidopsis reveals a novel allele of CRY2; Nat. Genet.
29 435-440
Fankhauser C and Chory J (1997) Light control of plant de-
velopment. Annu Rev Cell Dev Biol. 13 203-229
Flint-Garcia S A, Thornsberry J M and Buckler E S T (2003)
Structure of linkage disequilibrium in plants; Annu.
Rev. Plant Biol. 54 357-374
Frary A, Nesbitt T C, Grandillo S, Knaap E, Cong B, Liu J,
Meller J, Elber R, Alpert K B and Tanksley S D (2000)
fw2.2: a quantitative trait locus key to the evolution of
tomato fruit size; Science 289 85-88
Gazzani S, Gendall A R, Lister C and Dean C (2003) Analy-
sis of the molecular basis of flowering time variation in
Arabidopsis accessions; Plant Physiol. 132 1107-1114
Hayama R, Yokoi S, Tamaki S, Yano M and Shimamoto K
(2003) Adaptation of photoperiodic control pathways
produces short-day flowering in rice; Nature 422 719-
722
Hoeschele I, Uimari P, Grignola F E, Zhang Q and Gage K M
(1997) Advances in statistical methods to map quanti-
tative trait loci in outbred populations; Genetics 147
1445-1457
Hubbard L, McSteen P, Doebley J and Hake S (2002) Expres-
sion patterns and mutant phenotype of teosinte
branched1 correlate with growth suppression in maize
and teosinte; Genetics 162 1927-1935
Imaizumi T, Tran H G, Swartz T E, Briggs W R and Kay S A
(2003) FKF1 is essential for photoperiodic-specific light
signalling in Arabidopsis. Nature 426 302-306
Irish V F and Sussex I M (1990) Function of the apetala-1
gene during Arabidopsis floral development; Plant Cell
2 741-753
Johanson U, West J, Lister C, Michaels S, Amasino R and
Dean C (2000) Molecular analysis of FRIGIDA, a ma-
jor determinant of natural variation in Arabidopsis
flowering time; Science 290 344-347
Kardailsky I, Shukla V K, Ahn J H, Dagenais N, Christensen
S K, Nguyen J T, Chory J, Harrison M J and Weigel D
(1999) Activation tagging of the floral inducer FT; Sci-
ence 286 1962-1965
Kempin S A, Savidge B and Yanofsky M F (1995) Molecular
basis of the cauliflower phenotype in Arabidopsis; Sci-
ence 267 522-525
Kobayashi Y, Kaya H, Goto K, Iwabuchi M and Araki T
(1999) A pair of related genes with antagonistic roles in
mediating flowering signals; Science 286 1960-1962
Kojima S, Takahashi Y, Kobayashi Y, Monna L, Sasaki T,
Araki T and Yano M (2002) Hd3a, a rice ortholog of the
Arabidopsis FT gene, promotes transition to flowering
downstream of Hd1 under short-day conditions; Plant
Cell Physiol. 43 1096-1105
Koornneef M, Alonso-Blanco C, Blankestijn-de Vries H,
Hanhart C J and Peeters A J (1998) Genetic interactions
among late-flowering mutants of Arabidopsis; Genet-
ics 148 885-892
------, Alonso-Blanco, C. and Vreugdenhil D (2004) Naturally
occurring Genetic Variation in Arabidopsis Thaliana;
Annu. Rev. Plant Physiol. Plant Mol. Biol. 55 141-
172
------, Bentsink L and Hilhorst H (2002) Seed dormancy and
germination; Curr. Opin. Plant Biol. 5 33-36
Levy Y Y and Dean C (1998) The transition to flowering;
Plant Cell 10 1973-1990
Liu J, Van Eck J, Cong B and Tanksley S D (2002) A new
class of regulatory genes underlying the cause of pear-
shaped tomato fruit. Proc Natl Acad. Sci. USA 99
13302-13306
Luo D, Carpenter R, Vincent C, Copsey L and Coen E (1996)
Origin of floral asymmetry in Antirrhinum; Nature
383 794-799
Ma C X, Casella G and Wu R (2002) Functional mapping of
quantitative trait loci underlying the character process:
a theoretical framework. Genetics 161 1751-1762
------, Wu R and Casella G (2004) FunMap: functional map-
ping of complex traits; Bioinformatics
Maloof J N, Borevitz J O, Dabi T, Lutes J, Nehring R B,
Redfern J L, Trainer G T, Wilson J M, Asami T, Berry C
C, Weigel D and Chory J (2001) Natural variation in light
sensitivity of Arabidopsis; Nat. Genet. 29 441-446
QTL Mapping in understanding Plant Development 441
Mandel M A, Gustafson-Brown C, Savidge B and Yanofsky
M F (1992) Molecular characterization of the
Arabidopsis floral homeotic gene APETALA1; Nature
360 273-277
Michael T P, Salome P A and McClung C R (2003) Two
Arabidopsis circadian oscillators can be distinguished
by differential temperature sensitivity; Proc. Natl.
Acad. Sci. USA 100 6878-6883
Michaels S D and Amasino R M (1999) FLOWERING LO-
CUS C encodes a novel MADS domain protein that acts
as a repressor of flowering; Plant Cell 11 949-956
------, Bezerra I C and Amasino R M (2004) FRIGIDA-re-
lated genes are required for the winter-annual habit in
Arabidopsis; Proc. Natl. Acad. Sci. USA 101 3281-
3285
Michaels S D, He Y, Scortecci K C and Amasino R M (2003)
Attenuation of FLOWERING LOCUS C activity as a
mechanism for the evolution of summer-annual flow-
ering behavior in Arabidopsis; Proc. Natl. Acad. Sci.
USA 100 10102-10107
Nordborg M and Tavare S (2002) Linkage disequilibrium:
what history has to tell us; Trends Genet. 18 83-90
Putterill J, Robson F, Lee K, Simon R and Coupland G (1995)
The CONSTANS gene of Arabidopsis promotes flow-
ering and encodes a protein showing similarities to zinc
finger transcription factors; Cell 80 847-857
Satagopan J M, Yandell B S, Newton M A and Osborn T C
(1996) A bayesian approach to detect quantitative trait
loci using Markov chain Monte Carlo; Genetics 144 805-
816
Sheldon C C, Burn J E, Perez P P, Metzger J, Edwards J A,
Peacock W J and Dennis E S (1999) The FLF MADS box
gene: a repressor of flowering in Arabidopsis regulated
by vernalization and methylation; Plant Cell 11 445-
458
Simpson G G (2003) Evolution of flowering in response to
day length: flipping the CONSTANS switch; Bioessays
25 829-832
Suarez-Lopez P, Wheatley K, Robson F, Onouchi H, Valverde
F and Coupland G (2001) CONSTANS mediates be-
tween the circadian clock and the control of flowering
in Arabidopsis; Nature 410 1116-1120
Ungerer M C, Halldorsdottir S S, Purugganan M D and
Mackay T F (2003) Genotype-environment interactions
at quantitative trait loci affecting inflorescence devel-
opment in Arabidopsis thaliana; Genetics 165 353-365
Valverde F, Mouradov A, Soppe W, Ravenscroft D, Samach
A and Coupland G (2004) Photoreceptor regulation of
CONSTANS protein in photoperiodic flowering; Sci-
ence 303 1003-1006
Weinig C, Ungerer M C, Dorn L A, Kane N C, Toyonaga Y,
Halldorsdottir S S, Mackay T F, Purugganan M D and
Schmitt J (2002) Novel loci control variation in repro-
ductive timing in Arabidopsis thaliana in natural en-
vironments; Genetics 162 1875-1884
White S and Doebley J (1998) Of genes and genomes and the
origin of maize; Trends Genet. 14 327-332
Wu R, Ma C X, Lin M and Casella G (2004) A general frame-
work for analyzing the genetic architecture of develop-
mental characteristics; Genetics 166 1541-1551
Yan L, Loukoianov A, Blechl A, Tranquilli G, Ramakrishna
W, SanMiguel P, Bennetzen J L, Echenique V and
Dubcovsky J (2004) The wheat VRN2 gene is a flower-
ing repressor down-regulated by vernalization; Science
303 1640-1644
Yan L, Loukoianov A, Tranquilli G, Helguera M, Fahima T
and Dubcovsky J (2003) Positional cloning of the wheat
vernalization gene VRN1; Proc. Natl. Acad. Sci. USA
100 6263-6268
Yano M, Katayose Y, Ashikari M, Yamanouchi U, Monna L,
Fuse T, Baba T, Yamamoto K, Umehara Y, Nagamura Y
and Sasaki T (2000) Hd1, a major photoperiod sensitiv-
ity quantitative trait locus in rice, is closely related to
the Arabidopsis flowering time gene CONSTANS;
Plant Cell 12 2473-2484
Yanovsky M J and Kay S A (2002) Molecular basis of sea-
sonal time measurement in Arabidopsis; Nature 419
308-312
