an allelic series of mutants that permits a refined
analysis of gene function, in contrast to knockout
mutants. The mutagens most often used, because
they are effective in most species, are chemical and
those that generate single or few nucleotide
changes, such as ethylmethane sulphonate (EMS),
methyl nitrosourea or sodium azide. The last has
often been used with cereals since it has been
found to be more effective than the alkylating
agents. The effect of the mutagens is random
which is important for covering the genome, but is
restricted to certain transitions (G/C to A/T for
alkylating agents and A/T to G/C for azide) (Olsen
et al. 1993 ; van Arten 1998 ).
21.2 History of Lotus TILLING
For all the reasons given above, TILLING was
deployed to advanceLotus japonicusas a model
for legume biology and especially the rhizo-
bium–legume symbiosis. At the time the ATP
was being established, Martin Parniske at the
Sainsbury Laboratory, Norwich, UK, decided to
raise a population ofL. japonicusGifu EMS-
treated plants to isolate nodulation mutants and
develop the TILLING method for this species.
He was joined by groups working on legumes at
the adjacent John Innes Centre, and this collab-
oration established the Lotus TILLING Platform
(LTP), thefirst TILLING platform outside the
USA. The platform started operation in 2003 and
now operates under the RevGenUK banner
(http://revgenuk.jic.ac.uk/). It was originally
established for Lotus alone, but now includes
Medicago truncatula, brassicas and cereals. The
unique feature of LTP was that the mutagenized
populations included forward screens. This not
only helped to identify mutants of interest, but it
also provided a measure of how well the popu-
lation was mutagenized, in terms of mutation
density and saturation. The initial platform was
based on a Gifu population (Perry et al. 2003 ,
2009 ), but has since been supplemented with one
in the MG20 background.
21.3 The Mutagenized PopulationsThe detailed structures of the populations (total-
ling nearly 9,000 individuals) are presented in
Fig.21.1. Extensive data have already been pro-
vided on the Gifu population (Fig.21.1a) in Perry
et al. ( 2009 ). Essentially, three forward (thematic)
screens were conducted: one for nodulation
(nodule morphology, nodulation ability and root
morphology; NODPOP), one for general plant
morphology (by gross observation; DEVPOP)
and one on leaf starch accumulation (by iodine
staining of alcohol-cleared leaves at the end of
the day and end of the night) (Vriet et al. 2010 ).
DNA was collected from each plant in the screens
to use for TILLING since they represent specific
mutant-enriched subpopulations. Single plants
from each family were also used to collect leaves,
extract DNAs, harvest seeds and form a general
TILLING population (GENPOP). For the MG20
population (Fig.21.1a), further forward screens
were carried out. Firstly, to add to the nodulation
population by screening for nodule production
and nodule function (nod−/fix−), and to add to
GENPOP. In addition, two further screens were
completed: one for the ability of leaves to gen-
erate cyanide which was highly successful in
identifying cyanogenic glucoside mutants (Takos
et al. 2010 ) and an unsuccessful one based on the
well-established method of inducing isoflavo-
noids by applying glutathione (Shelton et al.
2012 ). In the latter screen, leaves were placed in
glutathione-containing induction medium and the
fluorescence of the medium measured after 24 h.
It was hypothesised that a lack offluorescence
would indicate an inability to produce isoflavo-
noids and increasedfluorescence, an overpro-
duction. Although 61 mutant families showing
fluorescence above or below the background level
were isolated (Fig.21.1b), none proved to be
mutant on further screening (Takos et al. 2011 ).
Two further populations were also generated.
Seeds from all the plants in a selection of ca. 2000
Gifu families were bulked and retained as a
backup population. Several groups have used this230 T.L. Wang and F. Robson