Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter.
Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 75
OCR for page 77
OCR for page 78
OCR for page 79
OCR for page 80
Representative terms from entire chapter:
insect resistance
Colloquium
Evolutionary dynamics of an Arabidopsis insect
resistance quantitative trait locus
Juergen Kroymann, Susanne Donnerhacke, Domenica Schnabelrauch, and Thomas Mitchell-Olds*
Department of Genetics and Evolution, Max Planck Institute for Chemical Ecology, Winzerlaer Strasse 10, 07745 Jena, Germany
Glucosinolate profiles differ among Arabidopsis thaliana ecotypes,
caused by the composition of alleles at several glucosinolate
biosynthetic loci. One of these, GS-Elong, harbors a family of
methylthioalkylmalate synthase (MAM) genes that determine the
side chain length of aliphatic glucosinolate structures. Fine map-
ping reveals that GS-Elong constitutes an insect resistance quan-
titative trait locus, caused by variation in glucosinolate profiles
conferred by polymorphism of MAM alleles In this region. A
sequence survey of randomly chosen ecotypes indicates that GS-
Elong is highly variable among A. thaliana ecotypes: indel poly-
morphisms are frequent, as well as gene conversion events be-
tween gene copies arranged in tandem. Furthermore, statistical
methods of molecular population genetics suggest that one of the
genes, MAM2, is subject to balancing selection. This may be caused
by ecological tradeoffs, i.e., by contrasting physiological effects of
glucosinolates on generalist vs. specialist insects.
Resistance to insect herbivores is genetically variable in many
plant populations (1~. Ecological and evolutionary interac-
tions between host plants and their insect enemies are often
mediated by secondary metabolites. Molecular genetics allows
us to clone and characterize genes controlling secondary me-
tabolism and insect resistance. When these genes are identified,
molecular population genetics provides statistical tests for neu-
tral evolution or natural selection, thus elucidating the evolu-
tionary forces responsible for genetic variation in ecologically
important traits. Here we show that a gene family involved in the
synthesis of secondary metabolites controls insect resistance,
shows complex molecular variation, and maintains excess amino
acid polymorphism because of balancing natural selection.
Glucosinolates are amino acid-derived plant secondary com-
pounds present in the Capparales (2, 3~. Their biosynthesis
occurs in three independent stages, chain elongation of the
amino acid, formation of the core structure (consisting of a
,B-thioglucose moiety and a sulfonated oxime), and side chain
modifications. Both side chain elongation and modification
contribute to the variation of glucosinolate structures, and >30
different glucosinolates have been identified in the model plant
Arabidopsis thaliana (~4, 5~. However, A. thaliana ecotypes vary
extensively both in their glucosinolate composition and quantity
(4~. Genetically, most of this natural variation can be explained
by the combination of alleles at five genetic loci within the A.
thaliana genome (4~. Among these, GS-Elong has a central role
as it controls side chain length of methionine-derived glucosi-
nolates, thereby determining potential for subsequent modifi-
cation steps (6, 7~. GS-Elong consists of a small gene family
encoding methylthioallylmalate synthase (MAM) enzymes re-
sponsible for carbon chain elongation in glucosinolate biosyn-
thesis. Leaves of the Columbia (Col-O) ecotype of A. thaliana
contain predominantly glucosinolates with four methylene
groups (C4) in their basic side chain. In contrast, Landsberg
erecta (
Growth Rate Measurement. Plants were harvested 16 days after
germination. Plants for each pot (i.e., for each line) were pooled,
dried in a hot air oven at 45°C, and then weighed. For each line, Sorbo
there were 14 or 15 replicates.
Glucosinolate Analysis. Leaves were harvested 18 days after plants
were moved to the growth rooms, i.e., 15-16 days after germi-
nation. Glucosinolate extraction and high-pressure liquid chro-
matography were carried out as described (7~.
Molecular Methods. Total leaf DNA was extracted with Qiagen
(Hilden, Germany) genomic-tips 100/G following the manufac-
turer's instructions. Total leaf RNA was isolated with Trizol
(Life Technologies). First-strand cDNA was synthesized from 1
,ug total RNA following ref. 9. Details on PCRs for M~37
(=atSg23000), MAM1 (=atSg23010), MAM2, and MAM-L
(=at5g23020) genes are given in the supporting information,
which is published on the PNAS web site, www.pnas.org.
PCR products were gel purified with QIAquick (Qiagen), and
cloned into pCRII TOPO TA, pCR2.1 TOPO TA, or pCR-XL-
TOPO vectors (all from Invitrogen). Plasmids were isolated
according to standard procedures.
Sequences were obtained either directly from gel-purified
PCR products, or from recombinant plasmids. In the latter case,
several independent clones were analyzed with vector- and
insert-specific primers. Sequencing was done on ABI 377 or 3700
DNA sequencers with Big Dye Terminators (Applied Biosys-
tems). Sequence of the entire Ler-0 GS-Elong region was assem-
bled from two overlapping Ler-0 bacterial artificial chromosomes
according to standard methods. Assembly and comparison of
DNA sequence data were carried out with DNASTAR (DNAS-
TAR, Madison, WI).
Genotyping of recombinant inbred lines followed ref. 7,
including additional markers (see supporting information). Re-
sulting PCR products were purified and analyzed by sequencing,
except for those amplified with recMS3f/recMS3r, which were
separated on 6% Metaphor (Biozym, Germany) agarose gels.
Statistical Methods. Calculation of Tajima's D, coalescent simu-
lations, and the McDonald-Kreitman test were carried out with
DNASP3.84 (10~. A neighborjoining tree (11) of MAM1 and
MAM2 alleles was constructed with TREECON (12) following ref.
13. Reliability of the branching order was estimated by boot-
strapping (100 replicates; ref. 14~.
For measurements of glucosinolates and resistance to S. exigua
and P. xylostella, we obtained least square means for each near
isogenic line, analyzing a randomized complete blocks design
with SYSTAT (SPSS, Chicago). Quantitative trait locus (QTL)
mapping with line means used a fixed-effects ANOVA with
markers in the GS-Elong interval cross-classified with the
AlkOhp marker (15), which was also examined in this experiment
(data not shown). Previous studies have shown a QTL influenc-
ing glucosinolate concentration in the GS-Elong region (16~;
hence, in these new experiments, we used a standard P = 0.05
significance threshold based on this a priori expectation.
Results
Complex Organization of GS-Elong in A. thaliana Ecotypes. The
organization of GS-Elong is highly variable in A. thaliana, and
indel polymorphisms of large regions are common (Fig. 1~. In
addition to a MAM-L (MAM-like) gene present in all ecotypes
investigated, GS-Elong may harbor two further loci, MAM1 and
MAM2. Some ecotypes contain both loci, whereas others possess
either MAM2 or MAM1, or, as the Ler-0 ecotype, a MAM2 in
addition to a truncated, nonfunctional MAM1 locus. In the case
of the Lm-2 ecotype, the 5' part of a MAM2-like sequence is
fused to the 3' part of a MAAll-like sequence, which may have
been caused by deletion of the intervening region. Reciprocal
14588 1 www.pnas.org/cgi/doi/ 10.1 073/pnas. 1734046100
ECOTYPE MAM2 MAM1
Cvl-O
Cal-O
Di-1
Condara
Hodja
Mr-O
Col-O
Aa~
Ag4
Gy-0
Mt~
Ema 1
Pla-0
Ler-O
BI-O
Di~
Petergof
Ka~
Lip-O
No-O
Sei~
Tsu-O
WI-O
Lm 2
-
MAM-L PHENOTYPE
4
4
3
3
3
4
4
4
4
4
4
4
3
3
3
3
3
3
3
3
~ 4
Fig. 1. GS-Elong region in A. thaliana ecotypes. Maximal levels of divergence
between MAM1 and MAM2 (nearly S%) occur in the Sorbo ecotype, which
likely represents the ancestral gene arrangement. Genes are patterned as
most similar to Sorbo MAM1 (vertically patterned) or Sorbo MAM2 (horizon-
tally patterned). Triangles indicate large deletions. Predominant glucosino-
late side chain length is indicated on the right: glucosinolates with four
methylene groups (4, e.g,, Col-0) or three methylene groups (3, e.g., Ler-0).
Notice that synthesis of G, glucosinolates is completely associated with occur-
rence of a full-length Sorbo-like MAM1 gene. Otherwise, C3 glucosinolates are
synthesized.
deletions of MAM2 in Col-0 and MAM1 in Ler-0 have the
consequence that the remaining MAM1 and MAM2 genes
segregate as alleles of each other, although they are phyloge-
netically paralogs resulting from a gene duplication event
(Fig. 1~.
6ene Conversion Between MAM1 and MAM2 Loci. An additional
layer of complexity is added by apparent interlocus gene con-
version between MAM1 and MAM2. Pairwise comparisons of
sequence similarity between MAM genes of ecotypes containing
both a MAM2 and a MAM1 locus reveal a maximal divergence
between the Sorbo MAM2 and MAM1 genes; both nucleotide
and protein sequences differ by ~5.0% (Table 1~. Therefore, the
Sorbo ecotype likely represents the basal configuration of MAM
genes at GS-Elong in A. thaliana. In contrast, in other ecotypes
MAM2 and MAM1 show regions of much greater similarity, with
a minimum value of ~1% divergence between paralogous loci in
the Condara and Hodja ecotypes, indicating that genetic infor-
mation was exchanged between MAM2 and MAM1 in these
haplotypes.
Sliding window analyses of nucleotide polymorphism between
MAM1 and MAM2 sequences were performed for those
ecotypes harboring both loci, as well as for all MAM1 and MAM2
Kroymann et a/.
Table 1. PainNise comparisons of amino acids (upper triangle) and nucleotides (lower triangle)
Aa-O
Ag-O
Col-O
Mt-O
Gy-O
.
Ema-1
Pla-O
Ler-Otunc
Sorbo
Cvi o
Cal-O
Di-l
Lm-2 *
Condara
Hodia
Mr-O
Mr-O
Cal-O
Ka-O
NmO
Sei-O
Tsu~
Wl-O
Lip-O
Sorbo
81-1
Di-g
Ler-O
Petergof
Cvi-O
Condara
Hodja
=_
o
0
v
2
<:
O.' .
*** .
.
o.o
0.4
0.4
1.0
1.6
1.3
1.5
1.3
2.3
3.9
2.6
4.
4.1
4.0
_-
_ _4.0
_ 4.7
4.5
4.5
37
_ 50
o
0.0
***
0.4
0.4
I O_
1.6
1.4
1.5
1.4
2.4
3.9
2.7
4.3
4
4.0
4.1
4.7
4.5
4.6
-
~.7_
5.0
__
_ _
0.2 0.2
0.2 0.2
*** 0.0
0.0 ***
0.9 1.0
1.6 1.6
1.2 1.3
.4 1.5
1.3 _ 1.3
2.3 2.4
37 3.8
2.6 2.6
4.2 4.3
4.0 4.2
3.9 4.0
4.0 4.1
4.6 4.7
4.4 4.6
4.5 4.6
3.6 3.8
4.9 5.1
MAM 1
1 1 1
06~0.4~ 1.2~ 1.2
0.6 1 0.4 1 1.2 1 1.2
0.8 1 0.6 1 1.4 1 1.4
0.& 1 0.6 1 1.4 1 1.4
1 1 1
- 1 - 1 - 1 -
***~.0 1 1.8 1 1.8
. . . .
1.7 1 *** 1 1.2 1 1.2
.9 1 1.0 1 *** 1 0.0-
~ ~ 1
1.8 1 0.9 1 0.2 1 ***
.
2.7 1 1.7 1 14 1 1 5
3.013.3T3.5l~4
3.3 :8~ 3.3 ~.2
4.8 1 3.9 ~ 4.5 1 4.3
4.4 1 3.5 1 3.7 1 3.5
4.4 1 3.4 1 3.6 1 o.5
5.0 1 3.~4.4 L4.2
4.8 r 37 1 43 1 4.1
49 1-37 1 43 1 42
1 1 1
4.0 1 2.6 1 34 1 33
-4.5r4.3~4.8l4.6
1 - ~
1 *** 1
1 1.61
~ 0.4
r—OS l
.
1 0.4 1
rO.S 1
1 2.6 1
1 o.9 1
r301
1 1.8:
r~71
1 1
1 1.8 1
l
1 2.6
1
1 24 1
T
1 2.4
1
2.0
3.
rl
1 -1 ~1
~i~ =~
L
00 1 3.8
2.0 1 4.i
2.0 1 4.] 1
1 1
1 - 1 - 1
1 2.0 T 3.61
1
1 1.8 1 38 1
r~O I 4.51
1 ~ 1
1 10 1 4.5 1
1 1
1 *** 1 3.4 1
1 1 1
37 1 ***
2.0 1 27
3.1 12.1
2., 1 1.4
23 1 15
1
9.3 1 1.5
2.9 1 1.9
1 1
~ 2.8 ~ 1.8
, ~
28 1 18
20 1 2.0
3;
1 °
1
1
4.1
4.3
4.3
I
47
,
4.5
47
1 47
1 4.1
14
2.0
T***
1 10
1 1.1
1
1 1.1
1 0.7
c
7
7
1.4
-
MAM2
c 1
cc 1
~ 1
1 1
1 43
i 45
1 4.5
L - _~
45 1
.3:
4.1
41
3.0
0.8
1-2.2 1
1 14 1
1 *** 1
1 1
1 ol l
0.2
0.7'
06
-0.6
0.9
rl2
c
Z~ ~ ~=
~ ~ ;
3.8 1 3.8 1 4.7 1 45 1 4.7
1 1 1 1
3.8 1 3.8 1 4.7 1 4 5 1 4 7
4.1 1 4.1 1 4.9 1 4.7 1 4.9
41 T 4.1 1 4.9 1 4.7 1 49
I 1 I I
4.1 1 4.1 1 4.9 1 4.7 1 4 9
. . .
3.8 1 3.8 1 4.7 1 4 5 1 4 7
3.6 1 3.6 1 4.9 1 4 7 1 4 9
3.6 1 3.6 1 4.9- 1 4.7 1 4.9
3.4 1 ,.4 1 3.8 1 3.6 1 3.8
1.6 ~ 164~0
2.6 1 2.6 r 2.6 7 2.4 7 2.6
1.8 1 1.8 ~ 0.6 1 0.4 1 0.6
0.8 1 0.8 1 1.2 1 1.0 1 1.2
*** 1 0O 1 16 1 14 1 16
1 1 1 1
0.0 1 *** 1 1.6 1 1.4 1 1.6
0.8 rO8 r*-** 1 0.21 0.4
0.6 ~ 0.7 1 0.4 1 *** 1 0.2
0.6 1 0.7 1 0 4 1 0 O 1 ***
1.0 1 1-0_1 1.2 1 1.2 1 1.3
rl2r'T'01°91°9-
1 °
1 '-
1 ~ 1
4.l
l
4.3
4.3 1
437
3.6
4.7
4.7
,.6
1
1.0
2.0
1.6
1 1.0
1
1 1.4
1
1.4
1 1.4
1
1.2
1.4
1
1***
1 1.7
o
4.,
4.3
4.5
4.5
4.5
4.3
4.9
4.9
3.8
0.8
2.2
0.6
1.2
1.6
1.6
0.4
0.2
0.4
1.0
Values indicate percent divergence. Nucleotide divergence was calculated forthe entire genes, including 1,098 positions 5' of the start
codons, i.e., including 5' untranslated and promoter sequences (except forthe truncated Ler-O MAM1, where onlythe remaining portion
of the gene was used) and 60 positions 3' of the stop codons, i.e., including 3' untranslated sequences.
*The Lm-2 sequence is a fusion between MAM2 and MAM1.
sequences surveyed (Fig. 2~. The nucleotide polymorphism
pattern between MAM1 and MAM2 in Sorbo largely reflects the
general pattern seen among all MAM1 and MAM2 genes (Fig. 2
A and C), further strengthening the hypothesis that Sorbo
reflects the basal configuration of MAM loci at GS-Elong in A.
thaliana. In contrast, all other ecotypes harboring two MAM loci,
i.e., Cal-O, Condara, Hodja, Cvi-O, and Mr-O, show at least partial
sequence identity between the respective genes. However, ob-
served regions of sequence identity differ between ecotypes,
suggesting that multiple gene conversion events occurred inde-
pendently of each other. Genetic information has evidently been
transferred in both directions, i.e., in some cases from the MAM1
to the MAM2 locus, and in others from MAM2 to MAM1. For
example, in the Condara and Hodja ecotypes most of the MAM1
locus is very similar to the Sorbo MAM2 sequence (Fig. ~B). In
contrast, in Cal-O, most of the 3' part of MAM1 resembles a
MAM2-like sequence (Fig. 2E). Finally, in Cvi-O, one central part
of MAM2 was converted into a MAM1-like sequence, whereas
the 3' part of MAM1 turned into a MAM2-like sequence
(Fig. 2D).
Qualitative Effects Caused by Variation in the Organization of GS-
Elong. Ecotypes with a funct~onal MAM1 sequence, like Col-O,
accumulate methionine-derived glucosinolates with four meth-
ylene groups (C4) in their side chain, whereas those with only a
single MAM2 sequence, like Ler-O, sequester predominantly C3
aliphatic glucosinolates (4, 7~. Plants with both a MAM1- and a
Kroymann et a/.
MAM2-like sequence produce primarily C4 glucosinolates, indi-
cating that MAM1 function is dominant over MAM2 function
with respect to short-chain glucosinolates. As can be seen from
Fig. 1, presence or absence of a full-length Sorbo-like MAM1
gene gives complete prediction of glucosinolate phenotype.
Therefore, loss of the MAM1 function, either by deletion of the
MAM1 allele or by gene conversion, results in a switch from a C4
to a C3-producing ecotype, provided that MAM2 function is
reta~ned.
Quantitative Effects Caused by MAM1/MAM2 Polymorphism. To
investigate the impact on plant performance of natural variation
between MAM genes, Col-O was crossed to a recombinant inbred
line (17), CL5, which has a I-er-O MAM2 allele and is predom-
inantly Col-O elsewhere in the genome (7~. A total of 5,000 F2
progeny were screened for recombination in a 210-kb interval
containing the GS-Elong region. Prereproductive rosette plants
from 58 recombinant, homozygous, near-isogenic F4 lines were
assayed for glucosinolates, growth rate, and resistance to larvae
of two herbivorous lepidopteran insects, P xylostella and S.
exigua, a crucifer specialist and generalist, respectively (Fig. 3~.
QTLs for total leaf aliphatic glucosinolate concentration and
resistarlce to S. exigua both center at a 15-kb nonrecombinant
region containing Col-O MAM1 or LRr-O MAM2, but not MAM-L
(Fig. 4; F = 92.34, df—1, 54, P = 0.00001, n = 292 for aliphatic
glucosinolates; F - 13.70, df = 1, 54, P = 0.0005, n = 1,212 for
S. exigua). These data show that the Ler-O MAM2 allele at
PNAS I November25, 2003 I vof. 100 | suppl. 2 1 14589
o.e
o.e
~ o.~-
- . -
0,2- h
o
a,. ~
o.
o.
0~2- h
o. t
o
ATE TM 4~
_E~
..-
A All
~ 1 ~~.~ ~1
..-
C Sorbo
.
llA ~
~~ 61~~ 0,
Cal-O
MATS TM 4,
=
. B
~ . ~ hi
D
Condara
Hodja
Cvi-O
Mr-O
i
AA L4h~ ah ~ fit | ~ , ~ n ,~ it.
O 1 2 3 4 ~ 0 1 2 3 4 s
kb
Fig. 2. Sliding window analysis of nucleotide polymorphism (I) among MAM1 and MAM2 genes. Arrows indicate regions where compared sequences are
identical because of one or more presumptive gene conversion events between paralogous genes. The peaks between 500 and 590 and between 1,820 and 1,830
nucleotides stem from complex changes resulting in poor sequence alignment. (A) All sampled MAM1 and MAM2 alleles (except truncated alleles). (B) MAM1
versus MAM2 in the Condara ecotype. Differences between MAM1 and MAM2 are very similar in Hodja. (C) MAM1 versus MAM2 in the Sorbo ecotype. (D) MAM1
versus MAM2 in the Cvi-0 ecotype. (E) MAM1 versus MAM2 in the Cal-0 ecotype. (F) MAM1 versus MAM2 in the Mr-0 ecotype. Window size, 50 nt; step width:
10 nt. Alignment gaps are included in scaling of the horizontal axis. The MAM gene structure is depicted above the panels.
GS-Elong causes increased aliphatic glucosinolate concentration
and greater resistance to the generalist herbivore, S. exigua, in
comparison to the Col-O MAM1 allele (Figs. 3 and 4~. Herbivory
***
~ 1
l | *** E . ~
15 ~ ~ l 1 ~ d1 ~ ~ 1
i l I ~ I ~ ~ I 1- 1 1 ~ ~ l
1 1111 1 1 11 1 ~ 1 ~—1 1 1 1~
I ~ ~ ~ ~ I 17 - 1 ~ ill
1 1 1 1 ~ 111 1 1 1 1~ - 1 1 ~ ~ ~
1 ~ 1 I~ ~ ~ ~ 1 1~ - 1 1 1 ~ 1
1 MU 1 ~~U 1 ~
OR GS S.e. P.x.
-160
-a
t]20
, _ _
¢100
- 80
- 60
- 40
- 20
o
Fig. 3. Quantitative effects in near-isogenic lines carrying the Col-0 MAM1
(white bars) or the Ler-0 MAM2 allele (gray bars) at GS-Elong. Data are least
square means (+ standard error) from ANOVA (Col-0 = 100) for growth rate
(GR), total aliphatic glucosinolates (GS), damage by S. exigua (S.e.), and by P.
xylostella (P.x.). ***, P < 0.001.
14590 1 www.pnas.org/cgi/doi/] 0.1 073/pnas.17340461 00
by the specialist insect, P. xylostella, was unaffected by the allelic
state at GS-Elong (F = 2.50, df = 1, 54, P = 0.12, n = 1,239~.
Finally, above-ground biomass was measured in a separate
experiment on 3,130 plants in the same 58 mapping lines at 16
80
To
a_
8
._
-
.
a 20
· ~
o
._
LL
60
40
'A
0 50 100 150 200
Kilobases
;20
115 ~
1 ~
410 ~
1 _
1
I Y
1 ~
~ 5 ~
1 ~
· ~
o
I o
Fig. 4. QTL mapping of total leaf glucosinolate content (dashed line) and
resistance to S. exigua (solid line). Statistical significance is indicated by F
ratios.
Kroymann et a/.
Table 2. Population genetic parameters in the A. thaliana GS-Elong region
Gene MYB37 MAM2 MAM1 MAM-L
Coding positions 721 1,518 1,518 1,509
No. of sequences 18 11 8 18
No. of haplotypes 4 4 3 7
Total 0.0014 0.0045 0.0017 0.0034
Synonymous 0 0030 0.0065 0.0039 0.0083
Non synonymous 0~0009 0.0038 0.0010 0.0019
Tajima's D 0.43 (NS) 1.86* -1.22 (NS) 0.40 (NS)
McDonald-Kreitman, G; P 0.003; 0.954 (NS) 6.098; 0.014* 0.108; 0.743 (NS) 0.092; 0.762 (NS)
All calculations are based on coding regions of genes. (:)nly the third exon of MYB37 was amplified, which comprises >70% of the
entire ORE. Statistical significance of D was investigated by coalescent simulations. Nucleotide diversity (Utotal) at MAM2 is 0.0072 if we
include alleles that have experienced gene conversion. NS, not significant.
*P < 0.05.
days after germination. There was no trace of a significant
growth rate QTL at GS-Elong (F = 0.128, df - 1, 54, P = 0.73~.
Nueleotide Polymorphism Patterns. Statistical methods of molecu-
lar population genetics were used to test for nonneutral nucle-
otide polymorphism at GS-Elong (18, 19~. Haplotypes for which
evidence for gene conversion was found (eg., Fig. 2) were
excluded from significance testing to conform to the statistical
assumption of identically and independently distributed point
mutations. With the remaining alleles, two independent statis-
tical tests reject the equilibrium neutral hypothesis at MAM2.
First, a positive Tajima's D indicates too many intermediate
frequency polymorphisms (D - 1.86, P < 0.05 or P < 0.01, based
on coalescent simulations without recombination or with free
recombination, respectively; Table 2~. Second, we found too
many amino acid polymorphisms segregating in A. thaliana.
When compared with the corresponding gene from closely
related Arabidopsis Iyrata, the McDonald-Kreitman test (19)
shows the ratio of replacement to synonymous nucleotide poly-
morphisms withinA. thaliana is significantly higher than between
species (G = 6.098; P = 0 014~.
D· ~
Iscusslon
Molecular Basis of an Insect Resistance QTL. Quantitative analyses
demonstrate that GS-Elong constitutes an insect resistance QTL,
caused by variation in glucosinolate quantity, quality, or both.
Recpirocal deletions of MAM1 and MAM2 have occurred in
Ler-0 and Col-0 ecotypes, respectively, so these paralogous loci
segregate as alleles in our mapping population. A Ler-0 MAM2
allele at GS-Elong confers higher resistance to the generalist
insect herbivore S. exigua than a Col-0 MAM1 allele. Herbivory
by a specialist insect, P. xylostella, was unaffected by the allelic
state at GS-Elong. This is consistent with studies demonstrating
that generalist insects are sensitive toward glucosinolate-based
plant defenses, whereas specialists may be able to cope with these
compounds (16, 20, 21~. Moreover, feeding and oviposition of
crucifer specialist insects may be stimulated by glucosinolates
and their degradation products (21, 22~. Indeed, P. xylostella has
evolved a counteradaptation that enables it to circumvent hy-
drolysis of glucosinolates by myrosinase, and, thus, avoids the
formation of toxic breakdown products (23~.
This insect resistance QTL displayed complex molecular
variation (Fig. 1) with multiple independent mutations causing
production of C3 glucosinolates, extensive gene conversion,
deletion of large genomic regions, and functionally important
genes appearing in Ler-0 that are absent from the A. thaliana
Col-0 sequence. Similarly, previous studies in model organisms
have shown that gene conversion has important effects on
sequence variation in Hsp70 genes in Drosophila (24), and
multiple independent deletions at the FR1 locus influence life
history variation in A. thaliana (25, 26~. Conceptually, the
Kroymann et a/.
GS-Elong region exemplifies complex dynamics predicted for the
evolution of gene families (27~. In practice, the difficulty of such
studies should not be underestimated. Our sustained attempts to
characterize allelic variation in the GS-Elong region by using
PCR ultimately failed, and we finally sequenced 300 kb of Ler-0
bacterial artificial chromosomes to identify genes which were
absent from Col-0.
Evidence for Natural Selection. Two independent statistical tests
reject a neutral evolutionary hypothesis at MAM2. We found too
many intermediate frequency nucleotide polymorphisms and too
many amino acid changes segregating in A. thaliana (P < 0.05 by
Tajima's D and McDonald-Kreitman tests; Table 2~. Although
these findings reject a standard equilibrium neutral model, could
these patterns be attributable to nonstandard or nonequilibrium
demographic processes rather than to nonneutral evolution? For
example, metapopulation structure and population decline can
cause positive values of Tajima's D (28, 29~. Likewise, relaxation
of selection caused by bottlenecks or fixation of deleterious
mutations in small populations can cause elevated levels of
nonsynonymous polymorphism (304. Because these population
processes affect multiple loci throughout the genome, we exam-
ined nucleotide polymorphism at three genes immediately flank-
ing MAM2 (Table 2~. These results show that sequence poly-
morphisms adjacent to MAM2 (~MYB37, MAM1, and MAML) are
compatible with an equilibrium neutral model, based on Taji-
ma's and McDor~ald-Kreitman analyses (all P > 0.05~. More-
over, Haubold et al. (31) examined nucleotide polymorphism at
14 loci in a 170-kb genomic region containing the MOM gene
family. They found that genes with contrasting patterns of
variation (or and D) are located within a few kilobases of one
another. That result corroborates our current finding, where
nonneutral variation at MAM2 contrasts sharply with neutral
patterns of polymorphism at adjacent loci. Furthermore, this
local scale of mlcleotide variation appears to be typical for A.
thaliana (~32~.
Molecular variation at MAM2 also contrasts with results from
other genes in A. thaliana. Although excess nonsynonymous
polymorphism has been observed in comparisons of several A.
thaliana genes with congeneric relatives (e.g., refs. 26, 33, and
34), this reflects locus-specific effects that are not found in most
genes (34-39~. Furthermore, MAM2 displays an excess of inter-
mediate-frequency nucleotide polymorphisms, in contrast to
most other A. thaliana genes (26, 33, 34, 37, 39), including data
from ~500 loci sampled throughout the A. thaliana genome (K.
Schmid and T.M.-O., unpublished data).
Natural genetic variation at MAM2 shows too much intermedi-
ate-frequency nucleotide polymorphism and too many amino acid
variants, relative to neutral predictions, suggesting that balancing
selection maintains functional diversity at this ecologically impor-
tant gene. Balancing selection refers to evolutionary mechanisms
PNAS I November25. 2003 1 vol. 100 1 suppl. 2 1 14591
that maintain more genetic variation than expected under neutrality
(40), such as genotype-by-environment interaction, frequency-
dependent selection, or trench warfare models of host-enemy
convolution (41~. However,MAM2 polymorphism has no impact on
glucosinolate identity (Fig. 1), suggesting that nonneutrality at
MAM2 is likely caused by selection on glucosinolate quantity and
not quality. In support of this interpretation, ecological analyses of
natural selection on A. thal~ana in the field find stabilizing selection
for intermediate glucosinolate concentrations (42~.
Allocation Costs or Ecological Tradeoffs? Balancing selection im-
plies that different selective factors favor contrasting pheno-
types. For example, genetic variation for insect resistance could
be maintained by tradeoffs among components of fitness, if gains
in one aspect of fitness were balanced by losses in other fitness
components (43~. One form of tradeoff, allocation costs, has
been proposed to explain genetic polymorphism in plant resis-
tance to insect herbivores. Allocation costs occur when defense
mechanisms are energetically expensive, so that genotypes with
strong defenses have fewer resources to invest in growth and
reproduction (44~. To test for possible allocation costs of glu-
cosinolate production, we quantified growth rate (biomass)
before bolting. Juvenile biomass is positively correlated with
individual fitness in Arabidops~s and Brassica (45, 46~. Further-
more, prereproductive vegetative growth rate measures resource
availability in the exact environment and growth stage where the
Ler-0 MAM2 allele causes increased aliphatic glucosinolate
concentration and greater resistance to the generalist herbivore,
and allows us to test for allocation costs independently of
tolerance (1~. However, highly replicated quantification of ju-
venile growth rate in the fine-scale mapping lines gave no
evidence for the existence of allocation costs. In this growth
1. Rausher, M. D. (2001) Nature 411, 857-864.
2. Halkier, B. A. (1999) Trends Plant Sci. 11, 425-431.
3. Rask, L., Ar~dreasson, E., Ekbom, B., Enksson, S., Pontoppidan, B. & Meijer,
J. (2000) Plant Mol. Biol. 42, 93-113.
4. Kliebenstein, D. J., Kroymann, J., Brown, P., Figuth, A., Pedersen, D.,
Gershenzon, J. & Mitchell-Olds, T. (2001) Plant Physiol. 126, 811-825.
5. Reichelt, M., Brown, P. D., Schneider, B., Oldham, N. J., Stauber, E., Tokuhisa,
J., Kliebenstein, D. J., Mitchell-Olds, T. & Gershenzon, J. (2002) Phytochem-
ist1y 59, 66~671.
6. Campos de Quiros, H., Magrath, R., McCallum, D., Kroymann, J., Schnabel-
rauch, D., Mitchell-Olds, T. & Mithen, R. (2000) Theor. Appl. Genet. 101,
429-437.
7. Kroymann, J., Textor, S., Tokuhisa, J. G., Falk, K L., Bertram, S., Gershenzon,
J. & Mitchell-Olds, T. (2001) Plant Physiol. 127, 1077-1088.
8. Shelton, A. M., Cooley, R. J., Kroening, M. K, Wilsey, W. T. & Eigenbrode,
S. D. (1991) J. Entomol. Sci. 26, 17-26.
9. Frohman, M. A., Dush, M. K. & Martin, G. R. (1988) Proc. Natl. Acad. Sci. USA
85, 8998-9002.
10. Rozas, J. & Rozas, R. (1999) Bioinformaizcs 15, 174-175.
11. Saitou, N. & Nei, M. (1987) Mol. Biol. Evol. 4, 406-425.
12. Van de Peer, Y. & De Wachter, R. (1997) Comput. Appl. Biosci.
13, 227-230.
13. Tajima, F. & Nei, M. (1984) Mol. Biol. Evol. 1, 269-285.
14. Felsenstein, J. (1985) Evolution (Lawrer~ce, Kans.) 39, 78~791. 38.
15. Kliebenstein, D. J., Lambrix, V. M., Reichelt, M., Gershenzon, J. & Mitchell- 39
Olds, T. (2001) Plant Cell 13, 681-693. 40
16. Kliebenstein, D. J., Pedersen, D., Barker, B. & Mitchell-Olds, T. (2002) 41
Genetics 161, 325-332.
17. Lister, C. & Dean, C. (1993) Plant J. 4, 745-750.
18. Tajima, F. (1989) Genetics 123, 585-595.
19. McDonald, J. H. & Kreitman, M. (1991) Nature 351, 652-654.
20. Blau, P. A., Feeny, P., Contardo, L. & Robson, D. S. (1978) Science 200,
1296-1298.
21. Raybold, A. F. & Moyes, C. L. (2001) Heredity 87, 383-391.
22. Pivnick, K. A., Jarvis, B. J. & Slater, G. P. (1994) J. Chem. Ecol. 20, 1407-1427.
23. Ratzka, A., Vogel, H., Kliebenstein,~. J., Mitchell-Olds, T. & Kroymann, J.
(2002) Proc. Natl. Acad. Sci. USA 99, 1122~11228.
14592 1 www.pnas.org/cgi/doi/10. 1 073/pnas. 1734046100
environment and developmental stage, MAM2 provides resis-
tance to generalist insect herbivores without physiological costs.
Further experiments will be required to measure fitness conse-
quences in other developmental stages and environments.
An alternative mechanism of tradeoffs is well documented in
the Brassica literature. Ecological costs occur when defensive
metabolites are toxic to some herbivores but stimulate feeding or
oviposition by adapted, specialist insects (43~. Indeed, our
insect-feeding assays detected enhanced resistance against the
generalist herbivore S. exigua, conferred by the MAM2 allele at
GS-Elong, whereas herbivory by the crucifer specialist P. xylos-
tella was unaffected by the allelic state of MAM genes at
GS-Elong. On the other hand, there is compelling evidence that
exposure to glucosinolates, and their degradation products does
stimulate feeding, reproduction, or oviposition of a variety of
crucifer specialists including butterflies, moths, aphids, beetles,
and others, but not by generalists (ref. 22, reviewed in ref. 21~.
Thus, balancing selection at MAM2 may be explained by eco-
logical tradeoffs caused by contrasting biological effects of
glucosinolates on specialist versus generalist herbivores, al-
though we cannot rule out that allocation costs might have
additional impact in some environments.
We thank Deana Pedersen (University of Montana, Missoula) for
measuring plant growth rate and Dr. Scott A. McCuine (Department of
Soil and Crop Sciences, Texas A&M University, College Station) for
providing Ler-O bacterial artificial chromosomes. T.M.-O. was supported
by European Union Contract No. QLRT-2000-01097, the Bundesmin-
isterium ffir Bildung und Forschung, U.S. National Science Foundation
Grant DEB-9527725, and the Max-Planck-Gesellschaft. J.K. was sup-
ported by the Deutsche Forschungsgemeinschaft and the Max-Planck-
Gesellschaft.
24. Bettencourt, B. R. & Feder, M. E. (2002) J. Mol. Evol. 54, 569-586.
25. Hagenblad, J. & Nordborg, M. (2002) Genetics 161, 289-298.
26. Le Corre, V., Roux, F. & Reboud, X. (2002) Mol. Biol. Evol. 19, 1261-1271.
27. Lynch, M., O'Hely, M., Walsh, B. & Force, A. (2001) Genetics 159,
1789-1804.
28. Charlesworth, B., Morgan, M. T. & Charlesworth, D. (1993) Genetics 134,
1289-1303.
29. Wakeley, J. & Aliacar, N. (2001) Genetics 159, 89~905.
30. Eyre-Walker, A., Keightley, P. D., Smith N. G. & Gaffney, D. (2002) Mol. Biol.
EYOI. 19, 2142-2149.
31. Haubold, B., Kroymann, J., Ratzka, \, Mitchell-Olds, T. & Wiehe, T. (2002)
Genetics 161, 1269-1278.
32. Tian, D., Araki, H., Stahl, E., Bergelson, J. & Kreitman, M. (2002) Proc. Natl.
A cad. Sci. USA 99,11525-11530.
33. Kawabe, A., Yamane, K. & Miyashita, N. T. (2000) Genetics 156,
1339-1347.
34. Olsen, K M., Womack, A., Garrett, A. R., Suddith, J. I. & Purugganan, M. D.
(2002) Genetics 160,1641-1650.
35. Aguade, M. (2001) MoL Biol. Evol. 18, 1-9.
36. Hauser, M. T., Harr, B. & Schlotterer, C. (2001) Mol. Biol. Evol. 18, 1754-1763.
37. Kawabe, A., Innan, H., Terauchi, R. & Miyashita, N. T. (1997) Mol. Biol. Evol.
14, 1303-1313.
Kawabe, A. & Miyashita, N. T. (1999) Genetics 153, 1445-1453.
Kuittinen, H. & Aguade, M. (2000) Genetics 155, 863-872.
Nordborg, M. & Innan, H. (2002) Curr. Opin. Plant Biol. 5, 69-73.
.. Bergelson, J., Kreitman, M., Stahl, E. A. & Tian, D. (2001) Science 292,
2281-2285.
42. Mauricio, R. & Rausher, M. D. (1997) Evolution (Lawrence, Kans.) 51,
1435-1114.
43. Purrington, C. B. (2000) Curr. Opin. Plant Biol. 3, 305-308.
44. Tian, D., Traw, M. B., Chen, J. Q., Kreitman, M. & Bergelson, J. (2003) Nature
423, 74-77.
45. Mitchell-Olds, T. (1996) Evolution (Lawrence, Kans.) 50, 140-145.
46. Mitchell-Olds, T. & Bradley, R. D. (1996) Evolution (Lawrence, Kans.) 50,
1859-1865.
Kroymann et a/.
