1
Supporting informations - legends
2
Table S1. Success rate of T. mirus inter-population crosses.
3
Each progenitor is characterized by its locality of origin (Table S3), rDNA genotype
4
(gen) (Figure S7) and phenotype (phe) (Figure 1a). Genotypically and phenotypically
5
indistinguishable progenitors mPa2P(10) and mPa2P(3-10) are sometimes collectively
6
considered as mPa2P line. Similarly, progenitors mPu1D(8) and mPu1D(11) are
7
collectively considered mPu1D. Each cross is characterized by the number of F1
8
hybrids detected among the total analyzed progeny derived from the corresponding
9
maternal plant. The F1 hybrids used for further selfings are underlined.
10
11
Table S2. Frequency of distinct rDNA phenotypes associated with individual
12
genotypes.
13
Individuals from F1–F3 generations
14
correspond to those used in Figure 3c. For both mPu1D × mPa2P crosses, genotypes
15
harboring the
16
deletion (-).
were considered. Data highlighted in yellow
d1-rDNA variant (+) are compared with genotypes distinct for its
17
18
Table S3. Tragopogon populations analyzed.
19
20
Table S4. Primers used.
21
aF
= forward direction, R = reverse direction
22
bp
- porrifolius specific, d - dubius specific, dp - nonspecific
23
24
Figure S1. Conservation of diagnostic restriction sites discriminating T. porrifolius-
25
derived p-ITS (p) from T. dubius-derived d-ITS (d).
1
26
(a) Restriction maps, drawn from available ITS accessions. Distances (double-
27
headed arrows) are approximately to scale. Owing to sequence variability in the ITS
28
regions found among populations of both T. porrifolius and T. dubius [(Suarez-
29
Santiago et al. 2011) and sequence databases], invariant occurrence of the specific
30
diagnostic restriction sites was confirmed on the genome-wide scale for the number
31
of diploid progenitor populations using gDNA-CAPS (cleaved amplified polymorphic
32
sequence) analyses with indicated primers (single head arrows) and restriction
33
endonucleases BstNI (B), NlaIII (N) and Tsp509I (T). The restricted fragments were
34
size-separated on 2% agarose gels. (b) Comparable results obtained using BstNI
35
and NlaIII mediated CAPS support the credibility of variable transcriptional patterns
36
found in distinct T. mirus samples. Genomic DNA (lanes g) as well as cDNA from
37
young (lanes c) and old (c*) leaves was obtained from the same individuals as in
38
Figure 3a [mPa2P(10) × mPu1D(11)].
39
40
Figure S2. Extensive, but not complete, transcriptional silencing of the P2-rDNA
41
variant is induced by the mRo1´D(33A) genome with specific macrodeletion of the long
42
d-rDNA variant.
43
Proportion of d- and p-ITS1 homeologs in the primary transcript was determined by
44
cleaved amplified polymorphic sequence (CAPS) analysis of cDNAs prepared from
45
young (lanes c) and old (c*) leaves and manifested as the rDNA phenotypes D, DP
46
and P (Figure 1a) beneath these lanes. Genomic ITS1 was characterized similarly
47
using gDNA-CAPS (lanes g). Contributions of parental (1′ = mRo1′D line; 2 = mPa2P
48
line) rDNA variants in hybrids were inferred from corresponding genotypes (square
49
brackets; Figure 7) distinguished using VspI mediated RFLP of intergenic spacer
50
(lanes s). Unlike the F1 hybrid (F1Hy3′D-1) bearing rDNA variants inherited from the
2
51
paternal genome (arrowhead), siblings originated from unwanted selfings of mPa2P(3-
52
10) frequently retained the transcriptionally dominant P2-rDNA variant. The self-
53
pollinated F1Hy3′D-1 provided a number of F2 individuals characterized by genotype
54
(1′, 3′, 7′, 5′ and 8′; Figure S7) and phenotype (D and DP). Selected F2 individuals
55
F2Hy3′D-2 and F2Hy7′DP-4 (framed) with phenotypes D and DP, respectively, were self-
56
pollinated, resulting in F3 progenies with successfully maintained maternal
57
phenotypes. Cross between mPa2P(3-10) and mRo1´D(33A) theoretically provides the
58
same genotypes as that of mPa2P(10) ×mPu1D(11) (Fig. 5). Owing to the absence of
59
the long d1-rDNA variant, however, it is impossible to distinguish genotypes 9 from 5′,
60
6 from 3′, and 2 from 7′ (Figure S7). Note that transcriptionally silent p2-rDNA
61
epigenetic variant emerged already in F1Hy3′D-1 and was stabilized in F2Hy3′D-2 and
62
transmitted into F3 generation.
63
64
Figure S3. Significantly higher sequence homogeneity among BstNI/NlaIII repeat
65
units derived from d-rDNA compared with the p-rDNA homeolog.
66
All pairwise comparisons of sequence identity generated from the indicated numbers
67
of BstNI/NlaIII monomers were used for statistical evaluations and manifested either
68
as box plots or as histograms.
69
70
Figure S4. Alterations of intergenic spacer (IGS) structure among native Tragopogon
71
populations.
72
Three T. mirus rDNA genotypes (1, 1′ and 2) are represented by the direct
73
progenitors mPu1D(8), mRo1′D(33A) and mPa2P(10), respectively, used for crossings
74
(Table S1). Genomic DNAs were digested with HaeIII (panel a), NdeI+SacI (b) and
75
VspI (c) and then Southern hybridized to the corresponding probe BstNI/NlaIII (a),
3
76
IGS (b) and 18S (c). Note in (a), HaeIII, which targets promoter sequences but not
77
BstNI/NlaIII repeats, provided 1–2 rather long (1.3–1.8 kb) and 3–4 short (<1 kb)
78
fragments for d-rDNA and p-rDNA homeologs, respectively. Note in (c) only VspI
79
digestion provided easily separable fragments d1 and p2 specific for mPu1D(8) and
80
mPa2P(10), respectively. This may reflect potential size polymorphisms at the
81
external transcribed spacer. Considerable IGS size variability in allotetraploids at the
82
d- and p-subgenome level was confirmed using pairs of primers DF+DR (panel d)
83
and PF+PR (e) (Table S4), respectively. (f) Cleaved amplified polymorphic sequence
84
(CAPS) analyses of ITS1 specific for p-rDNA homeolog (primers F2 and P2; Figure
85
S5; Table S4) revealed interpopulation variability in mutual content of two abundant
86
(pa and pc) ITS1 SNP variants faithfully resembling patterns found between pL and pS-
87
IGS size variants (e). The rDNA variants d1, d2, p1 and p2 were assigned using VspI-
88
RFLP (c) and species-specific PCR (d, e) analyses.
89
90
Figure S5. Two ITS1 variants are usually associated with the p-rDNA homeologs.
91
(a) Restriction map drawn for each variant pa and pc distinct in A/C SNP affecting the
92
diagnostic restriction site for ClaI. (b) Highly variable proportions of both variants
93
were detected among T. porrifolius populations by gDNA-CAPS analyses (ClaI
94
restriction and primers F2 and P2; lanes g). Similar analysis of T. mirus is shown in
95
Figures S4f and S6. Complementary DNA-CAPS (lanes c) revealed higher relative
96
transcriptional activity of the pa-rDNA variant compared with its pc-rDNA counterpart,
97
resulting in reciprocal transcriptional patterns found between pBr and pPu individuals.
98
99
100
4
101
Figure S6. Elimination of rDNA variants induced by interpopulation hybridization.
102
VspI-mediated RFLP analysis of the intergenic spacer (IGS) showed that cross
103
between mPa2P(3-10) and mRo1′D(33A) resulted in an F1 hybrid (F1Hy3′D-1)
104
heterozygous for both rDNA homeologs d and p. Its selfing therefore resulted in
105
segregation of rDNA variants (d1′, d2, p1, and p2) in F2 progeny covering maternal (3′),
106
grandparental (1′) and newly emerged (4′, 5′, 7′ and 8′) genotypes, as termed above
107
each lane. Relative abundances of individual genotypes are expressed as
108
corresponding chart. Plant F2Hy3′D-f2 (framed) was selfed and the long p-rDNA
109
fragments present in both natural progenitors (arrowheads) were sometimes
110
eliminated in F3 progeny, resulting in the unexpected genotypes 3′′ and 5′′. Their
111
relative abundances are expressed as corresponding chart. Cleaved amplified
112
polymorphic sequence analysis specific for T. porrifolius-derived ITS1 (CAPS p-ITS1;
113
primers F2-P2, Figure S5a, Table S4) revealed selective deletion of the pa-ITS1 SNP
114
variant only in such genotypes, suggesting a link between the pa-ITS1 SNP variant
115
and the long p-IGS variant. This cross theoretically provides the same genotypes as
116
does the mPa2P × mPu1D cross (Figure 5b). Owing to the deletion of the long d1
117
variant in mRo1´D(33A), however, it is impossible to distinguish genotypes 9 from 5′ ,
118
6 from 3′, and 2 from 7′ (Figure S7). This might be inferred from selfings, as shown
119
for the apparent genotype 7′ (framed) that provides a significant fraction of genotype
120
8′ in progeny (chart) and implies that this F2 progenitor is of genotype 7′ rather than 2.
121
Unexpected genotypes 3′′ and 5′′ were detected in F3 but not in F2. Therefore, the
122
frequency of individual genotypes in F2 and F3 progeny derived from genotype 3′ ,
123
were evaluated separately.
124
125
5
126
127
Figure S7. Variable rDNA genotypes in Tragopogon mirus.
128
Distinct pattern of individual rDNA variants (d1, d1′, d2, p1, p1′, p2 and p2′)
129
characterized each genotype 1, 2 and 1′ found in the direct progenitor lines mPu1D,
130
mPa2P and mRo1′D, respectively, as well as hybrid genotypes 3-9, 3′, 4′, 5′, 7′, 8′, 3′′
131
and 5′′. VspI-mediated RFLP distinguishes all genotypes, each being composed of
132
several restriction fragments corresponding to individual rDNA variants (Figure 5).
133
The origin of each fragment was deduced from species-specific d- and p-PCR
134
analyses using primer pairs DF+DR and PF+PR, respectively (Table S4). The rDNA
135
homeologs originating from T. dubius and T. porrifolius are abbreviated as d and p,
136
respectively. Short and long rDNA homologs are denoted by
137
Fragments originating from mPu1D (mRo1´D) and mPa2P are in red and blue,
138
respectively. Violet indicates the superposition of rDNA fragments originating from
139
both progenitors.
140
performed on p-ITS1 (Figures S4f and S6) provided the same pattern as p-PCR for
141
each genotype (not shown). Colocalization of the d1 and d2 variants derived from
142
mPu1D and mPa2P, respectively, was deduced from the relative intensity of the dL and
143
dS PCR products.
144
145
146
Figure S8. Detailed bisulfite methylation analysis of the p- and d- rDNA homeologs
147
present in natural T. mirus mPa2P(10) and F3 interpopulation hybrid F3Hy2D-11
148
individuals. They are identical in rDNA genotype 2 but distinct in reciprocal ND
149
(phenotypes P and D, respectively). (a) Distribution of mC along each sequenced
150
RNA Pol I promoter region. Positions of boundary cytosines 1 and 205/6 correspond
151
to positions -218 and -15 with respect to the transcription initiation site (TIS). Each
S
and L, respectively.
Cleaved amplified polymorphic sequence (CAPS) analysis
6
152
horizontal line, except the first reference line, represents the sequence of an
153
individual clone obtained from PCR amplification of bisulfite-treated DNA. Extensively
154
hypomethylated clones are highlighted by asterisks, and this demethylated at all
155
symmetrical CGs is denoted by double asterisks. (b) Reference sequences used for
156
alignments. The TIS is indicated by an arrow, and it also serves as prominent
157
diagnostic site used for distinguishing homeologous clones.
158
159
Figure S9.
Methylation analysis mediated by restriction endonucleases (RE).
160
Genomic DNAs from two [mPa2P(10) and mPa2P(3-10)] individuals of a natural T.
161
mirus mPa2P line as well as two [F3Hy2D-1 and F3Hy2D-11] individuals of hybrid
162
F3Hy2D line with genotype 2 already restored in the F2 progenitor (the family
163
relationships between plant individuals analyzed is shown in Figure 3) were digested
164
with methylation insensitive NdeI and SacI (lanes -), followed by methylation sensitive
165
BstBI, HaeII, and Sau3AI and insensitive isoschizomer MboI. Resultant fragments
166
were size separated in 1% agarose gels, Southern hybridized to IGS probe and
167
quantified. (a) Enzyme HaeII that cuts p-IGS but not d-IGS provided specific IGS
168
fragments (p) which appeared more prominent in mPa2P line than in F3Hy2D, as
169
follows from the ratios between the intensity of corresponding hybridization signal
170
and the reference signal (r) provided by NdeI and SacI digestion. These ratios are
171
expressed as box plots below each corresponding lane. In contrast, BstBI that cuts d-
172
IGS but not p-IGS provided specific fragments (d) that were more prominent in
173
F3Hy2D line compared with mPa2P. (b) Restriction digest of F3Hy2D individuals with
174
Sau3AI resulted in longer IGS fragments (mC) than those released by the
175
methylation insensitive isoschizomer MboI, suggesting high methylation levels at
176
recognition sites. In contrast, both mPa2P plants released also shorter IGS fragments
7
177
(C) identical to those produced by MboI, suggesting its hypomethylation. Consistent
178
with bisulfite sequencing, such restriction patterns suggest that the fraction of p-IGS
179
is hypomethylated in mPa2P line but not in F3Hy2D line. (c) Although Sau3AI targets
180
both d- and p-IGS homeologs (Figure 4c, d), hypomethylated fragments are derived
181
from p-rDNA (p>d) rather than from d-rDNA, as follows from the MboI restriction
182
profiles of mPa2P(3-10) and both diploid progenitors, T. dubius (dBr, dPu) and T.
183
porrifolius (pBr, pPu). Restriction with NdeI, SacI and MboI confirmed identical rDNA
184
genotypes in mPa2P and F3Hy2D lines, as also revealed with VspI restriction (Figure
185
5).
186
187
Figure S10. Enhanced levels of intergenic spacer (IGS) transcript in T. mirus
188
interpopulation hybrids compared with diploid progenitors.
189
Complementary DNAs from young leaves were amplified with either primers F1 and
190
R1 (IGS) or F2 and R2 (ITS1). The family relationships of both sister individuals
191
F3Hy2D-3 and F3Hy2D-4 is shown in Figure 3. Asterisk denotes the PCR product
192
specific for the d-rDNA homeolog. Despite variable numbers of spacer promoters, no
193
significant difference in total level of IGS transcripts was found between T. dubius
194
and T. porrifolius. This level was considerably increased in allotetraploids. In contrast,
195
the level of total genic primary transcript measured at the control ITS1 region
196
appeared invariable across T. dubius, T. porrifolius and T. mirus.
197
8