ArticlePDF Available

Transcriptome and Network Analyses of Heterostyly in Turnera subulata Provide Mechanistic Insights: Are S-Loci a Red-Light for Pistil Elongation?

Authors:

Abstract and Figures

Heterostyly employs distinct hermaphroditic floral morphs to enforce outbreeding. Morphs differ structurally in stigma/anther positioning, promoting cross-pollination, and physiologically blocking self-fertilization. Heterostyly is controlled by a self-incompatibility (S)-locus of a small number of linked S-genes specific to short-styled morph genomes. Turnera possesses three S-genes, namely TsBAHD (controlling pistil characters), TsYUC6, and TsSPH1 (controlling stamen characters). Here, we compare pistil and stamen transcriptomes of floral morphs of T. subulata to investigate hypothesized S-gene function(s) and whether hormonal differences might contribute to physiological incompatibility. We then use network analyses to identify genetic networks underpinning heterostyly. We found a depletion of brassinosteroid-regulated genes in short styled (S)-morph pistils, consistent with hypothesized brassinosteroid-inactivating activity of TsBAHD. In S-morph anthers, auxin-regulated genes were enriched, consistent with hypothesized auxin biosynthesis activity of TsYUC6. Evidence was found for auxin elevation and brassinosteroid reduction in both pistils and stamens of S- relative to long styled (L)-morph flowers, consistent with reciprocal hormonal differences contributing to physiological incompatibility. Additional hormone pathways were also affected, however, suggesting S-gene activities intersect with a signaling hub. Interestingly, distinct S-genes controlling pistil length, from three species with independently evolved heterostyly, potentially intersect with phytochrome interacting factor (PIF) network hubs which mediate red/far-red light signaling. We propose that modification of the activities of PIF hubs by the S-locus could be a common theme in the evolution of heterostyly.
Content may be subject to copyright.
plants
Article
Transcriptome and Network Analyses of Heterostyly
in Turnera subulata Provide Mechanistic Insights: Are
S-Loci a Red-Light for Pistil Elongation?
Paige M. Henning 1, Joel S. Shore 2and Andrew G. McCubbin 1, *
1School of Biological Sciences, Washington State University, PO Box 644236, Pullman, WA 99164-4236, USA;
paige.henning@wsu.edu
2Department of Biology, York University, 4700 Keele Street, Toronto, ON M3J1P3, Canada; shore@yorku.ca
*Correspondence: amccubbin@wsu.edu
Received: 30 April 2020; Accepted: 29 May 2020; Published: 3 June 2020


Abstract:
Heterostyly employs distinct hermaphroditic floral morphs to enforce outbreeding. Morphs
dier structurally in stigma/anther positioning, promoting cross-pollination, and physiologically
blocking self-fertilization. Heterostyly is controlled by a self-incompatibility (S)-locus of a small
number of linked S-genes specific to short-styled morph genomes. Turnera possesses three
S-genes, namely TsBAHD (controlling pistil characters), TsYUC6, and TsSPH1 (controlling stamen
characters). Here, we compare pistil and stamen transcriptomes of floral morphs of T. subulata to
investigate hypothesized S-gene function(s) and whether hormonal dierences might contribute
to physiological incompatibility. We then use network analyses to identify genetic networks
underpinning heterostyly. We found a depletion of brassinosteroid-regulated genes in short styled
(S)-morph pistils, consistent with hypothesized brassinosteroid-inactivating activity of TsBAHD.
In S-morph anthers, auxin-regulated genes were enriched, consistent with hypothesized auxin
biosynthesis activity of TsYUC6. Evidence was found for auxin elevation and brassinosteroid
reduction in both pistils and stamens of S- relative to long styled (L)-morph flowers, consistent with
reciprocal hormonal dierences contributing to physiological incompatibility. Additional hormone
pathways were also aected, however, suggesting S-gene activities intersect with a signaling hub.
Interestingly, distinct S-genes controlling pistil length, from three species with independently evolved
heterostyly, potentially intersect with phytochrome interacting factor (PIF) network hubs which
mediate red/far-red light signaling. We propose that modification of the activities of PIF hubs by the
S-locus could be a common theme in the evolution of heterostyly.
Keywords:
Turnera subulata; heterostyly; dierential gene expression; self-incompatibility;
network analysis
1. Introduction
Plants possessing heterostyly exhibit mating types that dier in floral morphology and biochemical
mating specificity [
1
,
2
]. Species may possess two (distyly) or three (tristyly) floral morphs that exhibit
reciprocal positioning of male (anthers) and female (stigma) organs [
3
]. These traits combine to
promote outcrossing and prevent inbreeding [
1
,
4
,
5
]. Often, morphs also exhibit dimorphisms in other
reproductive characters, including pollen size/number, stigma size/shape, stigmatic papillae length,
and site of incompatible pollen tube inhibition [59].
Heterostyly has received considerable attention in several areas of genetics and evolutionary
biology. These include supergene inheritance [
10
13
], frequency-dependent selection [
14
], and the
evolution and fitness consequences of self-fertilization [
15
17
]. Heterostyly occurs in at least 199
Plants 2020,9, 713; doi:10.3390/plants9060713 www.mdpi.com/journal/plants
Plants 2020,9, 713 2 of 20
genera in 28 flowering plant families with surprising phenotypic consistency, and has been proposed to
have been a major contributor to Angiosperm diversity [
18
20
]. Consequently, these breeding systems
present remarkable opportunities to study convergent evolution.
Although heterostyly involves multiple traits it is controlled by a single S(self-incompatibility)-
locus. Historically heterostyly S-loci were viewed as diallelic, short styled (S-) morphs, commonly
being heterozygous (Ss) and long styled (L-) morphs homozygous recessive (ss), but this dogma was
recently overturned. S-loci of Primula [
21
24
], Linum [
25
,
26
], Fagopyrum [
27
,
28
], and Turnera [
29
], have
been at least partially characterized. The studies revealed these S-loci to be hemizygous, consisting of
a small number of linked genes in S-morph genomes that lack allelic counterparts in L-morph plants.
The nature of S-locus- (S-) genes identified to date suggests that incompatibility in species with
heterostyly is not a “lock and key” recognition system as found in homomorphic SI systems (which
do not have morphological variation between mating types) [
21
,
23
,
30
]. In heterostylous species,
incompatibility most likely results from physiological dierences more analogous to incongruity
barriers that reproductively isolate species [
31
]. These dierences result from the action of a small
number of S-genes, all which have been identified in Primula and Turnera [
21
,
22
,
28
], providing
opportunities to study how these reproductive barriers evolve and operate.
The ~230 kb S-locus in Turnera contains just three genes [
29
]. Only one, TsBAHD, is expressed in
pistils and necessarily determines pistil length and physiology. TsBAHD possesses the functional and
conserved domains of BAHD acyltransferases and has highest homology to members of this gene family
that inactivate brassinosteroids by acylation [
23
,
32
]. As a result, TsBAHD has been hypothesized to
possess brassinosteroid inactivating activity, but as homology-based prediction of BAHD acyltransferase
function is unreliable, this needs to be confirmed [
33
]. Intriguingly, the Primula CYP734A50 S-gene
was recently demonstrated to control style length by inactivating brassinosteroids [
21
]. Hence,
the independently evolved heterostyly systems of Turnera and Primula may use a common biochemical
mechanism to regulate style length despite recruitment of distinct genes to their S-loci [29].
The other two Turnera S-genes are expressed in stamens, TsYUC6 in anthers and TsSPH1 in anthers
and filaments. TsYUC6 exhibits high homology to YUCCA flavin-dependent monooxygenases, which
catalyze the second step in the two-step process of auxin biosynthesis from L-tryptophan [
34
]. TsSPH1
is a member of the large plant-specific S-protein homolog (SPH) family named after the Papaver rhoeas
pistil (homomorphic) SI determinant PrsS [
35
37
]. PrsS encodes a small secreted polypeptide ligand
that interacts with a receptor to activate a signal transduction cascade, leading to programmed
cell death in incompatible pollen [
38
,
39
]. The function(s) of other members of the SPH family are
unknown [
36
]. The precise roles of TsYUC6 and TsSPH1 in determining stamen characteristics are
uncertain. Filament length is likely determined by TsSPH1 as it is missing in L-morph deletion mutants
with short filaments, but present in a long-homostyle mutant [
29
]. TsYUC6 could potentially regulate
pollen size and physiology through auxin pathways, but whether TsSPH1 may also be involved in
these traits is unclear.
Studies in Primula [
40
], Linum [
25
], Turnera [
41
,
42
] and Fagopyrum [
43
], have identified genes
outside the S-locus that are dierentially expressed (DE) between floral morphs, suggesting that
S-loci cause dierential gene expression. More recent reports have employed transcriptome level
expression analyses to study heterostyly in several species including Lithospermum multiflorum [
44
],
Primula vulgaris [
24
], P. maximowiczii [
45
], and P. oreodoxa [
46
], but the pathways behind heterostyly
remain largely uncharacterized.
Here, we combine the analysis of organ specific transcriptomes with network analyses of
dierentially expressed genes (DEGs) in an attempt to gain insights into the mechanisms of action of
S-genes in Turnera and the pathways through which they operate. We compare transcript abundance
between pistils and stamens of the S- and L-morphs of T. subulata at two developmental time points.
Then, using the DEG datasets, we investigate whether dierential gene expression provides evidence
to substantiate hypothesized activities of the recently identified S-genes, TsBAHD, and TsYUC6 [
29
].
Specifically, we consider whether TsBAHD controls pistil characters by degrading brassinosteroids
Plants 2020,9, 713 3 of 20
and TsYUC6 stamen characters by synthesizing auxin. This was achieved by assessing whether
brassinosteroid and/or auxin related or regulated genes exhibit dierential expression in pistils and
stamens between the floral morphs. In addition, we assess whether transcriptome data supports
the hypothesis that incompatibility in Turnera might result from physiological mismatch between
pistils with reduced brassinosteroid levels and pollen from anthers with elevated auxin (S-morph),
and elevated brassinosteroid pistils and pollen from reduced auxin anthers (L-morph). Lastly, we use the
DEG datasets to predict gene networks and pathways underpinning in heterostyly in Turnera, the results
of which leads us to propose a novel hypothesis regarding the evolution of these breeding systems.
2. Results
2.1. Dierentially Expressed Genes: Identification and Quantification
Autotetraploid T. subulata was chosen as the main focus for this study for a number of
reasons. These include, profusion and size of flowers (aiding collection and staging of floral tissues),
availability of germplasm (including S-locus mutants), and the volume of previous work involving this
species [23,41,42,4752]
. For T. subulata, RNAseq was performed on stamens and pistils (separately) from
3 mm and 16 mm floral buds for each morph. These bud stages (shown in Figure 1) were chosen to
represent stages early (targeting morphology) and late in development (targeting mating specificity).
RNAseq was also performed on whole buds (minus sepals) of the two morphs of a second, closely
related heterostylous species, T. joelii (2n) [
53
,
54
], allowing us to identify DEGs common to both species
(see below).
Plants 2020, 9, x FOR PEER REVIEW 3 of 20
hypothesis that incompatibility in Turnera might result from physiological mismatch between pistils with
reduced brassinosteroid levels and pollen from anthers with elevated auxin (S-morph), and elevated
brassinosteroid pistils and pollen from reduced auxin anthers (L-morph). Lastly, we use the DEG datasets
to predict gene networks and pathways underpinning in heterostyly in Turnera, the results of which leads
us to propose a novel hypothesis regarding the evolution of these breeding systems.
2. Results
2.1. Differentially Expressed Genes: Identification and Quantification
Autotetraploid T. subulata was chosen as the main focus for this study for a number of reasons.
These include, profusion and size of flowers (aiding collection and staging of floral tissues),
availability of germplasm (including S-locus mutants), and the volume of previous work involving
this species [23,41,42,47–52]. For T. subulata, RNAseq was performed on stamens and pistils
(separately) from 3 mm and 16 mm floral buds for each morph. These bud stages (shown in Figure
1) were chosen to represent stages early (targeting morphology) and late in development (targeting
mating specificity). RNAseq was also performed on whole buds (minus sepals) of the two morphs of
a second, closely related heterostylous species, T. joelii (2n) [53,54], allowing us to identify DEGs
common to both species (see below).
Figure 1. Development of the floral morphs in Turnera subulata. The L-morph (left) and S-morph
(right) of T. subulata. Three developmental time points are depicted from left to right young (3 mm
bud), mature (16 mm bud), and open flower. Red arrows highlight the position of the stamens while
white arrows the position of the pistil.
For each sample ~20 million filtered clean RNA-seq reads were generated (Table S1). Cleaned
sequence reads from libraries of both species were mapped to a predicted T. subulata transcriptome
(enabling direct cross referencing of differentially expressed transcripts between the two species),
with mapping rates of 71% for T. subulata and 52% for T. joelii. Read count tables and differential
analysis files were then generated using Blast2GO-Pro (v5.2.5) (Figure S1). Gene expression was
compared pairwise between morphs for each developmental stage and organ, and genes statistically
differing in expression (adjusted p-value < 0.05 and |log
2
FC| > 1.0) retained for further analysis (listed
in File S1 Table S2).
Initially, 5744 genes were designated as differentially expressed between stamens and/or pistils
of the two morphs of T. subulata. Despite triplicate biological replicates and stringent criteria to
identify DEGs, further investigation (by RT-PCR) suggested this initial dataset contained many false
positives (data not shown). Suspecting that allelic variation was a significant factor in false DEG
designation (heterostylous species are obligate outbreeders), we added additional screens. T. subulata
DEGs were cross-referenced with those identified from a second heterostylous species, T. joelii (2n)
(Figure 2A–C), and genes not designated as DE in both species removed. This resulted in removal of
73.95% (4053) of the DEGs from the initial dataset. T. subulata and T. joelii are very closely related
[53,54] and their S locus genes exhibit 94%–96% identity, but mapping T. joelii RNAseq reads to the
Figure 1.
Development of the floral morphs in Turnera subulata. The L-morph (left) and S-morph (right)
of T. subulata. Three developmental time points are depicted from left to right young (3 mm bud),
mature (16 mm bud), and open flower. Red arrows highlight the position of the stamens while white
arrows the position of the pistil.
For each sample ~20 million filtered clean RNA-seq reads were generated (Table S1). Cleaned
sequence reads from libraries of both species were mapped to a predicted T. subulata transcriptome
(enabling direct cross referencing of dierentially expressed transcripts between the two species), with
mapping rates of 71% for T. subulata and 52% for T. joelii. Read count tables and dierential analysis
files were then generated using Blast2GO-Pro (v5.2.5) (Figure S1). Gene expression was compared
pairwise between morphs for each developmental stage and organ, and genes statistically diering in
expression (adjusted p-value <0.05 and |log
2
FC| > 1.0) retained for further analysis (listed in File S1
Table S2).
Initially, 5744 genes were designated as dierentially expressed between stamens and/or pistils of
the two morphs of T. subulata. Despite triplicate biological replicates and stringent criteria to identify
DEGs, further investigation (by RT-PCR) suggested this initial dataset contained many false positives
(data not shown). Suspecting that allelic variation was a significant factor in false DEG designation
(heterostylous species are obligate outbreeders), we added additional screens. T. subulata DEGs were
Plants 2020,9, 713 4 of 20
cross-referenced with those identified from a second heterostylous species, T. joelii (2n) (Figure 2A–C),
and genes not designated as DE in both species removed. This resulted in removal of 73.95% (4053) of
the DEGs from the initial dataset. T. subulata and T. joelii are very closely related [
53
,
54
] and their Slocus
genes exhibit 94%–96% identity, but mapping T. joelii RNAseq reads to the T. subulata transcriptome
could potentially lead to loss of genes that are polymorphic between the species. To assess whether
this was the significance of this issue, all genes removed were assessed for read counts mapped from
T. joelii. It was determined that only 303 (5.27%) of the genes removed lacked sucient T. joelii reads
to enable dierential expression analysis (and hence may have been excluded as a result of mapping
failure). A peripheral benefit of cross-referencing with T. joelii was to remove T. subulata DEGs that
might have been associated with autoploidy rather than heterostyly in the latter species. As heterostyly
in Turnera is controlled by three genes with distinct biological activities, DEGs identified as either up-
or down-regulated in both stamens and pistils at the same developmental stage were also viewed as
likely to have been present for reasons other than heterostyly and removed (195 DEGs). Combined,
these screens reduced the total number of DEGs to 1429. To assess the quality of the final datasets,
the expression of the top ten up- and down-regulated genes from each dataset was assessed in the
relevant organ of each morph by RT-qPCR. In total, 80 DEGs were confirmed to be DE as indicated
in the RNAseq datasets (up or down) (Figure 3) providing confidence in these final datasets. Organ
datasets were then split into up- and down-regulated sets, for a total of eight, allowing the assessment
of processes predicted to be up- or down-regulated in the S-morph.
Plants 2020, 9, x FOR PEER REVIEW 4 of 20
T. subulata transcriptome could potentially lead to loss of genes that are polymorphic between the
species. To assess whether this was the significance of this issue, all genes removed were assessed for
read counts mapped from T. joelii. It was determined that only 303 (5.27%) of the genes removed
lacked sufficient T. joelii reads to enable differential expression analysis (and hence may have been
excluded as a result of mapping failure). A peripheral benefit of cross-referencing with T. joelii was
to remove T. subulata DEGs that might have been associated with autoploidy rather than heterostyly
in the latter species. As heterostyly in Turnera is controlled by three genes with distinct biological
activities, DEGs identified as either up- or down-regulated in both stamens and pistils at the same
developmental stage were also viewed as likely to have been present for reasons other than
heterostyly and removed (195 DEGs). Combined, these screens reduced the total number of DEGs to
1429. To assess the quality of the final datasets, the expression of the top ten up- and down-regulated
genes from each dataset was assessed in the relevant organ of each morph by RT-qPCR. In total, 80
DEGs were confirmed to be DE as indicated in the RNAseq datasets (up or down) (Figure 3)
providing confidence in these final datasets. Organ datasets were then split into up- and down-
regulated sets, for a total of eight, allowing the assessment of processes predicted to be up- or down-
regulated in the S-morph.
Figure 2. Isolation of organ specific differentially expressed genes (SvL). (A) Number of genes
identified in each dataset relative to the whole transcriptome, and the fraction designated as DEG
between the S- and L-morphs. (B and C), Venn diagrams depicting comparison of DEGs identified in
both T. subulata and T. joelii at the different developmental stages. Numbers indicate the number of
DEGs identified within each dataset (numbers in overlapping sectors indicate DEGs found in all
samples overlapping that region). White circles indicate genes retained in the final DEG dataset.
Mathematical symbols refer to the expression pattern (SvL) of a given gene. + represents upregulated
genes, represents downregulated, and genes with reciprocal expression, i.e., upregulated in one
tissue type and downregulated in the other.
Figure 2.
Isolation of organ specific dierentially expressed genes (SvL). (
A
) Number of genes identified
in each dataset relative to the whole transcriptome, and the fraction designated as DEG between the
S- and L-morphs. (
B
,
C
), Venn diagrams depicting comparison of DEGs identified in both T. subulata
and T. joelii at the dierent developmental stages. Numbers indicate the number of DEGs identified
within each dataset (numbers in overlapping sectors indicate DEGs found in all samples overlapping
that region). White circles indicate genes retained in the final DEG dataset. Mathematical symbols
refer to the expression pattern (SvL) of a given gene. +represents upregulated genes,
represents
downregulated, and
,
genes with reciprocal expression, i.e., upregulated in one tissue type and
downregulated in the other.
Plants 2020,9, 713 5 of 20
Plants 2020, 9, x FOR PEER REVIEW 5 of 20
Figure 3. RT-qPCR analysis of select DEGs. Genes identified as up-regulated in the S-morph have
been colored red, those down-regulated in the S-morph colored blue. A maximum allowed Cycle
Threshold (CT) of 36 was used, samples with higher CT values were assigned values of 36 for
graphing purposes. This graph depicts the Log
2
of relative expression (SvL); also known as the ΔΔCT
value. Error bars represent standard error of ΔΔCT values. p-values were adjusted using the
Benjamini-Hochberg method for FDR correction. * adjusted p-value < 0.05, ** adjusted p-value < 0.01,
*** adjusted p-value < 0.001. Descriptions of these genes can be found in File S1.
2.2. Identification of Previously Reported Turnera DEGs
As expected, S-genes TsSPH1, TsYUC6, and TsBAHD, were all identified in their respective
datasets. Three additional genes have been previously reported to exhibit morph specific expression:
a style specific polygalacturonase (TsPG), a pollen-specific polygalacturonase (TsPP) [41], and a style
specific α-dioxygenase (TsαDOX) [42], so we searched for these genes in our data.
Two neighboring (on a genome scaffold) predicted genes were identified with high identity to
TsPG, Tsub_00011274-RA aligns to the front of TsPG, and Tsub_00011275-RA to the end, suggesting
both contain ORFs of TsPG (supported by their similar Log
2
FCs in all analyses). In our datasets TsPG
was up-regulated in young and mature pistils and down-regulated in mature stamens of the S-morph.
In young stamens transcript counts were <10 in all replicates, consistent with no expression in this
sample. A full-length predicted gene for TsPP was identified (Tsub_00009438-RA) and determined to
be expressed only in mature S-morph stamens, consistent with the previous report [41]. TsαDOX was
also identified (Tsub_00010763-RA) and determined to be up-regulated in S-morph mature pistils, this
transcript is present in other samples, but at dramatically lower read counts (<120 v’s > 70,000),
consistent with the previous report that TsαDOX expression is specific to S-morph styles [42].
Figure 3.
RT-qPCR analysis of select DEGs. Genes identified as up-regulated in the S-morph have been
colored red, those down-regulated in the S-morph colored blue. A maximum allowed Cycle Threshold
(CT) of 36 was used, samples with higher CT values were assigned values of 36 for graphing purposes.
This graph depicts the Log
2
of relative expression (SvL); also known as the
∆∆
CT value. Error bars
represent standard error of
∆∆
CT values. p-values were adjusted using the Benjamini-Hochberg method
for FDR correction. * adjusted p-value <0.05, ** adjusted p-value <0.01, *** adjusted p-value <0.001.
Descriptions of these genes can be found in File S1.
2.2. Identification of Previously Reported Turnera DEGs
As expected, S-genes TsSPH1,TsYUC6, and TsBAHD, were all identified in their respective datasets.
Three additional genes have been previously reported to exhibit morph specific expression: a style
specific polygalacturonase (TsPG), a pollen-specific polygalacturonase (TsPP) [
41
], and a style specific
α-dioxygenase (TsαDOX) [42], so we searched for these genes in our data.
Two neighboring (on a genome scaold) predicted genes were identified with high identity to
TsPG, Tsub_00011274-RA aligns to the front of TsPG, and Tsub_00011275-RA to the end, suggesting
both contain ORFs of TsPG (supported by their similar Log
2
FCs in all analyses). In our datasets TsPG
was up-regulated in young and mature pistils and down-regulated in mature stamens of the S-morph.
In young stamens transcript counts were <10 in all replicates, consistent with no expression in this
sample. A full-length predicted gene for TsPP was identified (Tsub_00009438-RA) and determined
to be expressed only in mature S-morph stamens, consistent with the previous report [
41
]. Ts
α
DOX
was also identified (Tsub_00010763-RA) and determined to be up-regulated in S-morph mature pistils,
this transcript is present in other samples, but at dramatically lower read counts (<120 v’s >70,000),
consistent with the previous report that TsαDOX expression is specific to S-morph styles [42].
Plants 2020,9, 713 6 of 20
2.3. Analysis of Dierential Expression
Of the 1429 genes identified as DE in at least one comparison, 369 were DE in young stamens, 338 in
mature stamens, 253 in young pistils, and 538 in mature pistils (Figure 2B). Combining developmental
stages, 685 genes were DE in stamens and 760 in pistils. Of these, 1311 were present in more than one
binary comparison, and 121 exhibited reciprocal dierential expression between pistil and stamen
tissues (Figure 2B,C). One aim of this study was to identify gene networks involved in heterostyly in
Turnera to enable development of testable hypotheses. Tools that facilitate such analyses are essentially
non-existent for non-model species. One approach to resolve this problem is to convert non-model
species datasets into Arabidopsis genes. A caveat to this approach is the inference of orthology between
Arabidopsis and Turnera homologs. Though BLAST scores are not robust metrics for predicting orthology,
it is reasonable to assume designations to be accurate in regard to gene family, if not specific isoform.
Consequently, though caution is needed in extrapolating results in regard to specific isoform, for
predicting gene function and enabling hypothesis development these designations have considerable
value. We performed this conversion by identifying the Arabidopsis homolog with the lowest e-value
(using BLASTp) for each DEG. We initially performed ClueGo enrichment to assess enriched and
depleted Gene Ontology (GO) terms in the datasets.
2.4. ClueGO Analysis of Enriched and Depleted GO Terms
All converted datasets were analyzed using the Cytoscape application coupled with the ClueGO
plug-in, to identify enriched GO terms and clusters. A comprehensive list of enriched GO terms
and clusters can be found in File S2. GO terms and genes in up-regulated datasets are referred to as
enriched, and those in down-regulated datasets as depleted.
In young S-morph stamens, 26 GO terms were enriched, clustering into twelve groups (Figure 4A).
Fifty-eight DEGs associated with the term cell periphery (GO:0071944) and 16 with transmembrane
transport (GO:0055085). TsYUC6 associated with seven terms in this dataset (Table S2). Twenty-one
GO terms were depleted, clustering into six groups (Figure 4A). One depleted cluster was composed
entirely of pathways related to the conversion of 5-hydroxyindole acetaldehyde to 5-hydroxyindole
acetic acid. Again, many genes associated with the term cell periphery were depleted. The term
regulation of hormone levels (GO:0010817) was also depleted.
In mature S-morph stamens, eleven GO terms were enriched, clustering into three groups
(Figure 4B). Enriched clusters included fatty acid derivative metabolic processes (GO:1901568) and
active transmembrane transporter activity (GO:0022804), which encompassed the majority of enriched
GO terms and included several ABC transporters. Eight GO terms were depleted in mature stamens,
clustering into three groups (Figure 4B). Forty-five genes associated with the cell periphery (GO:0071944),
and again several ABC transporters associated with this term.
In young S-morph pistils, eighteen GO terms were enriched, clustering into four groups (Figure 4C).
Moreover, 77% of enriched terms clustered under the term secondary metabolites. Several of these
related to flavonoids, including anthocyanidin (GO:0046283), phenylpropanoid metabolic (GO:0009698),
and biosynthetic processes (GO:0009699). Nine terms were depleted and clustered into four groups
(Figure 4C). The largest cluster, drug transport (GO:0015893), included several ABC transporters.
TsBAHD did not associate with any enriched term in either dataset.
In mature S-morph pistils, 40 GO terms were enriched, clustering into 10 groups (Figure 4D).
The term hormone metabolic processes (GO:0042445), including jasmonic acid signaling (GO:0009867),
fatty acid derivative biosynthesis (GO:1901570), and metabolic (GO:1901568) processes were enriched.
A total of 64 GO terms were depleted, clustering into ten groups (Figure 4D). 53% of depleted GO
terms are related to photosynthesis. Both flavonoid metabolic process (GO:0009812) and biosynthetic
process (GO:0009813) terms were depleted.
Plants 2020,9, 713 7 of 20
Plants 2020, 9, x FOR PEER REVIEW 7 of 20
Figure 4. Enriched and Depleted ClueGO clusters of up-regulated DEGs (SvL). Pie charts depicting
ClueGO generated clusters and overall percentage of GO terms associated with each cluster. Clusters
contain GO terms from young stamen (A), mature stamen (B), young pistil (C), and mature pistil (D)
that have commonality. Enriched and depleted are relative to S- vs L-morph expression levels; i.e.,
enriched terms are upregulated and depleted terms downregulated in the S-morph. * represents
clusters containing KEGG or Reactome pathways. (Individual GO terms have been represented using
bar graphs generated via ClueGO, provided in Figure S2. For a comprehensive list of the clustered
GO terms, see File S2.) All clusters had a p-value 7.28 × 104.
Figure 4.
Enriched and Depleted ClueGO clusters of up-regulated DEGs (SvL). Pie charts depicting
ClueGO generated clusters and overall percentage of GO terms associated with each cluster. Clusters
contain GO terms from young stamen (
A
), mature stamen (
B
), young pistil (
C
), and mature pistil
(
D
) that have commonality. Enriched and depleted are relative to S- vs L-morph expression levels;
i.e., enriched terms are upregulated and depleted terms downregulated in the S-morph. * represents
clusters containing KEGG or Reactome pathways. (Individual GO terms have been represented using
bar graphs generated via ClueGO, provided in Figure S2. For a comprehensive list of the clustered GO
terms, see File S2.) All clusters had a p-value 7.28 ×104.
Plants 2020,9, 713 8 of 20
2.5. Kyoto Encyclopedia of Genes and Genomes (KEGG) Analysis
A total of 675 KEGG Orthology (KO annotations were assigned to the 8 datasets (Figure S4a).
At the global level, overview maps were similar between samples, but young tissue types had an
overview map specific to their enriched and depleted datasets. For young pistils this was degradation
of aromatic compounds (01220), and for young stamen 2-oxo-carboxylic acid metabolism (01210)
(Figure S4b). The terms metabolic pathways and biosynthesis of secondary metabolites were present
in all eight DEG datasets (Figure S4b; File S3). Genes involved in BR biosynthesis (map:00905) were
identified in depleted datasets of both young and mature pistils and the enriched set of mature pistils
consistent with TsBAHD being involved in BR metabolism (File S3). Eleven DEGs with BR related GO
terms were identified in pistils, four enriched and seven depleted. DWF4/Cyp90B1,FER,At4g30410,
and MYB56 were enriched. DWF4 is negatively regulated by BR and its upregulation consistent with
reduced BR content in S-morph pistils. At4g30410 and MYB56 are both antagonists of BR signaling,
and their upregulation is consistent with reduced BR signaling. FER is required for BR stimulated
elongation of hypocotyls [
55
]. Amongst the depleted DEGs, several are induced by BRs (CYP716A1,
CYC3; 1, and EXO), one a transcriptional repressor of auxin and BR regulated genes (PAR1) and two
encode BR degrading enzymes (BAS1 and BEN1), again consistent with a reduced BR environment in
S-morph pistils. Overall, these results support the hypothesis that TsBAHD functions as a BR degrading
enzyme. As a member of the YUCCA gene family, TsYUC6 is hypothesized to catalyze conversion of
indole-3-pyruvate (IPA) to indole-3-acetic acid (IAA) [
56
]. Consistent with this hypothesis, genes in
tryptophan metabolism (map:00380), which includes this conversion, were identified in both young
and mature stamen enriched DEGs (File S3). Other genes involved in tryptophan metabolism also
occur in the depleted dataset of young stamens. These include YUC2, which acts redundantly with
YUC6 in Arabidopsis [
57
], and components of alternative pathways to auxin synthesis, consistent with
negative feedback down-regulation resulting from auxin production by TsYUC6.
In stamens, 16 DEGs (in addition to TsYUC6) were annotated with auxin related GO terms (eight
enriched, eight depleted). Of the enriched genes, four are auxin induced (AIR3,IAA7, SAUR27,
DFL1) and four involved in auxin transport (PIN6,TT7,TT4,ABCB11), consistent with a higher auxin
environment in S-morph stamens. For depleted DEGS, the situation was less distinct. One, SGR5,
is involved in regulation of auxin gradients regulating YUC gene mediated auxin synthesis and its
down-regulation consistent with negative feedback in a high auxin environment. Two (BRU6 and
IAA14) are auxin upregulated, but were depleted, as both are negative regulators of auxin signaling at
low auxin levels, depletion might reflect abnormally high auxin levels in S-morph relative to L-morph
anthers. Another two (LAX3 and ABCG37) are auxin transporters. Of the others, AGO1 a key player in
microRNA pathways, regulates auxin-induced adventitious root formation, and ASK2 is a component
of an SCF complex required for auxin mediated degradation of repressors of the auxin response.
We next sought evidence for reciprocal changes in BRs and/or auxin level between pistils and
stamens of the two morphs, which might potentially contribute to physiological incompatibility. For
this analysis we added back DEGs that were enriched or depleted in both stamens and pistils of the
S-morph. In stamens, four BR related DEGs were identified, At5g36130 was enriched, and CDL1,PAR1,
and EXO depleted (the latter two being also depleted in pistils). This suggests that BR levels were
reduced in S-morph stamens as well as pistils. In pistils, 16 auxin-related DEGs were identified, only
one common to stamens (IAA14, depleted in both tissues). Enriched DEGs included auxin biosynthesis
genes (AAO2), auxin induced genes (At5g54490,WRKY23,DFL1), genes involved in auxin signaling
(ABCB19,AFB3,MYBR1,At3g22810), and an auxin influx carrier (LAX2). Depleted DEGs included
several auxin transporters (NRT1.1,CYP711A1,TT4,TT7,At5g14090), and a transcriptional activator
of YUCCA genes (SRS7). Overall, these results suggest that S-morph pistils have elevated auxin
levels relative to L-morph pistils, but that this is achieved via an alternative auxin synthesis pathway
(involving AAO2 rather than YUCCAs) and possibly decreased auxin transport.
Given the small number of genes controlling heterostyly, an interesting finding was that a
substantial number of DEGs (22) were annotated with GO terms related to other plant hormone
Plants 2020,9, 713 9 of 20
pathways. This suggests that the activities of TsBAHD, and TsYUC6, and TsSPH1 in pistils and stamens
respectively, have transcriptional repercussions beyond direct BR and auxin regulation. The diversity
of plant hormone pathways represented in the DEGs suggest that activities of the S-genes intersect
with major developmental networks or signaling hub(s). With this in mind, we investigated potential
gene networks represented within the datasets.
2.6. Prediction of DEG Association Networks in Stamens and Pistils
STRING analysis utilizes known and predicted genetic association data to identify genetic
networks in novel datasets. The analysis predicts both direct (physical) interactions and indirect
(functional) interactions, using data including co-expression, detection of shared selective signals across
genomes, text-mining of scientific literature and computational transfer of known interactions between
organisms based on gene ontology [
58
]. Young and mature DEG datasets were combined, including
enriched and depleted DEGs for each tissue, and used to generate organ specific gene association
networks using STRING. As dierential expression was the sole criterium used to identify the genes
included, these analyses were expected to identify components of pathways involved in heterostyly,
but not necessarily entire pathways. The networks generated can be found in Figures S5 and S6 and a
comprehensive list of predicted associations in File S4.
STRING analysis identified 6740 interactions between 606 designated Arabidopsis homologs in
pistils and 3964 interactions between 534 designated Arabidopsis homologs in stamens, and overall
networks for each tissue are available in Figures S5 and S6, File S4, and the NDEx repository (Table S3).
NCMine software extracts subgraphs from networks based on node-weighting using degree centrality,
reporting subgroups as functional modules [
59
]. Given the number of interactions predicted, we focus
on functional clusters directly involving S-genes identified. Descriptions of gene function are derived
from annotations in STRING unless otherwise referenced.
2.6.1. Predicted TsBAHD Gene Network Associations in Pistils
Of the 606 DEGs in pistils, 25 are first neighbors with TsBAHD (Figure S7), meaning they directly
form an edge with TsBAHD connecting it with the rest of the network. Four of these have connections
to BR signaling, as DWF4 (enriched), AHL24 (enriched), EIR1 (depleted) are BR induced auxin carriers,
and At3g21710 (VUP1) a BR/auxin regulator (enriched). DWF4 encodes a C-22 hydrolase critical
to BR synthesis, is negatively regulated by BR and has been suggested to be a point of cross-talk
between BR and auxin pathways [
60
]. AHL24 is a member of the AT-hook-containing protein family,
which inhibit transcription of genes involved in auxin synthesis and auxin-mediated cell elongation in
hypocotyls [
61
64
]. Several enriched DEGs genes annotated as being involved in abscisic acid (ABA)
signaling (DOX1,PHO1 and RBHOH) were also in this group.
Eight clusters containing TsBAHD were identified, containing a combined total of 52 DEGs.
All are available on the NDEx repository, in the interest of brevity, we focus on the two subnetworks
(Figures S8 and S9) with highest clustering coecients and lowest heterogeneity scores, and DEGs of
particular note.
Cluster 4 (clustering coecient 0.962, heterogeneity 0.066) contains several DEGs of interest
(Figure 5). AHL24 (enriched) is involved in regulation of hypocotyl elongation as noted above.
MIZ1 (enriched) negatively regulates cytokinin sensitivity in root development. DOX1 (enriched)
promotes establishment of systemic acquired resistance, in a salicylic acid-dependent manner,
in response to incompatible interaction. This gene is also implicated in the negative regulation
of ABA-mediated signaling pathway. EIR1 (depleted) is an auxin eux carrier
Plants 2020,9, 713 10 of 20
Plants 2020, 9, x FOR PEER REVIEW 10 of 20
elongation in hypocotyls [61–64]. Several enriched DEGs genes annotated as being involved in
abscisic acid (ABA) signaling (DOX1, PHO1 and RBHOH) were also in this group.
Eight clusters containing TsBAHD were identified, containing a combined total of 52 DEGs. All
are available on the NDEx repository, in the interest of brevity, we focus on the two subnetworks
(Figures S8 and S9) with highest clustering coefficients and lowest heterogeneity scores, and DEGs
of particular note.
Cluster 4 (clustering coefficient 0.962, heterogeneity 0.066) contains several DEGs of interest
(Figure 5). AHL24 (enriched) is involved in regulation of hypocotyl elongation as noted above. MIZ1
(enriched) negatively regulates cytokinin sensitivity in root development. DOX1 (enriched) promotes
establishment of systemic acquired resistance, in a salicylic acid-dependent manner, in response to
incompatible interaction. This gene is also implicated in the negative regulation of ABA-mediated
signaling pathway. EIR1 (depleted) is an auxin efflux carrier
Figure 5. Representative String Network: Predicted TsBAHD gene network association
cluster 4.
Cluster 4 is of particular interest as it contains several hormone related genes and a member
of the AHL family, which we hypothesize may be important for female morphological characteristics.
Cluster 8 (clustering coefficient 1.0, heterogeneity 0.0) contains only two DEGs in addition to
TsBAHD, FAF3 which is involved in repressing the WUS homeobox gene controlling the stem cell
pool, and ABCB15 a proton ATPase that can transport auxin [65].
2.6.2. Predicted TsYUC6 Gene Network Associations in Stamens
Of the 534 DEGs that form networks in stamens, 22 are predicted to be first neighbors with TsYUC6
(Figure S10). Six of these are annotated as having functions in auxin signaling, as PIN6 (enriched) is an
auxin efflux carrier, SAUR27 (enriched) an auxin responsive gene, BRU6 (depleted) catalyzes synthesis
of indole-3-acetic acid (IAA)-amino acid conjugates (a mechanism for plants to inactivate excess auxin),
Cyp79B2 (depleted) converts tryptophan to indole-3-acetaldoxime a pre-cursor for IAA in an alternative
auxin synthesis pathway, SGR5 (depleted) is an auxin gradient regulator, and LAX3 (depleted) an auxin
transporter. Several other first neighbors of YUC6 are involved in other plant hormone pathways:
GA20OX1 (enriched) in gibberellic acid (GA) synthesis, IPT5 (enriched) in cytokinin synthesis, and
CDL1 (depleted) in BR synthesis.
Thirteen clusters were identified amongst the 77 DEGs predicted to participate in sub-networks
(Figures S11–S14) containing YUC6 (NDEx repository). Cluster 4 (clustering coefficient 0.8,
heterogeneity 0.471) contains several genes involved in regulating GA homeostasis, GA2OX1
(enriched) is involved in GA synthesis, and GA2OX4 (depleted) and GA2OX6 (enriched) have similar
activities in interconverting GA variants.
Figure 5.
Representative String Network: Predicted TsBAHD gene network association cluster 4.
Cluster 4 is of particular interest as it contains several hormone related genes and a member of the
AHL family, which we hypothesize may be important for female morphological characteristics.
Cluster 8 (clustering coecient 1.0, heterogeneity 0.0) contains only two DEGs in addition to
TsBAHD,FAF3 which is involved in repressing the WUS homeobox gene controlling the stem cell pool,
and ABCB15 a proton ATPase that can transport auxin [65].
2.6.2. Predicted TsYUC6 Gene Network Associations in Stamens
Of the 534 DEGs that form networks in stamens, 22 are predicted to be first neighbors with
TsYUC6 (Figure S10). Six of these are annotated as having functions in auxin signaling, as PIN6
(enriched) is an auxin eux carrier, SAUR27 (enriched) an auxin responsive gene, BRU6 (depleted)
catalyzes synthesis of indole-3-acetic acid (IAA)-amino acid conjugates (a mechanism for plants to
inactivate excess auxin), Cyp79B2 (depleted) converts tryptophan to indole-3-acetaldoxime a pre-cursor
for IAA in an alternative auxin synthesis pathway, SGR5 (depleted) is an auxin gradient regulator,
and LAX3 (depleted) an auxin transporter. Several other first neighbors of YUC6 are involved in other
plant hormone pathways: GA20OX1 (enriched) in gibberellic acid (GA) synthesis, IPT5 (enriched) in
cytokinin synthesis, and CDL1 (depleted) in BR synthesis.
Thirteen clusters were identified amongst the 77 DEGs predicted to participate in sub-networks
(Figures S11–S14) containing YUC6 (NDEx repository). Cluster 4 (clustering coecient 0.8,
heterogeneity 0.471) contains several genes involved in regulating GA homeostasis, GA2OX1 (enriched)
is involved in GA synthesis, and GA2OX4 (depleted) and GA2OX6 (enriched) have similar activities in
interconverting GA variants.
Cluster 5 (clustering coefficient 0.8, heterogeneity 0.328), contains two genes involved in cytokinin
homeostasis, IPT5 (enriched) is involved in cytokinin synthesis and LOG5 (enriched) is a cytokinin
activating enzyme. In addition, this cluster contains a down-regulated auxin transporter-like protein,
LAX3.
Cluster 6 (clustering coecient 0.833, heterogeneity 0.2) contains two genes involved in auxin
synthesis, both depleted (BRU6 and YUC2), presumably a result of negative feedback resulting from
auxin synthesis by TsYUC6.
Cluster 13 (clustering coecient 0.423, heterogeneity 0.413) contains several genes involved with
auxin. ABCG37 (depleted) a negative regulator of auxin polar transport inhibitors, MIZ1 (depleted)
which plays a role in lateral root development by maintaining auxin levels as well as negatively
regulating cytokinin sensitivity in root development. Lastly, RCI3 (enriched) which is involved in
auxin catabolism.
Plants 2020,9, 713 11 of 20
2.6.3. Predicted TsSPH1 Gene Network Associations in Stamens
Only four genes are predicted to be first neighbors with TsSPH1 (Figure S15). TsSPH1 is
hypothesized to regulate filament elongation [
29
]. Three formed a single cluster (clustering coecient
0.833, heterogeneity 0.2) (available on the NDEx repository; Figure S16). These were At1g59835 (CEP2)
(enriched) and two RAPID ALKALINIZATION FACTOR-LIKE (RALFL) proteins, RALFL7 (depleted)
and RALFL16 (enriched). All three encode small extracellular signaling polypeptides.
CEP2 encodes a small extracellular peptide ligand mobile and is expressed in A. thaliana filaments.
Some CEPs have been shown to bind to leucine rich repeat receptor kinases to negatively regulate
root growth [
66
]. Both RALFL7 and RALFL16 are type IV RALFs, a clade predominantly expressed
in Arabidopsis inflorescence tissues, and RALFL16 is exclusive to floral tissues [
67
]. FERONIA is
the only known RALF receptor, and binds RALF to activate a signaling cascade that inhibits cell
elongation [
67
,
68
]. Though RALFL7 is depleted and RALFL16 enriched in S-morph stamen, read counts
suggest that overall RALFL proteins are reduced by ~10 fold in S-morph stamens (long filaments),
relative to L-morph (short filaments). This leads us to speculate that depletion of RALFL7 could be
integral to promoting cell elongation in S-morph filaments by removing inhibition of FERONIA.
2.6.4. Genes and Networks with Reciprocal Dierential Expression between Stamens and Pistils
We hypothesized that SI in Turnera might result from reciprocal dierences in gene expression in
stamens and pistils. Consequently, we investigated potential gene networks enriched in one organ
and depleted in the other. To identify such networks, we compared stamen and pistil datasets using
DyNet. A total of 385 interactions between 128 DEGs are shared between the organs. Of these,
76 showed reciprocal expression and 132 predicted interactions were identified. These genes clustered
into 5 GOGroups (Table S4). The DEGs clustered under the terms catalytic activity, nucleus, cell wall
organization, terpenoid biosynthetic process, and intrinsic components of membranes. These results
suggest that there may be reciprocal dierences in isoprenoid and terpenoid metabolism between
stamens and pistils of the two morphs. The majority of DEGs listed as intrinsic membrane components
are membrane transporters, a number of which possess ATPase activity. These could potentially be
related to reduced cell size in pistils and increased pollen/filament cell size in S- relative to L-morph
flowers. The presence of several ABC transporters in this group is also intriguing as these could
potentially be involved in the transport of auxin or peptide ligands.
3. Discussion
Several insights into the mechanisms underpinning heterostyly in Turnera were gained in this
study. Empirical evidence, in the form of dierential gene expression, was found for the hypothesized
enzyme activities of the S-genes TsBAHD and TsYUC6. Specifically, the identity and changes in
expression of DEGs in pistils support lower brassinosteroid levels in S- relative to L-morph pistils,
and higher auxin levels in S- relative to L-morph anthers. Somewhat surprisingly, the data also suggest
that auxin and brassinosteroid levels are altered in a consistent manner in anthers and pistils, leading us
to extend our hypothesis that incompatibility in Turnera results from physiological hormone mismatch
between pollen and pistils. Lastly, evidence was found for other hormone pathways being perturbed
by the action of the three S-genes, suggesting that a network signaling hub might underpin heterostyly.
The identities of DEGs forming genetic functional clusters directly involving S-genes were determined
by network analysis, generating results that lead us to propose a novel hypothesis concerning the
mechanism and evolution of heterostyly below.
Robust datasets are critical for meaningful bioinformatic analysis, and false positives were a
significant issue in our initial DEG datasets. This was most likely due to the inherent heterozygosity
of obligately outbreeding species causing miss-mapping to the transcriptome. This being the case,
increasing the number of biological replicates was unlikely to solve this issue. Our approach of
cross-referencing DEGs between two heterostylous species (which dier in the genes aected by
Plants 2020,9, 713 12 of 20
issues caused by heterozygosity) dramatically improved the quality of the DEG analysis, as confirmed
by assaying eighty DEGs by qPCR. The resulting datasets were inclusive in that they included the
three S-genes and all previously identified DEGs, and dierential expression and tissue specificity of
these genes was consistent with previous reports with a single exception. TsPG a polygalacturonase
previously reported to be style specific [
41
], was found in this study to be expressed in mature stamen
as well as pistils. As the previous report assessed expression in pollen rather than stamens, this
discrepancy is likely because TsPG is expressed in mature anther or filament tissues. This consistency
between our results and published data provided confidence that our DEG datasets were robust,
and included the majority of dierentially expressed genes, and hence further analyses would provide
meaningful information.
The analyses provided support for the hypothesis that TsBAHD functions as a BR degrading
BAHD acyltransferase to control pistil length. In S-morph pistils, genes involved in BR biosynthesis
were depleted, genes negatively regulated by BR (DWF4) and antagonists of BR signaling (At4g30410
and MYB56) enriched, and a BR regulated gene (PAR1) and BR degrading enzymes (BAS1 and BEN1)
were depleted. These results are consistent with a reduced BR environment in S-morph, relative
to L-morph, pistils. A recent study of pistils of heterostylous Primula oreodoxa reported that genes
related to plant hormone signal transduction and BR biosynthesis pathways were also dierential in
Primula pistils [
46
]. As style length in Primula is governed by the presence of the BR degrading enzyme
CYP734A50 [
21
], the data from Primula, combined with the results presented herein support that a
common mechanism underlies pistil characteristics in Primula and Turnera.
Support was also found for the hypothesis that TsYUC6 controls pollen and/or stamen
characteristics through auxin mediated pathways [
29
]. In Arabidopsis,AtYUC6 and AtYUC2 act
redundantly to regulate pollen development, filament elongation, and anther dehiscence [
57
,
69
,
70
].
As a YUCCA gene family member, TsYUC6 likely catalyzes conversion of indole-3-pyruvate to
indole-3-acetic acid [
56
]. We found evidence in the DEG datasets for elevated auxin levels in
S-morph stamen. Genes involved in tryptophan metabolism, which includes YUCCA mediated auxin
synthesis, were identified in both young and mature stamen enriched DEGs. Other genes associated
with tryptophan metabolism were depleted in young stamens, including YUC2 and components of
alternative pathways to auxin synthesis, likely a result of negative feedback caused by auxin production
by TsYUC6. Transcriptional evidence for the presence of elevated auxin in S-morph anthers included
enrichment of auxin induced genes (AIR3,IAA7, SAUR27,DFL1), genes involved in auxin transport
(PIN6,TT7,TT4,ABCB11), and depletion of the auxin repressed gene SGR5, auxin transporters (LAX3
and ABCG37), and Cyp79B2, which converts tryptophan to indole-3-acetaldoxime. BRU6 and IAA14
were also depleted, and both are rapidly but transiently induced at low auxin levels [
71
] and their
depletion likely reflects long term auxin elevation in S-morph anthers. Though it is not immediately
clear why some auxin transporters would be enriched and others depleted, overall, these results are
consistent with auxin being higher in S- than L-morph anthers, consistent with TsYUC6 mediating at
least some aspects of stamen phenotype by auxin synthesis.
The paucity of functional characterization of SPH1 gene family members makes hypothesizing
a mechanism of action for TsSPH1 challenging. Network analysis identified genes that might act in
genetic networks with TsSPH1 to influence stamen development. That RALFL7 is part of a network
with TsSPH1 and was depleted is interesting. RALF activates a signaling cascade mediated by
the receptor kinase FERONIA to inhibit cell elongation [
67
,
68
]. Whether RALFL7 possesses this
activity is unknown, but reducing RALFL7 could potentially promote cell elongation by releasing
inhibition of FERONIA resulting in the long S-morph filament phenotype. We previously proposed that
incompatibility in Turnera might result from physiological mismatch between low BR pistils/high auxin
pollen (S-morph) and high BR pistils/low auxin pollen (L-morph). We identified a number of genes
that exhibited reciprocal dierential expression between pistils and stamens, as might be expected of
genes contributing to a physiological barrier preventing self-fertilization. Genes involved in terpenoid
pathways and membrane transporters were found to be highly represented but no pathway with
Plants 2020,9, 713 13 of 20
clear implications to pollen-pistil compatibility was identified. Interestingly, however, DEGs were
identified that suggested that auxin and BR are altered in a consistent manner in pistil and stamen
tissues within a floral morph. despite dierent S-genes being expressed in each tissue. This suggests
that BR was reduced and auxin elevated in both pistils and stamens in S-morph relative to L-morph
flowers, and that this is caused by a dierent set of genes in each tissue. This is consistent with, but
extends upon, our previous hypothesis that reciprocal hormone environments between morphs might
be a responsible for compatibility [
29
]. The new data suggest that pollen that developed in a low
BR/high auxin environment (S-morph) is incompatible with low BR/high auxin S-morph pistils but
compatible with high BR/low auxin L-morph pistils, and vice versa (Figure 6).
Plants 2020, 9, x FOR PEER REVIEW 13 of 20
with TsSPH1 and was depleted is interesting. RALF activates a signaling cascade mediated by the
receptor kinase FERONIA to inhibit cell elongation [67,68]. Whether RALFL7 possesses this activity
is unknown, but reducing RALFL7 could potentially promote cell elongation by releasing inhibition
of FERONIA resulting in the long S-morph filament phenotype. We previously proposed that
incompatibility in Turnera might result from physiological mismatch between low BR pistils/high
auxin pollen (S-morph) and high BR pistils/low auxin pollen (L-morph). We identified a number of
genes that exhibited reciprocal differential expression between pistils and stamens, as might be
expected of genes contributing to a physiological barrier preventing self-fertilization. Genes involved
in terpenoid pathways and membrane transporters were found to be highly represented but no
pathway with clear implications to pollen-pistil compatibility was identified. Interestingly, however,
DEGs were identified that suggested that auxin and BR are altered in a consistent manner in pistil
and stamen tissues within a floral morph. despite different S-genes being expressed in each tissue.
This suggests that BR was reduced and auxin elevated in both pistils and stamens in S-morph relative
to L-morph flowers, and that this is caused by a different set of genes in each tissue. This is consistent
with, but extends upon, our previous hypothesis that reciprocal hormone environments between
morphs might be a responsible for compatibility [29]. The new data suggest that pollen that
developed in a low BR/high auxin environment (S-morph) is incompatible with low BR/high auxin
S-morph pistils but compatible with high BR/low auxin L-morph pistils, and vice versa (Figure 6).
Figure 6. Hypothesis for the basis of self-incompatibility in Turnera.
Pale grey indicates
stamens and pistil with relatively low brassinosteroid and high auxin concentration in the S-morph
flowers, dark grey the relatively high brassinosteroid and low auxin concentration in stamens and
pistil of the L-morph. It is hypothesized that pollen developing in one hormone environment cannot
grow on/in pistils with the same hormone environment (self- or within morph pollinations) and dies
after within-morph or self-pollination. In contrast, cross-pollination to pistils with the opposite
hormone environment leads to successful pollen tube growth and fertilization. S-gene labels indicate
the tissues in which they are expressed in the S-morph.
Beyond supporting the hypothesized enzyme activities of TsYUC6 and TsBAHD, the results of
this study provide evidence that auxin levels in stamens and BR levels in pistils are not the only
hormonal differences caused by the S-locus. Rather, multiple hormone pathways are affected in both
organs, suggesting that auxin and BR signaling in stamens and pistils (respectively) intersect with
additional signaling networks. To investigate this further, we used transcriptome data to predict gene
networks directly involving S-genes.
A particularly noteworthy DEG found in these networks was Tsub_00021189 (an AHL family
member), which was upregulated in S-morph pistils. AHLs regulate homeostasis of a variety of plant
Figure 6.
Hypothesis for the basis of self-incompatibility in Turnera. Pale grey indicates stamens
and pistil with relatively low brassinosteroid and high auxin concentration in the S-morph flowers,
dark grey the relatively high brassinosteroid and low auxin concentration in stamens and pistil of
the L-morph. It is hypothesized that pollen developing in one hormone environment cannot grow
on/in pistils with the same hormone environment (self- or within morph pollinations) and dies after
within-morph or self-pollination. In contrast, cross-pollination to pistils with the opposite hormone
environment leads to successful pollen tube growth and fertilization. S-gene labels indicate the tissues
in which they are expressed in the S-morph.
Beyond supporting the hypothesized enzyme activities of TsYUC6 and TsBAHD, the results of this
study provide evidence that auxin levels in stamens and BR levels in pistils are not the only hormonal
dierences caused by the S-locus. Rather, multiple hormone pathways are aected in both organs,
suggesting that auxin and BR signaling in stamens and pistils (respectively) intersect with additional
signaling networks. To investigate this further, we used transcriptome data to predict gene networks
directly involving S-genes.
A particularly noteworthy DEG found in these networks was Tsub_00021189 (an AHL family
member), which was upregulated in S-morph pistils. AHLs regulate homeostasis of a variety of
plant hormones including GA [
72
], jasmonic acid [
73
], cytokinin [
74
], auxin [
63
,
64
], and BR [
62
,
64
].
They anchor DNA to the nuclear matrix by binding AT rich conserved scaold matrix nuclear
attachment regions, to aect gene regulation through chromatin remodeling [
63
]. AHLs interact to
form homo-/hetero-complexes, which complex with other nuclear proteins that modulate plant growth
and development in response to environmental signals [
62
]. Two members of this gene family, namely
SUPPRESSOR OF PHYTOCHROME B-4 #3 (SOB3) and ESCAROLA (ESC), act redundantly to repress
hypocotyl elongation in light grown Arabidopsis seedlings [
61
]. Knocking-out additional AHL family
members, increases phenotype severity (hypocotyl length) [
62
]. Intriguingly, seed set is reduced in
Plants 2020,9, 713 14 of 20
Arabidopsis plants overexpressing a dominant mutant SOB3 allele, but whether this is caused by
breakdown of embryo development or a pollination defect (potentially analogous to SI in heterostyly)
has not been investigated [
61
]. We propose that Tsub_00021189 may play a key role in the short
pistil phenotype.
There has been much recent progress made in elucidating pathways in plant hormone signaling
and major signaling hubs identified. Surveying the biological eects of these pathways, the genetic
components of signaling hubs relative to the DEGs identified in this study, and the identity of S-genes
in this and other species, leads us to speculate a common theme in the evolution of heterostyly, at least
in regard to pistil characters. We hypothesize this to be recruitment to the S-locus of genes that intersect
with, and modify the action of, phytochrome interacting factor (PIF) signaling networks.
Phytochrome associated pathways can modify stamen and pistil morphologies, as both can
be modified by red/far-red light levels [
75
]. Brassica rapa plants exposed to low red/far-red have
shorter pistils and longer stamens than plants exposed to high red/far-red, increasing herkogamy in a
manner analogous to heterostyly [
75
]. Further, PIF signaling hubs intersect with and aect hormone
pathways including GA, auxin, BR, ABA, and jasmonic acid [
76
] in a manner comparable to the
dierential expression observed in this study. Several gene families represented by DEGs in this
analysis are significant players in PIF networks, (including SAURs,IAAs,DWFs,GA20OXs,PINs,AHLs).
In particular, AHLs are implicated in PIF pathways regulating cell elongation in hypocotyls [
61
,
64
,
77
].
Combined, these factors make PIF signaling networks attractive candidates to be hubs involved
in heterostyly.
Interestingly, all S-genes that control pistil characteristics identified to date potentially intersect
with PIF hubs. In Primula and Turnera, pistil length appears to be mediated by BR degradation,
albeit by genes encoding dierent enzymes [
21
,
23
], and BR related dierential expression in pistils
in Turnera also occurs in Primula [
46
]. In Fagopyrum, the only other heterostylous species for which a
pistil S-gene has been identified, style length is regulated by S-Locus Early Flowering 3 (S-ELF3) [
78
].
S-ELF3 is a homolog of Arabidopsis ELF3, a nuclear protein that regulates flowering by interacting
with the PIF4 network [
79
]. In each case the activities of pistil S-genes are analogous to at least
some of the actions of red/far-red light on a PIF network. As PIF hubs involve a large number of
interacting factors, there are numerous ways, involving many dierent genes, by which their activities
can be modulated. Heterostyly has evolved frequently and seemingly independently with remarkable
phenotypic consistency [
20
], and the large array of genes that participate in PIF networks provide
attractive potential targets through which this consistency might evolve. The recent development of
transformation systems for heterostylous species [
79
,
80
] has provided key tools to enable testing of
this hypothesis by direct manipulation of AHLs and other components of PIF pathways to investigate
their potential roles in heterostyly.
4. Materials and Methods
4.1. Tissue Collection, RNA Isolation, and RNA-Seq
Turnera subulata (autotetraploid) and T. joelii (2n) were grown in greenhouse conditions. Tissue
samples were chosen to represent stages early (targeting morphology) and late (targeting mating
specificity) in development. For T. subulata, stamens and pistils were collected separately from
3 mm and 16 mm floral buds for each morph. For T. joelii, whole 3 mm and 16 mm flower buds
(sepals removed) were harvested. RNA was isolated using Concert Plant RNA Reagent following the
“Small-Scale Isolation” protocol (Invitrogen, USA). RNA was treated with DNase I (Thermoscientific,
Carlsbad, CA, USA) and cleaned using an RNA-Clean & Concentrator-5 kit (Zymoresearch, Irving
CA, USA) following the manufactures’ protocol. RNA integrity was measured using an RNA Nano
6000 Assay Kit with an Agilent Bioanalyzer 2100 (Agilent Technologies, Santa Clara, CA, USA).
Library construction and sequencing by Illumina PE150 was performed by Novogene (Sacramento,
CA, USA). Paired-end reads with a Phred score of <30 were filtered out of the final data set (Table S1).
Plants 2020,9, 713 15 of 20
Blast2GO-Pro (v5.2.5) was used to map RNA-seq data to a reference T. subulata transcriptome to
quantify transcript abundance and generate read count tables.
4.2. Transcriptome Annotation
A transcriptome of 30,686 genes, predicted from the genome of T. subulata (2n) [Shore et al.
In preparation] was annotated using Blast2GO-Pro (v5.2.5) (NCBI Bioproject PRJNA589060, Genbank
Accession Number GICK01000000). BLASTx searches were restricted to Viridiplantae, default settings
were used for all parameters.
4.3. Quantification of Transcript Abundance and Dierentially Expressed Gene (DEG) Identification
RNAseq reads were mapped to the transcriptome using Blast2GO-Pro (v5.2.5) [
81
] using default
parameters, to generate read count tables for each sample. These were then used to identify dierentially
that were expressed between S- and L-morphs using the general linearized model likelihood method
coupled with relative log expression normalization also using Blast2GO-Pro (v5.2.5), which employs
edgeR for DE analysis [
81
]. Cut-oparameters for DEGs were an absolute log
2
fold change of >1
and a Benjamini–Hochberg adjusted p-value of <0.05. MA plots were generated using Blast2GO-Pro
(v5.2.5) and heatmaps with InCHlib using the Log
2
(CPM) and Z-score option [
82
]. T. subulata DEGs
were cross-referenced to T. joelii DEGs and only those common to both retained. Removed genes were
assessed for the number of read counts mapped from T. joelii to quantify potential interspecies mapping
failure. Genes that shared expression patterns between dierent organs within a developmental stage
were considered to possibly have been designated DEGs as a result of allelic variation and removed (but
later replaced for KEGG and STRING analyses). Blastx was used to identify closest Arabidopsis thaliana
homologs of DEGs for downstream analyses (with cut-os of >35% query coverage and e-value <0.05).
4.4. RT-qPCR Analysis of DEGs
qPCR was performed for the top ten up- and down-regulated genes from each dataset. RNA was
purified and DNase I treated as above. cDNA was synthesized using the SensiFAST cDNA synthesis
kit and manufacture’s protocol (Bioline, UK). All qPCR reactions used a SensiMix SYBR Low-ROX kit
with a slight modification of the manufacturers’ protocol (Bioline, UK). Primers (5
µ
M each) and 20 ng
RNA were combined in 20
µ
L reactions. Samples were run on an ABI 7500 fast machine (ABI, Vernon,
CA, USA) and normalized using UEV1D and
β
tubulin. Results were analyzed using ExpressionSuite
(v1.1) and DataAssist (v3.01) (ThermoFisher, Waltham, MA, USA). A threshold of 36 cycles was used
and p-values adjusted using the Benjamini–Hochberg method. Primer sequences and amplification
conditions can be found in Tables S5 and S6. If qPCR failed because of low expression or similarity to
paralogs preventing gene specific primer design, the next DEG in the list was assayed until ten were
successfully assayed for each category (File S1).
4.5. DE Gene Analysis
To analyze relationships between DEGs, STRING (v11.0) was used to generate association networks
using a low interaction score (0.100), and the A. thaliana database [
58
]. Pistil and stamen DEG data
sets were combined separately to generate tissue specific networks. Networks were uploaded to
the Network Data Exchange (NDEx) repository [
83
] for public access (UUIDs are listed in Table S3).
Networks were clustered using NCMine (v1.3.0), which uses a modified MCODE algorithm for
community detection, to identify functional subnetworks [
59
]. Pistil and stamen networks were
compared using DyNet (v1.0.0) using default settings [
84
]. STRING generated networks were analyzed
with Cytoscape (v3.7.1) coupled with ClueGO (v2.54), to perform more rigorous GO enrichment
analysis and create GO clusters [
85
,
86
]. Genes not mapped in STRING were included for ClueGO
analysis. A. thaliana GO, KEGG, and Reactome databases were used as lines of evidence. ClueGo
parameters used were, GO term levels of 0–20, kappa score of 0.400, and adjusted p-value score of <0.05
for individual terms. p-values were adjusted using the Bonferroni’s step-down method.
Plants 2020,9, 713 16 of 20
4.6. KEGG Analysis
KofamKOALA was used to assign KEGG orthology identifiers and identify associated KEGG
pathways using FASTA files generated through STRING [
87
]. Only significant annotations (p<0.05)
were used in reconstruction of KEGG pathways.
Supplementary Materials:
The following are available online at http://www.mdpi.com/2223-7747/9/6/713/s1.
Table S1: Resulting clean RNA-Seq reads as provided by NovoGene. Table S2: GO terms associated with TsYUC6
in stamens. Table S3: GO terms identified in the reciprocal dataset. Table S4: NDEx UUIDs for STRING generated
pathways. Table S5: Primers for RT-qPCR; Table S6: ABI fast 7500 conditions. Figure S1: MA plots and heatmaps
of DEGs in the four tissue types. Figure S2: Enriched GO terms. Figure S3: Depleted GO terms. Figure S4: KEGG
analysis of Turnera subulata; STRING networks. Figure S5: DEGs in pistils. Figure S6: DEGs in stamens. Figure S7:
First neighbors of BAHD. Figure S8: BAHD Cluster 8. Figure S9: First neighbors of YUC6. Figure S10: YUC6
Cluster 4. Figure S11: YUC6 Cluster 5. Figure S12: YUC6 Cluster 6. Figure S13: YUC6 Cluster 13. Figure S14:
First neighbors of SPH1. Figure S15: SPH1 Cluster 1. File S1: Descriptions of these DEGs, read count tables,
Arabidopsis homologs for each dierential dataset. File S2: Comprehensive lists of the clustered enriched and
depleted GO terms. File S3: Comprehensive list of Enriched and depleted pathways between samples types. File
S4: Comprehensive lists of genes in predicted associations identified in STRING analyses.
Author Contributions:
Conceptualization, A.G.M. and J.S.S.; methodology, A.G.M., J.S.S., and P.M.H.; software,
P.M.H.; validation, P.M.H.; formal analysis, P.M.H.; resources, J.S.S.; data curation, P.M.H.; writing—original
draft preparation, P.M.H.; writing—review and editing, A.G.M. and P.M.H.; visualization, P.M.H. and A.G.M.;
supervision, A.G.M. and J.S.S.; project administration, A.G.M. and J.S.S.; funding acquisition, A.G.M. and J.S.S. All
authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by NSF MCB170034 (AGM) and an NSERC grant (JSS).
Conflicts of Interest: The authors declare no conflict of interest.
Data Archiving:
The RNA-seq data and annotated transcriptome have been submitted to NCBI Bioproject
PRJNA589060. STRING Networks are available from the Network Data Exchange (NDEx) repository, UUIDs can
be found in Table S4).
References
1.
Bateson, W.; Gregory, R.P. On the inheritance of heterostylism in Primula.Proc. R. Soc. Lond. Ser. B
1905
,
76, 581–586.
2.
Barrett, S.C.H.; Cruzan, M.B. Incompatibility in heterostylous plants. In Genetic Control of Self-Incompatibility
and Reproductive Development in Flowering Plants; Williams, E.G., Ed.; Kluwer Academic Publishers: Dordrecht,
the Netherlands, 1994; pp. 189–219.
3.
Lewis, D.; Jones, D.A. The genetics of heterostyly. In Evolution and Function of Heterostyly; Barrett, S.C.H., Ed.;
Springer: New York, NY, USA, 1992; pp. 129–150.
4. Darwin, C. The Dierent Forms of Flowers on Plants of the Same Species; John Murray: London, UK, 1877.
5. Ganders, F.R. The biology of heterostyly. New Zealand J. Bot. 1979,17, 607–635. [CrossRef]
6.
Vuilleumier, B.S. The origin and evolutionary development of heterostyly in the Angiosperms. Evolution
1967,21, 210–226. [CrossRef]
7.
Dulberger, R. Structural dimorphism of stigmatic papillae in distylous Linum species. Am. J. Bot.
1974
,
61, 238–243. [CrossRef]
8.
Dulberger, R. Floral polymorphisms and their functional significance in the heterostylous syndrome.
In Evolution and Function of Heterostyly; Barrett, S.C.H., Ed.; Springer: New York, NY, USA, 1992; pp. 41–84.
9.
Orndu, R. Pollen flow in a population of Primula vulgaris Huds. Bot. J. Linn. Soc.
1979
,78, 1–10. [CrossRef]
10. Dowrick, V.P. Heterostyly and homostyly in Primula obconica.Heredity 1956,10, 219–236.
11.
Charlesworth, D.; Charlesworth, B. A model for the evolution of distyly. Am. Nat.
1979
,114, 467–498.
[CrossRef]
12. Thompson, M.J.; Jiggins, C.D. Supergenes and their role in evolution. Heredity 2014,113, 1–8. [CrossRef]
13.
Charlesworth, D. The status of supergenes in the 21st century, recombination suppression in Batesian mimicry
and sex chromosomes and other complex adaptations. Evol. Appl. 2016,9, 74–90. [CrossRef]
14.
Fisher, R.A. The theoretical consequences of polyploid inheritance for the mid style form in Lythrum salicaria.
Ann. Eugen. 1941,11, 31–38. [CrossRef]
15. Crosby, J.L. Selection of an unfavourable gene-complex. Evolution 1949,3, 212–230. [CrossRef] [PubMed]
Plants 2020,9, 713 17 of 20
16.
Bodmer, W.F. The genetics of homostyly in Primula vulgaris. Phil. Trans. R. Soc. Lond. Ser. B Biol. Sci.
1960
,
242, 517–549.
17.
Piper, J.G.; Charlesworth, B.; Charlesworth, D. Breeding system and evolution in Primula vulgaris and the
role of reproductive assurance. Heredity 1986,56, 207–217. [CrossRef]
18.
Barrett, S.C.H.; Jesson, L.K.; Baker, A.M. The evolution of stylar polymorphisms in plants. Ann. Bot.
2000
,85
(Suppl. SA), 253–265. [CrossRef]
19.
Naiki, A. Heterostyly and the possibility of its breakdown by polyploidization. Plant Species Biol.
2012
,
27, 3–29. [CrossRef]
20.
Barrett, S.C.H. ‘A most complex marriage arrangement’, recent advances on heterostyly and unresolved
questions. New Phytol. 2019,224, 1051–1067. [CrossRef]
21.
Huu, C.N.; Kappel, C.; Keller, B.; Sicard, A.; Takebayashi, Y.; Breuninger, H.; Nowak, M.D.; Bäurle, I.;
Himmelbach, A.; Burkart, M.; et al. Presence versus absence of CYP734A50 underlies the style-length
dimorphism in primroses. eLife 2016,5, e17956. [CrossRef]
22.
Nowak, M.D.; Russo, G.; Schlapbach, R.; Huu, C.N.; Lenhard, M.; Conti, E. The draft genome of Primula veris
yields insights into the molecular basis of heterostyly. Genome Biol. 2015,16, 12. [CrossRef]
23.
Li, J.; Cocker, J.M.; Wright, J.; Webster, M.A.; McMullan, M.; Dyer, S.; Swarbreck, D.; Caccamo, M.;
Oosterhout, C.V.; Gilmartin, P.M. Genetic architecture and evolution of the Slocus supergene in Primula
vulgaris.Nat. Plants 2016,2, 16188. [CrossRef]
24.
Burrows, B.A.; McCubbin, A.G. Sequencing the genomic regions flanking PvGLO sequences confirms the
presence of two GLO loci; one of which lies adjacent to the style-length determinant gene CYP734A50.
Plant Reprod. 2017,30, 53–67. [CrossRef]
25.
Ushijima, K.; Nakano, R.; Bando, M.; Shigezane, Y.; Ikeda, K.; Namba, Y.; Kume, S.; Kitabata, T.; Mori, H.;
Kubo, Y. Isolation of the floral morph-related genes in heterostylous flax (Linum grandiflorum), the genetic
polymorphism and the transcriptional and post-transcriptional regulations of the Slocus. Plant J.
2012
,
69, 317–331. [CrossRef] [PubMed]
26.
Ushijima, K.; Ikeda, K.; Nakano, R.; Matsubara, M.; Tsuda, Y.; Kubo, Y. Genetic control of floral morph and
petal pigmentation in Linum grandiflorum Desf.; a heterostylous flax. Hort. J. 2015,84, 261–268. [CrossRef]
27.
Yasui, Y.; Mori, M.; Aii, J.; Abe, T.; Masumoto, D.; Sato, S.; Hayashi, Y.; Ohnishi, O.; Ota, T. S-LOCUS
EARLY FLOWERING 3 is exclusively present in the genomes of short-styled buckwheat plants that exhibit
heteromorphic self-incompatibility. PLoS ONE 2012,7, 1–9. [CrossRef] [PubMed]
28.
Yasui, Y.; Hirakawa, H.; Ueno, M.; Matsui, K.; Katsube-Tanaka, T.; Yang, S.J.; Aii, J.; Sato, S.; Mori, M.
Assembly of the draft genome of buckwheat and its applications in identifying agronomically useful genes.
DNA Res. 2016,23, 215–224. [CrossRef]
29.
Shore, J.S.; Hamam, H.J.; Chafe, P.D.J.; Labonne, J.D.J.; Henning, P.M.; McCubbin, A.G. The long and short of
the S-locus in Turnera (Passifloraceae). New Phytol. 2019. [CrossRef]
30. Takayama, S.; Isogai, A. Self-incompatibility in plants. Ann. Rev. Plant. Biol. 2006,56, 467–489. [CrossRef]
31.
Hogenboom, N.G. Incompatibility and incongruity, two dierent mechanisms for the non-functioning of
intimate partner relationships. Proc. Roy. Soc. Lond. B 1975,188, 361–375.
32.
Wang, M.; Liu, X.; Wang, R.; Wanchun, L.; Rodermel, S.; Yu, F. Overexpression of a putative Arabidopsis BAHD
acyltransferase causes dwarfism that can be rescued by brassinosteroid. J. Exp. Bot.
2012
,63, 5787–5801.
[CrossRef]
33.
D’Auria, J.C. Acyltransferases in plants, a good time to be BAHD. Curr. Opt. Plant Biol.
2006
,9, 331–340.
[CrossRef]
34.
Cheng, Y.; Dai, X.; Zhao, Y. Auxin biosynthesis by the YUCCA flavin monooxygenases controls formation of
floral organs and vascular tissues in Arabidopsis. Gene Dev. 2006,20, 1790–1799. [CrossRef]
35.
Foote, H.C.C.; Ride, J.P.; Franklin-Tong, V.E.; Walker, E.A.; Lawrence, Z.; Franklin, C.H. Cloning and
expression of a distinctive class of self-incompatibility (S) gene from Papaver rhoeas L. Proc. Natl. Acad.
Sci. USA 1994,91, 2265–2269. [CrossRef] [PubMed]
36.
Kakeda, K.; Jordan, N.D.; Conner, A.; Ride, J.P.; Franklin-Tong, V.E.; Franklin, F.C.H. Identification of residues
in a hydrophilic loop of the Papaver rhoeas S protein that play a crucial role in recognition of incompatible
pollen. Plant Cell 1998,10, 1723–1731. [CrossRef] [PubMed]
37.
Ride, J.P.; Davies, E.M.; Franklin, F.C.H.; Marshall, D.F. Analysis of Arabidopsis genome sequence reveals a
large new gene family in plants. Plant Mol. Biol. 1999,39, 927–932. [CrossRef] [PubMed]
Plants 2020,9, 713 18 of 20
38.
Wilkins, K.A.; Bosch, M.; Haque, T.; Teng, N.; Poulter, N.S.; Franklin-Tong, V.E. Self-incompatibility-induced
programmed cell death in field poppy pollen involves dramatic acidification of the incompatible pollen tube
cytosol. Plant Physiol. 2005,167, 766–779. [CrossRef] [PubMed]
39.
De Graaf, B.H.J.; Vatovec, S.; Ju
á
rez-Diaz, J.A.; Chai, L.; Kooblall, K.; Wilkins, K.; Zou, H.; Forbes, T.;
Franklin, F.C.H.; Franklin-Tong, V.E. The Papaver self-incompatibility pollen S-determinant, PrpS, functions
in Arabidopsis thaliana.Curr. Biol. 2012,22, 154–159. [CrossRef]
40.
McCubbin, A.G.; Lee, C.; Hetrick, A. Identification of genes showing dierential expression between morphs
in developing flowers of Primula vulgaris.Sex Plant Reprod. 2006,19, 63–72. [CrossRef]
41.
Athanasiou, A.; Khosravi, D.; Tamari, F.; Shore, J.S. Characterization and localization of short-specific
polygalacturonase in distylous Turnera subulata (Turneraceae). Am. J. Bot. 2003,90, 675–682. [CrossRef]
42.
Khosravi, D.; Yang, E.C.C.; Siu, K.W.M.; Shore, J.S. High level of
α
-dioxygenase in short styles of distylous
Turnera species. I. J. Plant Sci. 2004,165, 995–1006. [CrossRef]
43.
Takeshima, R.; Nishio, T.; Komatsu, S.; Kurauchi, N.; Matsui, K. Identification of a gene encoding polygalacturonase
expressed specifically in short styles in distylous common buckwheat (Fagopyrum esculentum). Heredity
2019
,123,
492–502. [CrossRef]
44.
Cohen, J.I. De novo sequencing and comparative transcriptomics of floral development of the distylous
species Lithospermum multiflorum.Front. Plant Sci. 2016,7, 1934. [CrossRef] [PubMed]
45.
Lu, H.; Zou, Y.; Feng, N. Overexpression of AHL20 negatively regulates defenses in Arabidopsis.J. Int.
Plant Biol. 2010,52, 801–808. [CrossRef] [PubMed]
46.
Zhao, Z.; Luo, Z.; Yuan, S.; Mei, L.; Zhang, D. Global transcriptome and gene co-expression network analyses
on the development of distyly in Primula oreodoxa.Heredity 2019. [CrossRef] [PubMed]
47.
Tamari, F.; Khosravi, D.; Hilliker, A.J.; Shore, J.S. Inheritance of spontaneous mutant homostyles in Turnera subulata
xkrapovickasii and in autotetraploid T. scabra (Turneraceae). Heredity 2005,94, 207–216. [CrossRef] [PubMed]
48.
Labonne, J.D.J.; Hilliker, A.J.; Shore, J.S. Meiotic recombination in Turnera (Turneraceae), Extreme sexual
dierence in rates; but no evidence for recombination suppression associated with the distyly (S) locus.
Heredity 2007,98, 411–418. [CrossRef]
49.
Labonne, J.D.J.; Vaisman, A.; Shore, J.S. Construction of a first genetic map of distylous Turnera and a
fine-scale map of the S-locus region. Genome 2008,51, 471–478. [CrossRef]
50.
Labonne, J.D.J.; Goultiaeva, A.; Shore, J.S. High-resolution mapping of the S-locus in Turnera leads to the
discovery of three genes tightly associated with the S-alleles. Mol. Genet. Genom.
2009
,281, 673–685.
[CrossRef]
51.
Labonne, J.D.J.; Tamari, F.; Shore, J.S. Characterization of X-ray generated floral mutants carrying deletions
at the S-locus of distylous Turnera subulata.Heredity 2010,105, 235–243. [CrossRef]
52.
Labonne, J.D.J.; Shore, J.S. Positional cloning of the s haplotype determining the floral and incompatibility
phenotype of the long-styled morph of distylous Turnera subulata.Mol. Genet. Genom.
2011
,285, 101–111.
[CrossRef]
53.
Truyens, S.; Arbo, M.M.; Shore, J.S. Phylogenetic relationships, chromosome and breeding system evolution
in Turnera (Turneraceae): Inferences from ITS sequence data. Am. J. Bot. 2005,92, 1749–1758. [CrossRef]
54.
Rocha, L.; Ribeiro, P.L.; Endress, P.K.; Rapini, A. A brainstorm on the systematics of Turnera (Turneraceae,
Malpighiales) caused by insights from molecular phylogenetics and morphological evolution. Mol. Phylogenet.
Evol. 2019,137, 44–63. [CrossRef]
55.
Liao, H.; Tang, R.; Zheng, X.; Luan, S.; Yu, F. FERONIA receptor kinase at the crossroads of hormone signaling
and stress responses. Plant Cell Physiol. 2017,58, 1143–1150. [CrossRef] [PubMed]
56.
Zhao, Y. Auxin Biosynthesis, a simple two-step pathway converts tryptophan to indole-3-acetic acid in plants.
Mol. Plant 2012,5, 334–338. [CrossRef] [PubMed]
57.
Cecchetti, V.; Altamura, M.M.; Falasca, G.; Costantino, P.; Cardarelli, M. Auxin regulates Arabidopsis anther
dehiscence; pollen maturation; and filament elongation. Plant Cell
2008
,20, 1760–1774. [CrossRef] [PubMed]
58.
Szklarczyk, D.; Gable, A.L.; Lyon, D.; Junge, A.; Wyder, S.; Huerta-Cepas, J.; Simonovic, M.; Doncheva, N.T.;
Morris, J.H.; Bork, P.; et al. STRING v11, protein-protein association networks with increased coverage;
supporting functional discovery in genome-wide experimental datasets. Nucl. Acids Res.
2019
,47, D607–D613.
[CrossRef] [PubMed]
59.
Tadaka, S.; Kinoshita, K. NCMine Core-peripheral based functional module detection using near-clique
mining. Bioinformatics 2016,32, 3454–3460. [CrossRef] [PubMed]
Plants 2020,9, 713 19 of 20
60.
Yoshimitsu, Y.; Tanaka, K.; Fukuda, W.; Asami, T.; Yoshida, S.; Hayashi, K.-I.; Kamiya, Y.; Jikumaru, Y.;
Shigeta, T.; Nakamura, Y.; et al. Transcription of DWARF4 Plays a crucial role in auxin-regulated root
elongation in addition to brassinosteroid homeostasis in Arabidopsis thaliana. PLoS ONE
2011
,6, e23851.
[CrossRef]
61.
Street, I.H.; Shah, P.K.; Smith, A.M.; Avery, N.; Ne, M.M. The AT-hook-containing proteins SOB3/AHL29 and
ESC/AHL27 are negative modulators of hypocotyl growth in Arabidopsis.Plant J.
2008
,54, 1–14. [CrossRef]
62.
Zhao, J.; Favero, D.S.; Peng, H.; Ne, M.M. Arabidopsis thaliana AHL family modulates hypocotyl growth
redundantly by interacting with each other via the PPC/DUF296 domain. Proc. Natl. Acad. Sci. USA
2013
,
110, E4688–E4697. [CrossRef]
63.
Lee, K.; Seo, P.J. Coordination of matrix attachment and ATP-dependent chromatin remodeling regulate
auxin biosynthesis and Arabidopsis hypocotyl elongation. PLoS ONE 2017,12, e0181804. [CrossRef]
64.
Favero, D.S.; Le, K.N.; Ne, M.M. Brassinosteroid signaling converges with SUPPRESSOR OF
PHYTOCHROME B4-#3 to influence the expression of SMALL AUXIN UP RNA genes and hypocotyl growth.
Plant J. 2017,89, 1133–1145. [CrossRef]
65.
Kaneda, M.; Schuetz, M.; Lin, B.S.; Chanis, C.; Hamberger, B.; Western, T.L.; Ehiting, J.; Samuels, A.L.
ABC transporters coordinately expressed during lignification of Arabidopsis stems include a set of ABCBs
associated with auxin transport. J. Exp. Bot. 2011,62, 2063–2077. [CrossRef] [PubMed]
66.
Tabata, R.; Sumida, K.; Yoshii, T.; Ohyama, K.; Shinohara, H.; Matsubayashi, Y. Perception of root derived
peptides by shoot LRR-RKs mediates systemic N-demand signaling. Science
2014
,17, 343–346. [CrossRef]
[PubMed]
67.
Campbell, L.; Turner, S.R. A comprehensive analysis of RALF proteins in green plants suggest there are two
distinct functional groups. Front. Plant Sci. 2017,8, 37. [CrossRef] [PubMed]
68.
Haruta, M.; Sabat, G.; Stecker, K.; Minko, B.B.; Sussman, M.R. A peptide hormone and its receptor protein
kinase regulate plant cell expansion. Science 2014,343, 408–411. [CrossRef] [PubMed]
69.
Larsson, E.; Vivian-Smith, A.; Oringa, R.; Sundberg, E. Auxin homeostasis in Arabidopsis ovules is
anther-dependent at maturation and changes dynamically upon fertilization. Front. Plant Sci.
2017
,8, 1735.
[CrossRef]
70.
Yao, X.; Tian, L.; Yang, J.; Zhao, Y.; Zhu, Y.; Dai, X.; Zhao, Y.; Yang, Z.-N. Auxin production in diploid
microsporocytes is necessary and sucient for early stages of pollen development. PLoS Genet.
2018
,
14, e1007397. [CrossRef]
71.
Goda, H.; Sawa, S.; Asami, T.; Fujioka, F.; Shimada, Y.; Yoshida, S. Comprehensive comparison of
auxin-regulated and brassinosteroid-regulated genes in Arabidopsis.Plant Physiol.
2004
,134, 1555–1573.
[CrossRef]
72.
Matsushita, A.; Furumoto, T.; Ishida, S.; Takahashi, Y. AGF1; an AT-hook protein; is necessary for the negative
feedback of AtGA3ox1 encoding GA 3-oxidase. Plant Physiol. 2007,143, 1152–1162. [CrossRef]
73.
Endt, D.V.; Silva, M.S.E.; Kijne, J.W.; Pasquali, G.; Memelink, J. Identification of a bipartite
jasmonate-responsive promoter element in the Catharanthus roseus ORCA3 transcription factor gene that
interacts specifically with AT-hook DNA-binding proteins. Plant Physiol. 2007,144, 1680–1689. [CrossRef]
74.
Rashotte, A.M.; Carson, S.D.; To, J.P.; Kieber, J.J. Expression profiling of cytokinin action in Arabidopsis.
Plant Physiol. 2003,132, 1998–2011. [CrossRef]
75.
Weinig, C. Phytochrome photoreceptors mediate plasticity to light quality in flowers fo the Brassicaceae.
Am. J. Bot. 2002,89, 230–235. [CrossRef] [PubMed]
76.
Paik, I.; Kathare, P.K.; Kim, J.I.; Huq, E. Expanding roles of PIFs in signal integration from multiple processes.
Mol. Plant 2017,10, 1035–1046. [CrossRef] [PubMed]
77.
Xiao, C.; Chen, F.; Yu, X.; Lin, C.; Fu, Y.F. Over-expression of an AT-hook gene, AHL22, delays flowering and
inhibits the elongation of the hypocotyl in Arabidopsis thaliana.Plant Mol. Biol.
2009
,71, 39–50. [CrossRef]
[PubMed]
78.
Nieto, C.; Lopez-Salmeron, V.; Daviere, J.M.; Prat, S. ELF3-PIF4 interaction regulates plant growth
independently of the evening complex. Curr. Biol. 2015,25, 187–193. [CrossRef] [PubMed]
79.
Hayta, S.; Smedley, M.A.; Li, J.; Harwood, W.A.; Gilmartin, P.M. Plant Regeneration from Leaf-derived Callus
Cultures of Primrose (Primula vulgaris). HortScience 2016,51, 558–562. [CrossRef]
Plants 2020,9, 713 20 of 20
80.
Chafe, P.D.J.; Lee, T.; Shore, J.S. Development of a genetic transformation system for distylous Turnera
joelii (Passifloraceae) and characterization of a self-compatible mutant. Plant Cell Tissue Organ Cult.
2015
,
120, 507–517. [CrossRef]
81.
Robinson, M.D.; McCarthy, D.J.; Smyth, G.K. edgeR, a Bioconductor package for dierential expression
analysis of digital gene expression data. Bioinformatics 2010,26, 139–140. [CrossRef]
82.
Skuta, C.; Bartunek, P.; Svozil, D. InCHlib—Interactive cluster heatmap for web applications. J. Cheminform.
2014,6, 44. [CrossRef]
83.
Pratt, D.; Chen, J.; Welker, D.; Rivas, R.; Pillich, R.; Rynkov, V.; Ono, K.; Miello, C.; Hicks, L.; Szalma, S.; et al.
NDEx, the Network Data Exchange. Cell Syst. 2015,1, 302–305. [CrossRef]
84.
Goenawan, I.H.; Bryan, K.; Lynn, D.J. DyNet, visualization and analysis of dynamic molecular interaction
networks. Bioinformatics 2016,32, 2713–2715. [CrossRef]
85.
Bindea, G.; Mlecnik, B.; Hackl, H.; Charoentong, P.; Tosolini, M.; Kirilovsky, A.; Fridman, W.H.; Pag
è
s, F.;
Galon, J. ClueGO, a Cytoscape plug-in to decipher functionally grouped gene ontology and pathway
annotation networks. Bioinformatics 2009,20, 1091–1093. [CrossRef] [PubMed]
86.
Shannon, P.; Markiel, A.; Ozier, O.; Baliga, N.S.; Wang, J.T.; Ramage, D.; Schwikowski, B.; Ideker, T. Cytoscape,
a software environment for integrated models of biomolecular interaction networks. Genome Res.
2003
,
13, 2498–2504. [CrossRef] [PubMed]
87.
Aramaki, T.; Blanc-Mathieu, R.; Endo, H.; Ohkubo, K.; Kanehisa, M.; Goto, S.; Ogata, H. KofamKOALA,
KEGG ortholog assignment based on profile HMM and adaptive score threshold. bioRxiv 2019. [CrossRef]
[PubMed]
©
2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).
... The basis of style length and female mating type in both Primula and Turnera is inactivation of BR. Indirect alterations to the Phytochrome Interacting Factor (PIF) signaling hubs via inactivation of BR or direct alteration of PIF signaling hubrelated genes may be the basis of repression of style elongation in Fagopyrum, Primula, and Turnera [16], a hypothesis that has been computationally supported [11]. Auxin appears to play a role in stamen elongation in Fagopyrum [11], potentially similar to filament elongation in Turnera, as suggested by computational analysis [16]. ...
... Indirect alterations to the Phytochrome Interacting Factor (PIF) signaling hubs via inactivation of BR or direct alteration of PIF signaling hubrelated genes may be the basis of repression of style elongation in Fagopyrum, Primula, and Turnera [16], a hypothesis that has been computationally supported [11]. Auxin appears to play a role in stamen elongation in Fagopyrum [11], potentially similar to filament elongation in Turnera, as suggested by computational analysis [16]. The commonality in the evolution of the S-locus in Linum and Primula suggests a further convergent nature at the genomic level. ...
... SPH1, a member of the S-protein homolog family, likely controls filament length [3]. YUC6, a member of the YUCCA flavin monooxygenase family, controls male mating-type [15] likely via alterations of auxin concentration in the S-morph relative to the L-morph [16]. Currently, it is unknown how the S-locus evolved in Turnera. ...
Article
Full-text available
A majority of Turnera species (Passifloraceae) exhibit distyly, a reproductive system involving both self-incompatibility and reciprocal herkogamy. This system differs from self-incompatibility in Passiflora species. The genetic basis of distyly in Turnera is a supergene, restricted to the S-morph, and containing three S-genes. How supergenes and distyly evolved in Turnera, and the other Angiosperm families exhibiting distyly remain largely unknown. Unraveling the evolutionary origins in Turnera requires the generation of genomic resources and extensive phylogenetic analyses. Here, we present the annotated draft genome of the S-morph of distylous Turnera subulata. Our annotation allowed for phylogenetic analyses of the three S-genes’ families across 56 plant species ranging from non-seed plants to eudicots. In addition to the phylogenetic analysis, we identified the three S-genes’ closest paralogs in two species of Passiflora. Our analyses suggest that the S-locus evolved after the divergence of Passiflora and Turnera. Finally, to provide insights into the neofunctionalization of the S-genes, we compared expression patterns of the S-genes with close paralogs in Arabidopsis and Populus trichocarpa. The annotation of the T. subulata genome will provide a useful resource for future comparative work. Additionally, this work has provided insights into the convergent nature of distyly and the origin of supergenes.
... In recent years, three S-locus genes (hemizygous in S-morphs and absent in L-morphs) have been identified: TsSPH1, likely involved in filament elongation; TsYUC6 (a member of the YUCCA gene family), involved in auxin biosynthesis and likely controlling pollen development; and TsBAHD, inactivating brassinosteroids and controlling both female incompatibility and style length, similarly to CYP T in Primula 20,21,31 . Differentially-expressed genes (DEGs) between L-and S-morphs of Turnera subulata flowers have been characterized in a recent study 22 . Among these DEGs were several genes related to the phytochrome interacting factor (PIF) signaling network, a large and highly interconnected network that mediates several plant morphogenetics processes, whose key regulators are the PIF transcription factors 32 . ...
... Among these DEGs were several genes related to the phytochrome interacting factor (PIF) signaling network, a large and highly interconnected network that mediates several plant morphogenetics processes, whose key regulators are the PIF transcription factors 32 . This observation, together with the fact that phytochrome-associated pathways can modulate the morphology of sexual organs in Brassica rapa when exposed to different red:far-red ratios 33 , led the authors to propose that the recruitment of genes from the PIF network might represent a commonality among S-loci controlling distyly in different species 22 . However, this hypothesis has never been tested. ...
... Third, a comparative transcriptomic analysis among P. veris, F. esculentum and T. subulata identified, for the first time, homologous genes that are differentially expressed between L-and S-morphs, thus potentially involved in controlling distyly, in all three species. Last, we tested the previously proposed hypothesis that S-loci evolved by recruiting genes from the phytochrome-interacting factor (PIF) signaling network 22 . The results presented here address fundamental questions on the function and evolution of distyly by identifying genes involved in the expression of distyly in Primula and F. esculentum and revealing commonalities that might explain the convergent evolution of distyly in these two species and T. subulata. ...
Article
Full-text available
Distyly, a floral dimorphism associated with heteromorphic self-incompatibility and controlled by the S- locus supergene, evolved independently multiple times. Comparative analyses of the first transcriptome atlas for the main distyly model, Primula veris , with other distylous species produced the following findings. A set of 53 constitutively expressed genes in P. veris did not include any of the housekeeping genes commonly used to normalize gene expression in qPCR experiments. The S- locus gene CYP T acquired its role in controlling style elongation via a change in expression profile. Comparison of genes differentially expressed between floral morphs revealed that brassinosteroids and auxin are the main hormones controlling style elongation in P. veris and Fagopyrum esculentum , respectively. Furthermore, shared biochemical pathways might underlie the expression of distyly in the distantly related P. veris , F. esculentum and Turnera subulata , suggesting a degree of correspondence between evolutionary convergence at phenotypic and molecular levels. Finally, we provide the first evidence supporting the previously proposed hypothesis that distyly supergenes of distantly related species evolved via the recruitment of genes related to the phytochrome-interacting factor (PIF) signaling network. To conclude, this is the first study that discovered homologous genes involved in the control of distyly in distantly related taxa.
... BAHD, a member of the BAHD acyltransferase family, pleiotropically controls both female morphology and the female mating type [19,20]. Interestingly, like Primula CYP734A50, Turnera BAHD establishes female phenotypes by inactivating BR in S-morph pistils, illustrating a common molecular basis behind pistil phenotypes in these two genera [20,21]. SPH1 is predicted to be involved in filament elongation; this gene is not expressed in short homostyle mutants [9]. ...
... Finally, YUC6 is hypothesized to determine the male mating type [9]. Based on homology, YUC6 is predicted to be a member of the YUCCA gene family of flavin-containing monooxygenases, which are involved in synthesizing the plant hormone auxin, specifically indole-3-acetic acid (IAA) [9]; previous RNAseq analysis identified several auxin-related genes as differentially expressed between the anthers of two floral morphs of distylous T. subulata [21]. ...
... In addition to the identification of S-genes, work in Fagopyrum [26], Primula [27][28][29], Turnera [21,30,31], and also Lithospermum [32] suggest that genes outside of the S-loci are likely regulated by the S-gene products and involved in various aspects of distyly. Hence, the molecular basis of distyly likely involves a cascade of changes in transcription to produce the overall phenotypes observed. ...
Article
Full-text available
In heterostylous, self-incompatible Turnera species, a member of the YUCCA gene family, YUC6, resides at the S-locus and has been hypothesized to determine the male mating type. YUCCA gene family members synthesize the auxin, indole-3-acetic acid, via a two-step process involving the TAA gene family. Consequently, it has been speculated that differences in auxin concentration in developing anthers are the biochemical basis underlying the male mating type. Here, we provide empirical evidence that supports this hypothesis. Using a transgenic knockdown approach, we show that YUC6 acts pleiotropically to control both the male physiological mating type and pollen size, but not the filament length dimorphism associated with heterostyly in Turnera. Using qPCR to assess YUC6 expression in different transgenic lines, we demonstrate that the level of YUC6 knockdown correlates with the degree of change observed in the male mating type. Further assessment of YUC6 expression through anther development, in the knockdown lines, suggests that the male mating type is irreversibly determined during a specific developmental window prior to microsporogenesis, which is consistent with the genetically sporophytic nature of this self-incompatibility system. These results represent the first gene controlling male mating type to be characterized in any species with heterostyly.
... First BRs have the ability to directly affect pollen germination and tube growth [30], hence presence/absence of active BR could potentially affect pollen germination and/or tube growth of the mating types differentially, creating a mechanism of incompatibility. BRs also have the potential to affect gene expression, however, and as we have previously reported there is considerable differential gene expression between L-and S-morph pistils necessarily caused by the difference in the abundance of active BRs [31], hence BR mediated differential gene expression in the pistils could also mediate mating type. To discriminate between these two possibilities, we assessed the effects of adding BRs to Turnera pollen germinated in vitro. ...
... The alternative way by which BRs might generate a self-incompatibility mechanism is through their effects on transcription. A large number of genes have been reported to be differentially expressed between the pistils of the flower morphs in Turnera [31] and the results presented suggest that this differential expression likely underlies self-incompatibility, presumably by changing the physiology of the stigma and stylar transmitting tract. Interestingly this provides support for a long standing hypothesis, that self-incompatibility in heterostyly is more analogous to mechanisms hypothesized for incongruity-interspecies barriers, than homomorphic self-incompatibility systems [38]. ...
Article
Full-text available
Heterostyly is a breeding system that promotes outbreeding through a combination of morphological and physiological floral traits. In Turnera these traits are governed by a single, hemizygous S-locus containing just three genes. We report that the S-locus gene, BAHD, is mutated and encodes a severely truncated protein in a self-compatible long homostyle species. Further, a self-compatible long homostyle mutant possesses a T. krapovickasii BAHD allele with a point mutation in a highly conserved domain of BAHD acyl transferases. Wild type and mutant TkBAHD alleles were expressed in Arabidopsis to assay for brassinosteroid (BR) inactivating activity. The wild type but not mutant allele caused dwarfism, consistent with the wild type possessing, but the mutant allele having lost, BR inactivating activity. To investigate whether BRs act directly in self-incompatibility, BRs were added to in vitro pollen cultures of the two mating types. A small morph specific stimulatory effect on pollen tube growth was found with 5 µM brassinolide, but no genotype specific inhibition was observed. These results suggest that BAHD acts pleiotropically to mediate pistil length and physiological mating type through BR inactivation, and that in regard to self-incompatibility, BR acts by differentially regulating gene expression in pistils, rather than directly on pollen.
... Scale bar: 5mm (B) Quantification of flower phenotypes of SS plants compared to homozygous recessive L-morphs (ss) and heterozygous S-morphs (Ss). (C-E) Comparison of pollen size of L-morph (C), heterozygous S-morph (D) and homozygous dominant S-morph (E); Scale bar: 25 µm; The letters indicate significant differences as determined by a Tukey's HSD test4.4.4 BR levels are not causal for style length differences in Amsinckia spectabilisBrassinosteroids play a role in determining the style length in other heterostylous species, as it was shown for CYP734A50 in Primula 235 and for TsBAHD in Turnera subulata273 . Both factors act as BR inactivators that cause decreased style growth and are therefore responsible for the short style in S-morph plants. ...
Thesis
Der Lebenszyklus von Pflanzen ist geprägt von sich wiederholenden Wachstums- und Entwicklungsphasen, die auf wiederkehrenden Abläufen, bestehend aus Zellteilung, Zellvergrößerung und Zelldifferenzierung, basieren. Diese Dissertation ist aus zwei Projekten aufgebaut, die sich beide mit unterschiedlichen Blickwinkeln des Zellwachstums beschäftigen. Im ersten steht die Charakterisierung einer Arabidopsis thaliana Mutante, die eine generelle Zellvergrößerung aufweist, im Vordergrund. Das zweite fokussiert sich auf zwei natürlich vorkommende Blütenmorphologien in Amsinckia spectabilis (Boraginaceae), die sich, aufgrund von Zelllängenunterschieden, in Griffellänge und Höhe der Staubblattposition unterscheiden. Es wurde gezeigt, dass die EMS-Mutante eop1 durch größere Zellen 26% größere Blütenblätter aufweist. Außerdem wurden weitere Phänotypen beschrieben, wie zum Beispiel, vergrößerte Kotyledonen, (ebenfalls aufgrund von Zellvergrößerung), Fruchtblätter, Kelchblätter, Rosettenblätter und Pollen. Die Gesamtwuchshöhe der Mutante zeigte sich ebenfalls erhöht und zusätzliche Trichomäste erklärten den haarigen Phänotyp. Feinkartierung enthüllte eine C zu T Transition des letzten Nukleotids des Introns 7 des INCURVATA 11 (ICU11) Gens, einer 2-oxoglutarat/Fe(II)-abhängigen Dioxygenase, als ursächlichen SNP, welcher missgespleißte mRNA verursacht. Zwei T-DNA Insertionslinien (icu11-2 & icu11-4), ebenfalls mit vergrößerten Blütenblättern, bestätigten ICU11 als kausales Gen, und erlaubten somit die Analyse von drei verschiedenen icu11 Allelen. Ein Vergleich der verursachten molekularen Veränderung durch die jeweiligen Mutationen ermittelte Unterschiede in den drei Mutanten, wie zum Beispiel Überexpression von ICU11, als auch die Modifikation von ICU11 mRNA. Zusammen bildete das die Grundlage für die Untersuchung des molekularen Mechanismus, der für den beobachteten Phänotyp verantwortlich ist. Verschiedene Ansätze ermittelten widersprüchliche Ergebnisse hinsichtlich der Proteinfunktion von ICU11 in den drei Mutanten. So zeigte eine Komplementierungsanalyse, dass alle drei Mutationen austauschbar sind, was, zusammen mit der Beobachtung, dass eine ICU11 Überexpression im Wildtyp zu einem icu11-ähnlichen Phänotyp zeigte, dazu führte, dass die icu11 Mutanten als gain-of-function Mutationen eingeordnet wurden. Im Widerspruch dazu stand die Entdeckung, dass sich icu11-4 durch ein genomisches ICU11 Transgen retten ließ. So wurde ein Model, basierend auf der Annahme, dass eine ICU11 Überexpression die Proteinfunktion ebenso hemmt wie ein nichtfunktionales Protein, vorgeschlagen. Außerdem wurde eine erhöhte Resistenz der icu11-3 (eop1) gegenüber Paclobutrazol, einem Gibberellin (GA)-Inhibitor, und die Aktivierung der Expression von AtGA20ox2, einem Haupt-GA-Biosynthese-Gen, festgestellt. Zusätzlich wurde eine zytoplasmatische Lokalisation von ICU11 detektiert, sodass ein Einfluss von ICU11 auf die GA- Biosynthese und somit auf das Gesamt-GA-Level angenommen wird, der den beobachteten (GA-überdosierten) Phänotyp erklären könnte. Das zweite Projekt strebte die Identifizierung der genetischen Grundlage des S-Locus in Amsinckia spectabilis an, da die Gattung Amsinckia einige untypische Charakteristiken für eine heterostyle Art, wie zum Beispiel das Fehlen einer offensichtlichen Selbstinkompatibilität (SI), sowie die mehrmalige Entwicklung zu Homostyly und 100% autonomem Selbsten, aufweist. Die Analyse basierte auf drei Amsinckia spectabilis Varianten: einer heterostylen Form, bestehend aus zwei Blütenmorphologien mit gegensätzlich positionierten Sexualorganen (S-Morph: hohe Staubblattposition und kurzer Griffel und L-Morph: niedrige Staubblattansätze und langer Griffel), und zwei homostylen Formen, einer großblütigen teilweise selbstenden und einer kleinblütigen voll selbstenden. Natürliche Populationen weisen ungefähr ein 1:1 S:L Morph-Verhältnis auf, welches sich durch vorherrschend disassortative Paarung beider Morphs erklären lasst. Dadurch kann das dominante S-Allel ausschließlich heterozygot auftreten (heterozygot (Ss) im S-morph und homozygot rezessiv (ss) im L-morph). Die Suche nach Morph-spezifischen Phänotypen offenbarte 56% längere L-Morph Griffel und 58% höhere S-Morph Staubblattansätze. Zusätzlich wurden 21% größere S-Morph Pollen, sowie das Fehlen einer offensichtlichen SI gefunden. Dies war die Grundlage für die Annahme, dass der Amsinckia spec. S-Locus mindestens aus G- (Griffel), A- (Staubblatt) und P- (Pollen) Locus besteht. Vergleichende Transkriptom-Analyse beider Morphs offenbarte 22 unterschiedlich exprimierte Marker, die in 2 Contigs der PacBio Genom-Assemblierung eines SS-Individuums lokalisiert werden konnten. Dies erlaubte die genetische Einengung des S-Locus auf einen Bereich von circa 23 Mb. Gegensätzlich zu bisher aufgeklärten S-Loci in anderen Pflanzenarten konnte kein Hinweis auf eine hemizygote Region gefunden werden, die die supprimierte Rekombination am S-Locus erklären könnte, sodass eine Inversion als Ursache dieser vermutet wurde.
... This leads us to hypothesize that TsBAHD causes short styles in Turnera by acetylating and inactivating BRs, employing the same mechanism as that found in Primula but effected by a different class of BR-inactivating enzyme [11,13]. Support for this hypothesis has been found in the form of patterns of BR regulated gene expression in transcriptome analyses, where depletion of BR regulated genes was found in S-morph relative to L-morph pistils [20]. However, an empirical test of this hypothesized activity is lacking, but necessary. ...
Article
Full-text available
Heterostyly distinct hermaphroditic floral morphs enforce outbreeding. Morphs differ structurally, promote cross-pollination, and physiologically block self-fertilization. In Turnera the self-incompatibility (S)-locus controlling heterostyly possesses three genes specific to short-styled morph genomes. Only one gene, TsBAHD, is expressed in pistils and this has been hypothesized to possess brassinosteroid (BR)-inactivating activity. We tested this hypothesis using heterologous expression in Arabidopsis thaliana as a bioassay, thereby assessing growth phenotype, and the impacts on the expression of endogenous genes involved in BR homeostasis and seedling photomorphogenesis. Transgenic A. thaliana expressing TsBAHD displayed phenotypes typical of BR-deficient mutants, with phenotype severity dependent on TsBAHD expression level. BAS1, which encodes an enzyme involved in BR inactivation, was downregulated in TsBAHD-expressing lines. CPD and DWF, which encode enzymes involved in BR biosynthesis, were upregulated. Hypocotyl growth of TsBAHD dwarfs responded to application of brassinolide in light and dark in a manner typical of plants over-expressing genes encoding BR-inactivating activity. These results provide empirical support for the hypothesis that TsBAHD possesses BR-inactivating activity. Further this suggests that style length in Turnera is controlled by the same mechanism (BR inactivation) as that reported for Primula, but using a different class of enzyme. This reveals interesting convergent evolution in a biochemical mechanism to regulate floral form in heterostyly.
Preprint
Full-text available
Distyly is a floral dimorphism associated with a heteromorphic self-incompatibility mechanism that prevents inbreeding and promotes outcrossing in 28 angiosperm families. Distyly is controlled by a cluster of genes, i.e. the S-locus supergene, that evolved independently in all distylous species studied to date. Here we present a transcriptome atlas for Primula veris , the main model for studying distyly since Darwin, which helped to elucidate the mechanism through which the S-locus gene CYP T acquired its role in controlling style elongation. We also identified genes that are differentially expressed between the two floral morphs of P. veris and Fagopyrum esculentum , revealing the main molecular mechanisms and hormones underlying the expression of distyly in these two species. Further, we discovered shared homologous genes that likely control distyly in P. veris , F. esculentum and Turnera subulata , suggesting that the convergent evolution of distyly at the phenotypic level is coupled, to some extent, also at the molecular level. Additionally, we provided statistical support to the hypothesis that distyly supergenes of different species evolved by the recruitment of genes related to the phytochrome-interacting factor (PIF) signaling network. Overall, this is the first study to identify homologous genes that underlie the control of distyly in distantly-related taxa.
Article
Full-text available
KofamKOALA is a web server to assign KEGG Orthologs (KOs) to protein sequences by homology search against a database of profile hidden Markov models (KOfam) with pre-computed adaptive score thresholds. KofamKOALA is faster than existing KO assignment tools with its accuracy being comparable to the best performing tools. Function annotation by KofamKOALA helps linking genes to KEGG resources such as the KEGG pathway maps and facilitates molecular network reconstruction. Availability KofamKOALA, KofamScan, and KOfam are freely available from GenomeNet (https://www.genome.jp/tools/kofamkoala/) Supplementary information Supplementary data are available at Bioinformatics online.
Article
Full-text available
Distyly is an intriguing floral adaptation that increases pollen transfer precision and restricts inbreeding. It has been a model system in evolutionary biology since Darwin. While the S‐locus determines the long‐ and short‐styled morphs, the genes were unknown in Turnera. We have now identified these genes. We used deletion mapping to identify, and then sequence, BAC clones and genome scaffolds to construct S/s haplotypes. We investigated candidate gene expression, hemizygosity, and used mutants, to explore gene function. The s‐haplotype possessed 21 genes collinear with a region of chromosome 7 of grape. The S‐haplotype possessed three additional genes and two inversions. TsSPH1 was expressed in filaments and anthers, TsYUC6 in anthers, and TsBAHD in pistils. Long‐homostyle mutants did not possess TsBAHD and a short‐homostyle mutant did not express TsSPH1. Three hemizygous genes appear to determine S‐morph characteristics in T. subulata. Hemizygosity is common to all distylous species investigated, yet the genes differ. The pistil candidate gene, TsBAHD, differs from that of Primula, but both may inactivate brassinosteroids causing short styles. TsYUC6 is involved in auxin synthesis and likely determines pollen characteristics. TsSPH1 is likely involved in filament elongation. We propose an incompatibility mechanism involving TsYUC6 and TsBAHD. This article is protected by copyright. All rights reserved.
Article
Full-text available
Common buckwheat (Fagopyrum esculentum) is a heteromorphic self-incompatible (SI) species with two types of floral architecture: thrum (short style) and pin (long style). The floral morphology and intra-morph incompatibility are controlled by a single genetic locus, S. However, the molecular mechanisms underlying the heteromorphic self-incompatibility of common buckwheat remain unclear. To identify these mechanisms, we performed proteomic, quantitative reverse-transcription PCR, and linkage analyses. Comparison of protein profiles between the long and short styles revealed a protein unique to the short style. Amino-acid sequencing revealed that it was a truncated form of polygalacturonase (PG); we designated the gene encoding this protein FePG1. Phylogenetic analysis classified FePG1 into the same clade as PGs that function in pollen development and floral morphology. FePG1 expression was significantly higher in short styles than in long styles. It was expressed in flowers of a short-homostyle line but not in flowers of a long-homostyle line. Linkage analysis indicated that FePG1 was not linked to the S locus; it could be a factor downstream of this locus. Our finding of a gene putatively working under the regulation of the S locus provides useful information for elucidation of the mechanism of heteromorphic self-incompatibility.
Article
Full-text available
Proteins and their functional interactions form the backbone of the cellular machinery. Their connectivity network needs to be considered for the full understanding of biological phenomena, but the available information on protein-protein associations is incomplete and exhibits varying levels of annotation granularity and reliability. The STRING database aims to collect, score and integrate all publicly available sources of protein-protein interaction information, and to complement these with computational predictions. Its goal is to achieve a comprehensive and objective global network, including direct (physical) as well as indirect (functional) interactions. The latest version of STRING (11.0) more than doubles the number of organisms it covers, to 5090. The most important new feature is an option to upload entire, genome-wide datasets as input, allowing users to visualize subsets as interaction networks and to perform gene-set enrichment analysis on the entire input. For the enrichment analysis, STRING implements well-known classification systems such as Gene Ontology and KEGG, but also offers additional, new classification systems based on high-throughput text-mining as well as on a hierarchical clustering of the association network itself. The STRING resource is available online at https://string-db.org/.
Article
Full-text available
Gametophytic development in Arabidopsis depends on nutrients and cell wall materials from sporophytic cells. However, it is not clear whether hormones and signaling molecules from sporophytic tissues are also required for gametophytic development. Herein, we show that auxin produced by the flavin monooxygenases YUC2 and YUC6 in the sporophytic microsporocytes is essential for early stages of pollen development. The first asymmetric mitotic division (PMI) of haploid microspores is the earliest event in male gametophyte development. Microspore development in yuc2yuc6 double mutants arrests before PMI and consequently yuc2yuc6 fail to produce viable pollens. Our genetic analyses reveal that YUC2 and YUC6 act as sporophytic genes for pollen formation. We further show that ectopic production of auxin in tapetum, which provides nutrients for pollen development, fails to rescue the sterile phenotypes of yuc2yuc6. In contrast, production of auxin in either microsporocytes or microspores rescued the defects of pollen development in yuc2yuc6 double mutants. Our results demonstrate that local auxin biosynthesis in sporophytic microsporocytic cells and microspore controls male gametophyte development during the generation transition from sporophyte to male gametophyte.
Article
Full-text available
Heterostyly is a widespread floral adaptation to promote outbreeding, yet its genetic basis and evolutionary origin remain poorly understood. In Primula (primroses), heterostyly is controlled by the S-locus supergene that determines the reciprocal arrangement of reproductive organs and incompatibility between the two morphs. However, the identities of the component genes remain unknown. Here, we identify the Primula CYP734A50 gene, encoding a putative brassinosteroid-degrading enzyme, as the G locus that determines the style-length dimorphism. CYP734A50 is only present on the short-styled S-morph haplotype, it is specifically expressed in S-morph styles, and its loss or inactivation leads to long styles. The gene arose by a duplication specific to the Primulaceae lineage and shows an accelerated rate of molecular evolution. Thus, our results provide a mechanistic explanation for the Primula style-length dimorphism and begin to shed light on the evolution of the S-locus as a prime model for a complex plant supergene.
Article
Heterostylous genetic polymorphisms provide paradigmatic systems for investigating adaptation and natural selection. Populations are usually comprised of two (distyly) or three (tristyly) mating types, maintained by negative frequency‐dependent selection resulting from disassortative mating. Theory predicts this mating system should result in equal style‐morph ratios (isoplethy) at equilibrium. Here, I review recent advances on heterostyly, focusing on examples challenging stereotypical depictions of the polymorphism and unresolved questions. Comparative analyses indicate multiple origins of heterostyly, often within lineages. Ecological studies demonstrate that structural components of heterostyly are adaptations improving the proficiency of animal‐mediated cross‐pollination and reducing pollen wastage. Both neutral and selective processes cause deviations from isoplethy in heterostylous populations, and, under some ecological and demographic conditions, cause breakdown of the polymorphism, resulting in either the evolution of autogamy and mixed mating, or transitions to alternative outcrossing systems, including dioecy. Earlier ideas on the genetic architecture of the S‐locus supergene governing distyly have recently been overturned by discovery that the dominant S‐haplotype is a hemizygous region absent from the s‐haplotype. Ecological, phylogenetic and molecular genetic data have validated some features of theoretical models on the selection of the polymorphism. Although heterostyly is the best‐understood floral polymorphism in angiosperms, many unanswered questions remain.
Article
Distyly is a genetically controlled flower polymorphism that has intrigued both botanists and evolutionary biologists ever since Darwin’s time. Despite extensive reports on the pollination and evolution of distylous systems, the genetic basis and mechanism of molecular regulation remain unclear. In the present study, comparative transcriptome profiling was conducted in primrose (Primula oreodoxa), the prime research model for heterostyly. Thirty-six transcriptomes were sequenced for styles at different stages and corolla tube in the three morphs of P. oreodoxa. Large numbers of differentially expressed genes (DEGs) were detected in the transcriptomes of styles across different morphs. Several transcription factors (TFs) and phytohormone metabolism-related genes were highlighted in S-morphs. A growing number of genes showed differential expression patterns along with the development of styles, suggesting that the genetic control of distyly may be more complicated than ever expected. Analysis of co-expression networks and module-trait relationships identified modules significantly associated with style development. CYP734A50, a key S-locus gene whose products degrade brassinosteroids, was co-expressed with many genes in the module and showed significant negative association with style length. In addition, crucial TFs involved in phytohormone signaling pathways were found to be connected with CYP734A50 in the co-expression module. Our global transcriptomic analysis has identified DEGs that are potentially involved in regulation of style length in P. oreodoxa, and may shed light on the evolution and broad biological processes of heterostyly.
Article
With 145 species, Turnera is the largest genus of Turneraceae (Malpighiales). Despite several morphotaxonomic and cytogenetic studies, our knowledge about the phylogenetic relationships in Turnera remains mainly based on morphological data. Here, we reconstruct the most comprehensive phylogeny of Turnera with molecular data to understand the morphological evolution within this group and to assess its circumscription and infrageneric classification. We analyzed two nuclear and six plastid markers and 112 taxa, including species and infraspecific taxa, 97 from Turnera, covering the 11 series of the genus. Bayesian inference, maximum parsimony and maximum likelihood analyses show that Turnera, as traditionally circumscribed, is not monophyletic. The genus is divided into two well-supported independent clades; one of them is sister to the genus Piriqueta and is here segregated as the new genus Oxossia. According to our reconstructions, Turnera probably evolved from an ancestor without extrafloral nectaries and with solitary, homostylous flowers with yellow petals. The emergences of extrafloral nectaries and distyly, both common in extant taxa, played an important role in the diversification of the genus. An updated infrageneric classification reflecting the relationships within Turnera is now possible based on morphological synapomorphies and is here designed for further studies.
Article
Intermorph differences in the wall structure and constituents of stigmatic papillae are described for distylous Linum pubescens and L. grandiflorum. In the long-styled morph of both species the wall portion around the apex of the papilla has a thickened cellulose-pectin layer. In the short-styled morph of L. pubescens a cap zone is interposed between the cuticle and apical portion of the papilla wall. The subcuticular cap space contains pectins and lipid particles. Similar particles are also present in deposits on the cuticle surface. Papillae of the short-styled morph in L. grandiflorum lack a cap zone and have only epicuticular lipid deposits. Other distylous Linum species in which the two morphs differ in wall structure of the papillae are L. mucronatum, L. flavum, L. perenne, L. austriacum, and L. maritimum. Studies of stigma dimorphism can help to elucidate evolutionary relations between dimorphic and monomorphic Linum species.