- © 2019 American Society of Plant Biologists. All rights reserved.
Abstract
Understanding how plants adapt to ambient temperatures has become a major challenge prompted by global climate change. This has led to the identification of several genes regulating the thermal plasticity of plant growth and flowering time. However, the mechanisms accounting for the natural variation and evolution of such developmental plasticity remain mostly unknown. In this study, we determined that natural variation at ICARUS2 (ICA2), which interacts genetically with its homolog ICA1, alters growth and flowering time plasticity in relation to temperature in Arabidopsis (Arabidopsis thaliana). Transgenic analyses demonstrated multiple functional effects for ICA2 and supported the notion that structural polymorphisms in ICA2 likely underlie its natural variation. Two major ICA2 haplogroups carrying distinct functionally active alleles showed high frequency, strong geographic structure, and significant associations with climatic variables related to annual and daily fluctuations in temperature. Genome analyses across the plant phylogeny indicated that the prevalent plant ICA genes encoding two tRNAHis guanylyl transferase 1 units evolved ∼120 million years ago during the early divergence of mono- and dicotyledonous clades. In addition, ICA1/ICA2 duplication occurred specifically in the Camelineae tribe (Brassicaceae). Thus, ICA2 appears to be ubiquitous across plant evolution and likely contributes to climate adaptation through modifications of thermal developmental plasticity in Arabidopsis.
INTRODUCTION
Plants must adapt their development and physiology to environmental cues that change across geography and time. In the past few years, climatic factors such as temperature have become a major focus, and large efforts are currently devoted to understanding the mechanisms involved in plant adaptation to ambient temperature (Bita and Gerats, 2013). In particular, annual plants are characterized by their developmental plasticity in relation to temperature, which is presumed to reflect adaptations to seasonal fluctuations (Nicotra et al., 2010). High, but not stressful, temperatures influence overall plant growth during the vegetative phase, producing a set of morphological changes that are referred to as thermomorphogenesis. These changes resemble the plant responses to shade avoidance and include long hypocotyls and petioles, as well as upward leaf bending (hyponasty; McClung et al., 2016; Quint et al., 2016). Genetic and molecular studies, mostly in the model plant Arabidopsis (Arabidopsis thaliana), are elucidating the mechanisms underlying such temperature-mediated responses. Recently, PHYTOCHROME B has been proposed to function as a thermosensor through its temperature-dependent reversion from the active to the inactive form (Jung et al., 2016; Legris et al., 2016). In addition, the transcription factor PHYTOCHROME INTERACTING FACTOR4 (PIF4) has been shown to play a central role in controlling thermomorphogenesis (Quint et al., 2016). PIF4 is transcriptionally and posttranslationally regulated to integrate signals from light, the circadian clock, and several phytohormones such as gibberellins and brassinosteroids. PIF4 subsequently regulates auxin biosynthesis and signaling genes. Moreover, histone modifications and chromatin remodelling also contribute to thermomorphogenesis, as the dynamics of H2A.Z nucleosomes and histone deacetylation appear to be associated with the thermoregulation of gene expression (Quint et al., 2016; Tasset et al., 2018). On the other hand, ambient temperature also affects the major developmental transitions in the plant life cycle, such as flowering initiation, which is often, but not always, advanced by high temperature (Verhage et al., 2014; Méndez-Vigo et al., 2016). The thermoregulation of flowering time involves different molecular mechanisms, including the temperature-dependent splicing and nonsense-mediated mRNA decay of FLOWERING LOCUS M (FLM) and the temperature-induced degradation of SHORT VEGETATIVE PHASE (SVP; Lee et al., 2013; Sureshkumar et al., 2016); the regulation of the florigen-encoding genes FLOWERING LOCUS T (FT) and TWIN SISTER OF FLOWERING LOCUS T (TSF) by PIF4 and SVP (Kumar et al., 2012; Fernández et al., 2016); and a potential role for the regulation of microRNAs miR156 and miR172 by the SQUAMOSA PROMOTER BINDING-LIKE (SPL) gene family (Verhage et al., 2014).
In addition to environmental acclimation to high ambient temperature via developmental plasticity, plant evolutionary adaptation to climate involves genetic modifications of such phenotypic responses among natural populations (Franks et al., 2014). Arabidopsis studies have already identified several genes contributing to the natural variation in thermomorphogenic growth, such as the circadian clock gene EARLY FLOWERING3 (ELF3; Box et al., 2015; Raschke et al., 2015) and likely PIF4, a gene downregulated by ELF3 (Brock et al., 2010). Furthermore, natural variants of FLM, FLOWERING LOCUS C (FLC), and FRIGIDA affect the plasticity of flowering time in relation to high ambient temperature (Balasubramanian et al., 2006; Lutz et al., 2015, 2017; Méndez-Vigo et al., 2016; Sanchez-Bermejo and Balasubramanian, 2016). Structural and regulatory mutations in these genes have been shown to alter temperature-mediated plasticity. In particular, a missense mutation in ELF3 and multiple deletions and insertions in the promoter or intron regions of the other genes have been reported as functional polymorphisms (Lutz et al., 2015; Raschke et al., 2015; Méndez-Vigo et al., 2016). In addition, natural variation at the ICARUS1 (ICA1) gene has been shown to produce a very severe growth defect characterized by short hypocotyls, small leaf size, late flowering, and low fertility, specifically at elevated temperature (>27°C), but not at 21°C (Zhu et al., 2015). Two natural loss-of-function mutations have been found in Arabidopsis accessions Don-0 and Sij-4 that behave similarly to null mutant alleles in the Columbia (Col) background. Plants carrying such natural or artificial ica1 loss-of-function mutations also display larger cells, blockage of the G2/M transition of the cell cycle, and increased endoreduplication at high temperature, as well as enhanced sensitivity to DNA damage (Zhu et al., 2015). Thus, ICA1 is an essential gene required for growth at high temperature that affects the plastic responses mediated by ambient temperature and other fundamental processes.
ICA1 is the homolog of Thg1 from humans and yeasts. Thg1 encodes a tRNAHis guanylyl transferase, an enzyme that regulates tRNAHis maturation by catalyzing the addition of a G nucleotide specifically to the 5′ end of tRNAHis (Gu et al., 2003; Heinemann et al., 2012). This gene is universal across all life kingdoms, from archaea to animals and plants, and the enzymatic activity of Thg1 has been extensively studied because, contrary to other nucleic acid polymerases, it synthesizes 3′-to-5′ RNA (Heinemann et al., 2012; Lee et al., 2017). Most animal and fungal genomes contain only one Thg1 copy encoding a simple Thg1 protein characterized by the catalytic and the C-terminal domains. By contrast, Arabidopsis contains two ICA/Thg1 genes, each encoding two complete Thg1 units connected by a 40-amino acid peptide, which indicates that several duplication events have occurred during Arabidopsis ICA/Thg1 evolution (Placido et al., 2010; Zhu et al., 2015). However, the overall evolution, including copy number and structure, for this conserved protein family remains unknown in plants.
Understanding the functional relationships between duplicated genes is fundamental for elucidating many biological processes because gene duplication has been a chief mechanism to generate genetic diversity during plant evolution (Panchy et al., 2016; Soltis and Soltis, 2016). The particular functional relationships between duplicated genes have led to different types of genetic interactions or epistasis, which underlie the gene networks regulating phenotypic diversity (Phillips, 2008). Multiple approaches are currently being used to address the mechanisms that preserve duplicated genes by neofunctionalization or subfunctionalization and their genetic interactions (Lynch, 2007). These include intra- and interspecific genome-wide comparisons estimating diversity and differentiation rates (Lynch and Katju, 2004; Panchy et al., 2016), as well as functional analyses of duplicated genes using artificial genetic reconstructions (Hanikenne et al., 2008; Gallego-Bartolomé et al., 2010; Vlad et al., 2014). In addition, several studies have dissected the functions of natural alleles segregating in duplicated genes at the intraspecific level, which are presently shaping the evolutionary fate of duplication events (Kliebenstein et al., 2001; Kroymann et al., 2003; Kliebenstein, 2008; Caicedo et al., 2009; Vlad et al., 2010; Prasad et al., 2012; Trontin et al., 2014).
In this study, we addressed the genetic, functional, and evolutionary relationships between the pair of duplicated genes, ICA2 and ICA1, in Arabidopsis. The strong, high temperature–specific growth defects of Don-0 accession revealed substantial natural variation at ICA2, which shows synergistic epistasis with ICA1. We characterized the functional bases of ICA2 variation for several adaptive traits, such as growth and flowering time, in relation to ambient temperature. In addition, we inferred the role of ICA2 in climate adaptation via association analyses with climatic factors. Finally, we determined the evolutionary patterns of ICA/Thg1 copy number and structure across the plant phylogeny. These results shed light on the conservation of the ICA/Thg1 genes during plant evolution, the maintenance of Arabidopsis ICA duplication in the Camelineae tribe, and the contribution of ICA2 to intraspecific variation for thermal developmental plasticity in Arabidopsis.
RESULTS
Fine Mapping Reveals That the Don-0 Growth Defect Is Caused by Two Closely Linked, Epistatic Loci
Previous analysis of the growth defect in the Sij-4 × Col crossing population at high temperature showed that this defect is determined by a recessive allele mapping to a 37-kb interval in ICA1. By contrast, similar analysis of Don-0 × Landsberg erecta (Ler) or Sij-4 × Ler crosses could only locate ICA1 to a large genomic region of ∼400 kb because recombinant plants between markers inside this region did not yield clear results (Zhu et al., 2015). This suggested the presence of a potential genetic modifier within this ∼400-kb mapping interval. To determine whether another locus near ICA1 could contribute to the growth phenotype of the Don-0 × Ler cross, we analyzed a family of 2200 plants segregating for this genomic segment (Figure 1). A total of 33 recombinant plants in the 400-kb region were isolated, which were classified as Ler, Don-0, or segregating families according to the severe growth defect observed in the vegetative phase of the F3 offspring grown at 28°C. The application of a genetic model with a single gene causing this phenotype led to inconsistent fine mapping from multiple recombinant classes (e.g., recombinant classes i and m in Figure 1B). In addition, we did not find recombinant plants showing intermediate phenotypes, but the strong growth defect was only present in recombinants carrying homozygous Don-0 alleles in almost the complete 400-kb region. These results suggest that the growth phenotype was caused by Don-0 recessive alleles in at least two linked loci with synergistic epistasis. In particular, two classes of recombinant genotypes showing disparate phenotypes (Figure 1B, recombinant classes a and g) allowed us to map two loci to adjacent genomic regions of 95 and 139 kb, as defined by the physical distance between the markers flanking the recombination events of these plants. Genotypic and phenotypic analysis of the selfed progeny from six recombinants between both loci (classes b, c, d and h, i, j in Figure 1B) showed that the Ler and Don-0 alleles at these loci display significant epistasis at high temperature (P < 0.001). This interaction explained 59% of the phenotypic variance for rosette size measured at 28°C (Figure 1C). Therefore, natural variation in two closely linked and interacting genes, named ICA1 and ICA2, accounts for the vegetative growth defect of Don-0. Furthermore, only 25% of the offspring from recombinant plants homozygous for ICA1-Don-0 alleles and heterozygous ICA2-Don-0/ICA2-Ler displayed the high temperature growth defect (chi-square P > 0.05 for 3:1 segregation test). Genotyping of these offspring showed that heterozygous ICA2 individuals had normal growth at 28°C, which indicates that a single copy of ICA2-Ler is sufficient to restore normal growth at high temperature, that is, the ICA2-Ler allele is dominant versus ICA2-Don-0.
High Temperature Growth Defect and Fine Mapping of ICARUS2.
(A) Vegetative growth of 20-d-old Ler, Don-0, Col, ica1-2, and ica2-1 plants at 21 and 28°C.
(B) Genetic map of chromosome 2 and graphical genotypes of the different classes of Don-0/Ler recombinant plants isolated in the ∼400-kb genomic interval. Recombinants are classified and arranged according to Don-0/Ler alleles in ICA1 and ICA2 genomic regions. The recombinant class (Rec. class) and the number of independent recombinant plants with similar genotype analyzed (# of rec.) is shown in the left side of the panel. The 28°C growth phenotype of the offspring derived from each recombinant was recorded as Ler, Don-0, or segregating families and is presented in the right side of the panel. The two most informative classes of recombinants are marked with asterisks (*). They corresponded to genotypes shared by multiple independent recombinants but displaying disparate growth phenotypes, as indicated on the right. The size and location of ICA1 and ICA2 mapping intervals, delimited by the markers flanking the recombination events of these two classes of genotypes, are depicted with dark blue rectangles. The positions of the pair of duplicated genes At2g31580 and At2g32320 are indicated with dark blue dashed lines. Chro. 2, chromosome 2.
(C) Genetic interaction between ICA1 and ICA2 in the Ler/Don-0 mapping population. Panel shows the 28°C rosette size (mean ± se of 14 to 25 plants) of homozygous plants for Ler or Don-0 alleles at ICA1 and ICA2 regions.
(D) The 28°C growth phenotypes of 14-d-old F1 plants derived from crosses between Col line 35S:amiR-ICA1 and accessions carrying a Don-0 (top panel) or Ler-like (bottom panel) allele at At2g32320.
ICA2 Encodes a Thg1 Homolog That Interacts with ICA1
At2g31580, which was previously demonstrated to carry natural loss-of-function mutations underlying ICA1 in Don-0 and Sij-4 (Zhu et al., 2015), is located in the ICA1 mapping interval. Analysis of the ICA2 genomic region identified the ICA1 homolog At2g32320 located 291 kb from ICA1 (Figure 1B). Since both genes encode Thg1-like proteins that share 86% identity, we hypothesized that At2g32320 is ICA2. To evaluate this, we first analyzed the sequence of At2g32320 in the four parental accessions. Ler and Don-0 differed in 60 single-nucleotide polymorphisms and 16 small insertion/deletions across the gene, including eight missense mutations that mainly affect the catalytic and C-terminal domains of the first Thg1 unit (Figure 2A). By contrast, Col and Sij-4 showed similar protein structure to that of the Don-0 accession, explaining our ability to map ICA1 into a smaller interval in the Sij-4 × Col cross (Zhu et al., 2015). In addition, these findings suggest that Don-0/Sij-4/Col At2g32320 protein is required for ICA1 loss-of-function mutations to display the growth defect. To determine whether the genetic interaction between ICA1/At2g31580 and At2g32320 accounts for the high temperature growth defect, we generated artificial ICA1 loss-of-function mutants in accessions encoding At2g32320 proteins similar to those of Don-0 or Ler. To this end, the previously developed line 35S:amiR-ICA1, carrying a microRNA that targets the last ICA1 coding exon (Zhu et al., 2015), was crossed with nine accessions and F1 plants were analyzed (Figure 1D). All F1 lines had normal phenotypes at 21°C. However, at 28°C, all F1 individuals derived from four accessions with Don-0-like At2g32320 showed the severe growth phenotype, whereas the F1 plants from five other accessions encoding Ler-like At2g32320 proteins grew normally. Therefore, the temperature-dependent growth defect caused by ICA1 loss-of-function depends on the natural allele at At2g32320.
Genetic Complementation of Developmental and Cellular Traits by Natural Alleles of ICA2.
(A) Genomic and protein structure of ICA2 from parental accessions (Ler, Don-0, Sij-4, and Col) showing protein domains and missense mutations. SNP, single-nucleotide polymorphism; UTRs, untranslated regions.
(B) Phenotypic analyses of ica1-2 homozygous transgenic lines carrying ICA2-Ler or ICA2-Don-0 grown at 28°C, for rosette size (left panel), flowering time (middle panel), and fruit length (right panel). Transgenic lines in the various panels are arranged in the same order. Upper and lower 95% confidence intervals for Col and ica1-2 phenotypes are shown as blue-shaded areas.
(C) Flow cytometry analyses of nuclear DNA content in parental, transgenic, and mutant lines grown at 28°C. Left panel shows representative ploidy profiles of Ler and Don-0 leaves. Right panel displays the percentage of leaf cells with 2C to 64C ploidy levels in six transgenic lines carrying ICA2-Ler or ICA2-Don-0 transgenes compared with Col, ica1-2, and ica2-1 genotypes.
(D) Sensitivity to DNA damage measured by the first leaf assay in 10-d-old seedlings treated with BLEO. The percentage of BLEO-treated plants with true leaves is shown in relation to mock samples for parental, mutant, and ica1-2 transgenic lines. In (C) and (D), each bar represents the mean ± se of three biological replicates, and statistical comparisons of each line and Col are indicated on the bars.
(E) ICA1 and ICA2 expression in parental accessions at different temperatures. Each panel shows the mean ± se of three biological replicates.
(F) Correlations between ICA2 expression and rosette size (left panel), flowering time (middle panel), and fruit length (right panel) in ICA2 transgenic lines grown at 28°C. The expression of each line is relativized to that of untransformed Col plants. Statistical significances of Pearson correlation coefficients and significant regression lines are shown in each panel. ns, nonsignificant; *P < 0.001.
In contrast to ica1 loss-of-function mutants, a T-DNA insertional mutant of At2g32320 in the Col background showed no obvious growth defect during the vegetative or reproductive phase at different temperatures (Figures 1A and 2B; Supplemental Figure 1). Nevertheless, we could not identify a double mutant for ICA1 and At2g32320 from a large F2 population of 2346 plants (Supplemental Figure 2). This result strongly suggests that the At2g32320 T-DNA allele is not functional and that it interacts with ica1-2 to cause lethality. To prove that At2g32320 is ICA2, we obtained 16 independent homozygous transgenic lines carrying a 5.6-kb genomic construct with the coding and promoter regions of At2g32320 from Ler or Don-0 in the genetic background of a Col ica1-2 loss-of-function mutant that carries a Don-0-like allele of At2g32320 (Figure 2; Supplemental Figure 3). All of these lines showed normal growth at 21°C, similar to Col plants. However, they strongly differed at 28°C when we analyzed them in detail by measuring rosette diameter, flowering time, leaf number, and fruit length (Supplemental Table 1). Most of the lines with the At2g32320 transgene from Ler displayed growth and flowering phenotypes similar to those of the Col wild-type accession, indicating that this transgene fully complemented the phenotypic defects observed in ica1-2 (Figure 2B). On the contrary, all but one of the eight lines with the At2g32320 transgene from Don-0 showed phenotypes similar to ica1-2 (small rosettes, late flowering, and small and sterile fruits) or intermediate between ica1-2 and Col, which correspond to no and partial complementation, respectively. Overall, an At2g32320 transgene from Ler, but not from Don-0, complemented the growth defects produced by ICA1 loss-of-function alleles, which is in agreement with the observed epistasis between Ler and Don-0 alleles at ICA1/ICA2 (Figure 1C). Hence, we concluded that ICA2 is At2g32320 and that the temperature-dependent developmental phenotypes of the Don-0 accession are caused by natural alleles in two homologous genes encoding Thg1 proteins.
Natural Variation at ICA2 Affects the Cell Cycle and Sensitivity to DNA Damage
To determine whether ICA2 is involved in the same cellular processes that were previously described for ICA1 (Zhu et al., 2015), we also analyzed ICA2 transgenic and mutant lines for cellular DNA content and sensitivity to DNA damage. Flow cytometry measurements showed that Don-0 and ica1-2 plants contained larger proportions of cells with high ploidy levels than the wild-type Ler and Col at 28°C, but not 21°C (Figure 2C; Supplemental Figure 4). By contrast, the DNA content profiles of ica2-1 plants were similar to those of Col at both temperatures. In addition, at 21°C, ica1-2 transgenic lines carrying ICA2-Ler or ICA2-Don-0 alleles did not differ in terms of DNA content patterns (Supplemental Figure 4). However, both types of lines differed significantly at 28°C, since the ICA2-Ler transgenic lines were similar to Col, whereas lines with ICA2-Don-0 transgenes displayed larger proportions of cells with 16C and 32C ploidy levels (chi-square P < 0.001), as observed for ica1-2 (Figure 2C).
We measured sensitivity to DNA damage in the plants using the first leaf assay, which is based on growth arrest of the first pair of leaves in response to treatment with the DNA damaging agent Bleocin (BLEO) at 21°C (Rosa et al., 2013; Zhu et al., 2015). This assay showed that Don-0 and ica1-2 were significantly more sensitive to BLEO damage than Col and Ler, but no significant difference was found between ica2-1 and Col plants (Figure 2D). On the contrary, both types of ica1-2 transgenic lines differed significantly in terms of sensitivity to DNA damage (P < 0.001), since lines carrying ICA2-Ler transgenes showed similar proportions of plants with true leaves after BLEO treatment to Col (89.2% ± 1.1%), whereas lines with ICA2-Don-0 had lower values (53.8% ± 1.9%), like ica1-2 (Figure 2D). Thus, the effects of ICA1 on both cellular traits depended on the ICA2 allele, indicating that ICA2 also interacts with ICA1 for the regulation of cell division and for the capacity to repair DNA damage.
The Phenotypic Effects of ICA2 Are Not Caused by cis-Regulatory Polymorphisms
Since natural alleles of ICA2 differ in terms of mutations in the coding and noncoding gene regions, we aimed to find out whether the functional differences between Don-0 and Ler could be explained by structural or cis-regulatory polymorphisms. We analyzed ICA2 expression in the Don-0 and Ler parental accessions, but no significant difference was detected (Figure 2E). We then quantified ICA2 expression at 28°C in all lines carrying ICA2 transgenes from Don-0 or Ler in the ica1-2 loss-of-function background (Figure 2F). On average, transgenic lines with ICA2-Ler and ICA2-Don-0, which differ strongly in their developmental and cellular phenotypes, did not differ significantly in terms of ICA2 expression (P = 0.33). This result indicates that natural alleles of ICA2 do not bear cis-regulatory mutations affecting the overall gene expression level, as both parental transgenes included their own native promoters and untranslated regions. Therefore, the functional and phenotypic variation between natural alleles of ICA2 in Don-0/Col and Ler is likely caused by structural polymorphisms. Nevertheless, independent lines carrying the same transgene showed up to fivefold differences in expression (P < 10−5). The ICA2-Don-0 line with the highest expression level was the only line with this transgene displaying almost full complementation (Figures 2B and 2F). On the contrary, the transgenic line with nearly the lowest ICA2-Ler expression level showed the weakest complementation. Accordingly, significant correlations were detected between ICA2 expression and the three quantitative traits affected by ICA2 (Figure 2F). These results demonstrate that both natural alleles of ICA2 encode functional proteins, but the Ler allele is more active than the Don-0 allele. In addition, between-line variation in ICA2 expression (due to the different transgene insertion sites) also affects ICA2 function and growth at high temperature.
Natural Variation in ICA2 Affects Growth at Different Temperatures
To characterize the effects of ICA2 allelic variation in the presence of active ICA1 alleles, we also analyzed 13 independent homozygous transgenic lines carrying Don-0 or Ler ICA2 transgenes in the ica2-1 genetic background. Quantitative analysis of leaf size traits at different temperatures showed that at 28°C, but not 21°C, ica2-1 had smaller petioles and blade length or area than Col (P < 0.01), indicating that ICA2 weakly but significantly affects the thermomorphogenic response (Figure 3). In addition, homozygous lines for ICA2-Ler and ICA2-Don-0 transgenes differed in terms of leaf size traits at both 21 and 28°C (0.001 < P < 0.02; Supplemental Table 1). On average, the ICA2-Ler plants had larger leaf petioles and blades than ICA2-Don-0 plants (Figures 3A, 3B, and 3C), again pointing out that the ICA2-Ler allele is more active than ICA2-Don-0. Parallel analyses of flowering time at different temperatures showed that neither the ica2-1 mutation nor ICA2 transgenes had a significant effect on this trait (P > 0.05). However, all transgenic lines displayed accelerated flowering mediated by high temperature similar to Col (P < 0.001; Figure 3D). Therefore, natural variation in ICA2 did not affect flowering time in an active ICA1 genetic background.
Phenotypic Effects of ICA2 Natural Alleles at Different Ambient Temperatures.
(A) to (D) Reaction norms of ica2-1 homozygous transgenic lines carrying ICA2 transgenes from Ler (left panel) or Don-0 (middle panel), as well as Col and ica2-1 control lines (right panel), for petiole length (A), blade length (B), leaf area (C), or flowering time (D) estimated at 21 and 28°C.
(E) and (F) Expression of ICA2 (E) and ICA1 (F) in the same transgenic and control lines. In (A) to (D), panels show the mean ± se of each line derived from 20 to 30 plants grown at 21 or 28°C. In (E) and (F), panels show the mean ± se of each line grown at 21 or 28°C as estimated from three technical replicates. Values of all lines are relative to Col expression at 21°C. The effects of ICA2 transgene and temperature were statistically tested by mixed-model analysis of variance. In each panel, the same or different letters close to the four groups of lines (two transgenes and two temperatures) indicate nonsignificant and significant differences, respectively, as tested by Tukey’s test (P < 0.05).
To quantify the effects of temperature and ICA2 alleles on gene expression, we also analyzed ICA2 and ICA1 mRNA levels in all of these lines. Most of the transgenic lines had higher ICA2 expression levels than the Col parental accession, although there was substantial variation among lines (Figure 3E). Overall, no significant difference in expression was detected between ICA2 transgenes (P = 0.8). By contrast, a small effect of temperature on ICA2 expression was detected, which depended on the transgene, because two lines carrying ICA2-Ler showed lower expression at 28 versus 21°C (Figure 3E). These results suggest that the position of the ICA2 transgene insertion affects the temperature response of the transgene. In addition, ICA1 expression showed much lower variation among transgenic lines, and no significant difference was found between transgenes at both temperatures (Figure 3F).
ICA2 Diversity Shows Broad Geographic Distribution and Climatic Association
To determine the intraspecific relevance of the structural polymorphisms of ICA2 found in Don-0 and Ler, we analyzed the global diversity of ICA2 in Arabidopsis (1001 Genomes Consortium, 2016). Clustering analysis of 791 sequences identified two major haplogroups present in 73% and 15% of the worldwide accessions corresponding to the Don-0/Col and Ler alleles, respectively (Figure 4A). Both haplogroups differed in the eight Don-0/Ler polymorphic amino acids and in numerous polymorphisms in introns and the 3′ region, which displayed moderate-to-strong linkage disequilibrium. Network analyses of protein sequences showed that the ICA2-Don-0/Col haplogroup is closer to the A. lyrata haplotype and, consequently, to the ancestral allele, whereas ICA2-Ler is the most distant allele (Figure 4B).
Genetic Diversity and Geographic and Climatic Distribution of ICA2 in Arabidopsis.
(A) Neighbor-joining tree showing the genetic relationships among 791 ICA2 worldwide genomic sequences. Branches corresponding to partitions reproduced in <50% bootstrap replicates are collapsed.
(B) Haplotype network of ICA2 proteins from worldwide accessions of Arabidopsis and A. lyrata. Each node corresponds to an amino acid substitution, with black dots corresponding to predicted nodes that are not present in the analyzed sequences. Areas of nodes are proportional to frequency.
(C) Geographical and climatic distribution of major ICA2 haplogroups in Eurasia. Temperature seasonality is measured as the sd of monthly temperature averages (units of the scale are degrees Celsius × 100). In (A) to (C), accessions and amino acids belonging to Don-0/Col or Ler haplogroups are shown in magenta and sea green, respectively.
Geographic analyses of ICA2 diversity revealed a strong spatial structure, as accessions with the Ler haplogroup are mainly distributed in Northern Europe and Asia (Figure 4C). Approximately 90% of the ICA2-Ler accessions belonged to genetic groups 3, 5, and 6 established from genome sequences (1001 Genomes Consortium, 2016), which are only found in Scandinavia and Asia. To further examine whether the distribution of ICA2 haplogroups is related to climate, we also tested associations using climatic data from the original population locations (Figure 4C). Analyses of all 687 accessions with Ler or Don-0/Col haplogroups detected significant associations of ICA2 with temperature seasonality and isothermality on a global scale (P < 0.008). Moreover, since the strong genetic structure of the worldwide collection might mask or confound climate associations, we performed similar tests including only the 159 accessions of genetic groups 3, 5, and 6, finding similar results at −log(P) = 5.3. Overall, it appears that the ICA2-Ler haplogroup is mainly distributed in locations with moderately low temperature seasonality and isothermality, whereas the ICA2-Don-0/Col haplogroup showed a wider climatic distribution. Thus, demographic and climatic factors linked to annual and daily fluctuations in temperature likely contributed to the current geographic distribution of ICA2 haplogroups.
Interspecific ICA/Thg1 Gene Diversity across the Plant Phylogeny Discloses Ancient Gene Duplications and Rearrangements
To explore the evolution of the Arabidopsis complex locus including both ICA1 and ICA2, we analyzed the number, structure, and physical positions of ICA genes in the well-annotated genomes available for 58 species across the phylogenetic tree of green plants (Figure 5). Overall, most diploid species carried only one or two ICA/Thg1 genes, encoding proteins with an average identity of 60.1% ± 13.4%, which indicates a strong conservation of the low number and structure of ICA genes during plant evolution. As described for most fungi and animal species, the oldest clades of Chlorophyta and Embriophyta carry only one ICA gene encoding a single Thg1 unit (simple ICA/Thg1 gene). By contrast, the moss Spagnum fallax, the basal group of Tracheophyta (Selaginella moellendorffii), and the early diverging monocots (Zostera marina) and eudicots (Aquilegia coerulea) carried two or three gene copies, indicating that a first duplication of the simple ICA/Thg1 gene occurred in an ancestral tracheophyte ∼450 million years ago (Mya; Vargas and Zardoya, 2014). Furthermore, ICA genes from all analyzed grass monocots and core eudicots encode proteins with two Thg1 units, which we refer to as double ICA/Thg1 genes (Figure 5). Therefore, the transition from simple to double ICA/Thg1 genes could be dated to the diversification of mono- and dicotyledonous plants ∼120 Mya (Vargas and Zardoya, 2014). We also performed maximum-likelihood (ML) phylogenetic analysis of ICA/Thg1 proteins, which, as expected, reflected the overall phylogeny of green plants from a common ancestor with fungi and animals (Figure 6A). However, this analysis did not detect groups of orthologous genes supporting a common ancestral duplication in different early diverging monocots and eudicots with at least two copies of the simple ICA/Thg1 gene (Z. marina, Musa acuminata, and A. coerulea); instead, all proteins of each species clustered together. This finding suggests that multiple independent duplications of the ICA/Thg1 simple gene might have occurred in those basal groups and/or that both copies of ICA/Thg1 genes coevolved in these plants to maintain high similarity.
Evolution of the Number and Structure of ICA Proteins in Viridiplantae.
Sizes and domain structures of proteins encoded by ICA genes present in the genomes of 58 plant species are depicted in the right part of figure. For each species, the genome size, chromosome number, ploidy level of the dominant phase of the life cycle, number of ICA genes per haploid genome, and number of Thg1 units encoded per gene are also shown. Ploidy level indicated as 2* corresponds to species evolved by recent polyploidization. Species are arranged according to the plant phylogeny as established in Phytozome (http://phytozome.jgi.doe.gov). Major ICA duplications or gene size doubling are indicated in red or purple, respectively, in the left side of the panel. Main tribes of Brassicaceae carrying ICA duplications are indicated in green. ICA genes that are closely linked in the same chromosomes are depicted as connected by a dashed line, indicating the size of the DNA genomic interval between them. chrom., chromosomes.
Evolutionary and Microsynteny Relationships of ICA1 and ICA2 in Brassicaceae.
(A) ML phylogenetic tree obtained for 94 ICA proteins from 58 species across the plant phylogeny and seven model organisms from other life kingdoms, as described in Figure 5. The bootstrap consensus tree inferred from 1000 replicates is taken to represent the evolutionary history of the taxa analyzed (Tamura et al., 2011). Double and simple ICA proteins detected by the genome and protein structure analysis described in Figure 5 are shown with bold or normal text, respectively. For the five Brassicaceae species belonging to the Camelineae tribe and carrying multiple ICA genes (shown in light blue), ICA proteins are named by adding ICA1 or ICA2 to the species name based on their closer relationship to ICA1 or ICA2 from Arabidopsis. For the remaining plant genomes with more than one ICA gene, ICA proteins are named by adding a number from one to three to the species name. Red and green depict mono- and dicotyledonous clades, respectively, whereas dark blue spans the Brassicaceae plant family.
(B) and (C) Microsynteny within ICA1/ICA2 genomic region between A. lyrata (B) or C. rubella (C), and Arabidopsis. Microsyntenic blocks are shown as red rectangles above the coding sequences connected by shaded lines. The positions of ICA1 and ICA2 are framed in blue. Microsynteny analyses were performed using the Web-based SynMap browser as implemented in the Comparative Genomics CoGe platform (https://genomevolution.org/coge/; Lyons and Freeling, 2008; Haug-Baltzell et al., 2017).
Plants of most core eudicot families carried only one double ICA/Thg1 gene, with the main exception being some recent polyploid species, such as soybean (Glycine max) and apple tree (Malus domestica), which had larger numbers of these genes. However, 9 of 11 analyzed Brassicaceae plants had two or more genes, pointing to specific ICA/Thg1 duplications in this family. In particular, Arabidopsis, Arabidopsis lyrata, Capsella rubella, and the hexaploid Camelina sativa, all belonging to the Camelineae tribe, contained one to three pairs of ICA/Thg1 genes closely linked at physical distances smaller than 1 Mb (Figures 5 and 6). The similar arrangement of most genes within ICA intervals of these species revealed strong microsynteny and common ancestry for the ICA genomic region of Camelineae (Figures 6B and 6C; Supplemental Figure 5). In addition, diploid and polyploid species of the Brassiceae tribe also contained multiple ICA/Thg1 copies, but these were located on different chromosomes. The ML phylogenetic analysis showed that the first cluster separating in the Brassicaceae family includes ICA2 from Arabidopsis and half of the ICA proteins from other Camelineae species (Figure 6A). Moreover, all ICA/Thg1 sequences from the Brassiceae tribe clustered together with Arabidopsis and other Camelineae ICA1 proteins. Therefore, two independent ICA/Thg1 duplications have occurred within the Brassicaceae family, but only the older duplication of Camelineae maintained the close linkage of the two ICA/Thg1 genes.
DISCUSSION
Genetic Interactions among Natural Alleles Uncover a Conserved Function for the Duplicated ICARUS Genes in Arabidopsis
The genetic and functional analyses performed in this study show that the pair of duplicated genes ICA1 and ICA2 of Arabidopsis behaves as a rather complex locus, since both genes are closely linked in the genome and their natural alleles interact to produce a severe growth defect that is conditional upon high ambient temperature. This locus illustrates the extreme complexity that genetic and functional relationships between duplicated genes can achieve in plants. Most previous analyses addressing the function of natural alleles in duplicated genes have focused on recent duplications segregating within a single species and mainly affecting tandemly repeated genes (Kliebenstein et al., 2001; Kroymann et al., 2003; Caicedo et al., 2009; Prasad et al., 2012; Trontin et al., 2014; see Vlad et al., 2010 as nontandem repeat duplication). However, Arabidopsis ICA duplication is conserved across the Camelineae tribe of Brassicaceae, providing an example of the natural variation in older duplications, which have been mainly approached in terms of inter- and intraspecific hybrid incompatibilities (Bomblies et al., 2007; Bikard et al., 2009; Alcázar et al., 2010, 2012; Bomblies, 2010).
Phenotypic analyses of plants carrying different natural alleles at ICA1 and ICA2 show that ICA2-Don-0/Col is the recessive allele and that only homozygous plants for recessive alleles at both ICA genes display severe developmental defects along the Arabidopsis life cycle, specifically at high temperature. This synergistic, double dominant epistasis between ICA1 and ICA2 indicates that both genes are involved in the same biological process, because no additional or intermediate phenotypes were observed in heterozygous plants for each ICA gene or in homozygous plants for the recessive allele only in one of the two genes. The genetic interaction affected not only plant growth and flowering traits but also cellular phenotypes, such as the high nuclear cell content caused by endoreduplication and the sensitivity to DNA damage. These results demonstrate that the temperature-dependent phenotypes previously described for ICA1 (Zhu et al., 2015) are determined by the genetic interaction between both ICA1 and ICA2 genes. Therefore, ICA2 is also involved in regulating the cell cycle and cell division to modulate cell size and subsequently plant growth, mainly at high temperature. Hence, in contrast to the many duplicated genes that have diverged from each other by subfunctionalization or neofunctionalization (Lynch, 2007), ICA1 and ICA2 have a general conserved function in Arabidopsis.
The characterization of the epistasis between ICA1 and ICA2 in the regulation of plant growth at high temperature provides additional functional information about these genes. First, both Don-0 and Ler natural alleles of ICA2 are active, but ICA2-Ler is dominant and a gain-of-function allele, as shown by the analysis of transgenic lines in the ica1 and ica2 backgrounds. Second, the differential behavior of artificial loss-of-function mutations of ICA1 and ICA2 genes points to the major relevance of ICA1 for growth at high temperature, although only in the background of the most frequent and weaker ICA2-Col allele. Third, concordant with the lethality of null mutants described in yeast (Gu et al., 2003), the combination of null alleles in both ICA genes is likely to be lethal in Arabidopsis, as we could not obtain an ica1-2 ica2-1 double mutant.
In agreement with the functional conservation observed in Arabidopsis, ICA genes show very limited divergence over hundreds of million years, as the number, structures, and sequences of ICA proteins appeared highly conserved across the plant phylogeny. Two major ancient structural events have occurred during plant evolution. The first duplication of the simple ICA gene encoding a single Thg1 unit took place at the origin of tracheophyte plants ∼450 Mya (Vargas and Zardoya, 2014). Second, with the early divergence of monocots and eudicots ∼120 Mya, either the two simple ICA/Thg1 genes were rearranged, or one of them was duplicated in tandem and the other lost, to render a single gene encoding two Thg1 units. The evolution of such a double ICA/Thg1 gene ensured that both Thg1 units display similar expression and protein stoichiometry. The presence of one double ICA gene has been maintained throughout most angiosperms, with the main exception being Brassicaceae, where we detected two independent duplications that occurred in the ancestors of the Camelineae and Brassiceae tribes. Taking into account our current view of the evolution of Brassicaceae, these ICA duplications can be dated to the Middle Miocene, ∼12 to 18 Mya (Huang et al., 2016), although ML phylogenetic analysis supports the notion that the Camelineae duplication is older than that of the Brassiceae. Thus, ICA genes appear to comprise a very small gene family well suited for phylogenomic studies across the plant kingdom because they are universal, essential, and rather conserved.
Natural Structural Variation in ICARUS2 Is Likely Involved in Growth Adaptation to Temperature Fluctuations
In contrast to ICA1, for which various independent and very low frequency loss-of-function mutations have been found in nature (Zhu et al., 2015), the natural gain-of-function allele of ICA2 described here shows a high worldwide frequency. Our molecular and environmental characterizations provide several clues about the functional bases of natural variation in ICA2 and its role in the adaptation of Arabidopsis to climate. On one hand, the phenotypic behavior and gene expression of ICA2 transgenic lines at different temperatures indicate that the functional difference between Don-0/Col and Ler alleles is likely caused by one or several amino acid substitutions affecting ICA2 structure. Considering that Thg1 proteins from bacteria, animals, and fungi function as tetramers to catalyze the addition of a single G nucleotide to the 5′ end of the tRNAHis (Heinemann et al., 2012; Lee et al., 2017), we hypothesize that structural mutations in ICA2 in Arabidopsis might alter the interactions among Thg1 units and/or the catalytic activity of the protein complex. Supporting this hypothesis, the amino acid substitution Arg-31 to Cys-31 found in ICA2 is located in the activation site of ICA/Thg1 proteins, which binds to a GTP molecule through the activation pocket formed by amino acids 29 to 37 (Lee et al., 2017). Nevertheless, we cannot exclude the possibility that natural alleles of ICA2 might also differ in terms of the precise cellular and temporal regulation of gene expression or in transcript splicing, because artificial modification of ICA2 expression levels showed significant phenotypic effects.
On the other hand, the natural variation in ICA2 affects growth mainly at high temperature, hence contributing to the variation in plasticity in relation to temperature. This is reflected by the severe growth defect displayed only at high temperature by plants harboring loss-of-function alleles in both ICA genes. Also, the ica2-1 mutant and ICA2 natural variants affect relevant traits targeted by high temperature, such as leaf size, in the presence of functional ICA1 alleles. These phenotypes suggest that ICA2-Don-0/Col increases thermomorphogenesis and leaf growth plasticity. In addition, the more active ICA2-Ler allele also accelerates flowering initiation compared with ICA2-Don-0 when combined with ICA1 loss-of-function alleles at high temperature. Most likely, the strong flowering time effect caused by ICA1 and ICA2 interaction in the ica1 background underlies a large-effect quantitative trait locus previously identified specifically at 28°C in an experimental Don-0/Ler population (Méndez-Vigo et al., 2016). This effect is not only a consequence of the potential developmental delay produced by the severe growth defect, because the ica1-2 mutant and lines carrying ICA2-Don-0 transgenes in the ica1-2 background exhibited delayed flowering, but also developed more leaves than the wild type (Supplemental Figure 3B). Interestingly, this quantitative trait locus has been shown to genetically interact with the flowering integrator FLC, which also affects the thermal plasticity of flowering (Méndez-Vigo et al., 2016). Thus, ICA2 and ICA1 appear to be a pair of molecular components that contribute to the natural variation in the temperature-mediated regulation of leaf growth and flowering time.
The significant associations between ICA2 haplogroups and environmental variables related to the daily and annual fluctuations in temperature suggest that the natural variation in ICA2 is involved in adaptation to these climatic factors. It must be emphasized that the strong growth defect caused by the ICA1 and ICA2 interaction has been shown to be quantitatively variable and reversible depending on temperature (Zhu et al., 2015). This further supports the notion that, under natural conditions, ICA2 is likely more involved in regulating the cell cycle and plant growth in response to thermal fluctuations versus average temperatures. Nevertheless, deciphering the evolutionary processes that maintain the natural variation in ICA2 awaits additional studies quantifying the fitness effects of ICA2 in natural and/or fluctuating environments.
It remains unknown how Arabidopsis ICA genes affect growth and development differentially depending on ambient temperature. Various outcomes from the analyses of accessions and transgenic lines support the notion that this is not determined by the differential regulation of ICA expression. First, ICA2 and ICA1 did not show significant differences in expression, indicating a rather similar contribution of both genes at the transcriptional level. Second, ICA2 and ICA1 displayed comparable expression levels at 21 and 28°C, which implies that the growth defect caused specifically at high temperature is not a consequence of reduced expression of ICA2 at 28°C. These results suggest that other proteins that interact with the ICA tetramer and show temperature-mediated transcriptional or posttranscriptional regulation might provide the temperature conditionality of the phenotypic effects of ICA. Favoring this view, Thg1 proteins from yeast and human have been shown to interact with the origin recognition complex involved in DNA replication and cell division (Rice et al., 2005) and with mitofusin 2, a fundamental component for mitochondria fusion (Edvardson et al., 2016). Alternatively, given the function of Thg1 genes in tRNAHis maturation (Heinemann et al., 2012; Lee et al., 2017), the temperature dependence of ICA might be determined through the regulatory functions mediated by uncharged or modified tRNAs or the recently identified category of functional small RNAs derived from tRNAs, which have been shown to regulate gene expression under stress conditions (Heinemann et al., 2012; Raina and Ibba, 2014; Alings et al., 2015; Zhu et al., 2018). Currently, a large body of studies on tRNA biology have linked tRNA metabolism with the transcriptional and translational regulation of specific genes, as well as with cancer and other diseases, including the association of a human Thg1 mutation with cerebellar ataxia (Edvardson et al., 2016; Orioli, 2017). However, the relevance of such interactions and regulations is mostly unknown in plants, and further research is needed to elucidate the precise functional mechanisms underlying the universal ICA/Thg1 genes.
METHODS
Plant Materials
The parental Arabidopsis (Arabidopsis thaliana) accessions Don-0, Ler, Col, and Sij-4, as well as the loss-of-function T-DNA mutant ica1-2 in the Col background, were described previously (Zhu et al., 2015). The T-DNA insertion mutant ica2-1 in the Col background was obtained from Nottingham Arabidopsis Stock Centre (N680284) as line SALK_131808. A previously developed 35S:amiR-ICA1 line in the Col-0 background (Zhu et al., 2015) was used to silence ICA1 in different wild accessions. This line was crossed with the following accessions: Col, Sij-4, Uod-1 (N6975), and Pi-0 (N7298) belonging to the ICA2-Don haplogroup and Ler, Bd-0 (N7013), Ga-0 (N6919), Kondara (6929), and Sij-2 (N10009) in the ICA2-Ler haplogroup. This artificial microRNA targets the ICA1 sequence 5′-GGGAAGAACACCTACACTTC-3′, which is fully conserved among crossed accessions and absent in ICA2 gene. The artificial microRNA was a single gene target, as shown by WMD3 (http://wmd3.weigelworld.org).
Fine mapping of ICA1 and ICA2 was performed using selfed progeny of 2200 plants obtained from the recombinant inbred line Ler/Don-0-70 derived from a cross between the Don-0 and Ler accessions (Méndez-Vigo et al., 2016). This line was heterozygous for a genomic region spanning the ICA1 and ICA2 loci but homozygous Ler (60%) or Don-0 (35%) for the rest of the genome. Genetic markers developed for fine mapping are described in Supplemental Table 2.
The genetic interaction between Ler and Don-0 alleles of ICA1 and ICA2 was tested by analyzing plants carrying the four genetic combinations that were homozygous Ler or Don at these loci in an otherwise homogeneous background. These plants were obtained from the selfed progeny of six recombinants selected from the fine mapping population as heterozygous Ler/Don-0 for ICA2 and homozygous ICA1-Ler (recombinants classes b, c, and d in Figure 1B) or ICA1-Don (recombinants h, i, and j in Figure 1B). Between 25 and 30 offspring plants from each recombinant were grown at 28°C, phenotyped, and genotyped with the markers developed for fine mapping (Supplemental Table 2). At least 14 offspring plants of each of the four homozygous genotypes were used for phenotypic comparison.
Growth Conditions and Phenotypic Analyses
Plants were grown in pots with soil and vermiculite at a 3:1 proportion in growth chambers supplemented with light to provide a long-day photoperiod (16 h of cool-white fluorescent light, photon flux of 80 µmol/m2 s) and set up at different temperatures of 21 or 28°C.
The general phenotypic effects of ICA genes on growth-related traits at different temperatures were analyzed as described previously (Zhu et al., 2015). Vegetative growth was quantified by measuring the maximum rosette diameter (for plants in the ica1-2 background) or the length of leaves at the fifth to sixth position (in ica2-1 background) after flowering initiation. In the latter case, we estimated the leaf area and the lengths of petioles and blades using ImageJ 1.46 software (http://rsb.info.nih.gov/ij/). Flowering time was calculated as the number of days from seedling planting to opening of the first flower, and total leaf number was measured as the final number of rosette plus cauline leaves. Growth during the reproductive phase was estimated based on the length of the first 10 fruits in the main inflorescence, which is an indirect measurement of fertility and seed production (Alonso-Blanco et al., 1999). For phenotypic analyses, all transgenic lines and parental controls were grown simultaneously in the same experiment. A three complete blocks design with randomization was used, with each block containing one pot with six plants per line.
DNA Content and DNA Damage Measurements
DNA contents in cells were measured by flow cytometry using leaves of 14-d-old plants. Fresh tissue from a single plant grown at 21 or 28°C was chopped in Galbraith nuclear buffer (Galbraith et al., 1983) with a sharp razor. Nuclei were stained with propidium iodide and loaded into a cytometer (Cytomics FC500, Beckman Coulter) to obtain the distribution of DNA content in cell populations. The numbers of cells with high ploidy levels (C16, C32, and C64) were statistically compared among genotypes and transgenic lines using the average counts of three biological replicates (samples collected from different plants) in chi-square tests in contingency tables.
The first leaf assay was used to measure DNA damage sensitivity using the method of Rosa et al. (2013) as described in Zhu et al. (2015). In brief, 4-d-old seedlings grown in plates with 0.5× Murashige and Skoog (MS) solid medium were treated with a DNA damaging agent, the radiomimetic drug BLEO (Millipore; 1 µg/mL in 0.5× MS liquid medium), for 5 d. The BLEO-treated seedlings were thoroughly washed with 0.5× MS liquid medium and transferred to solid MS plates. The development of the first pair of leaves was scored 24 h later. Plants for this experiment were grown at 21°C in a growth chamber supplemented with light to provide a long-day photoperiod (16 h of cool-white fluorescent light, photon flux of 80 µmol/m2 s).
ICA2 Sequences, Constructs, and Transgenic Lines
An ICA2 genomic fragment of 5.6 kb was sequenced from Don-0 and Ler, including 1.3 kb of promoter sequence and 5′ untranslated region; 3.5 kb of coding sequence; and 0.8 kb of the 3′ region. Both the 5′ and 3′ regions overlapped with neighboring genes. For sequencing, 10 overlapping 0.8- to 1.2-kb fragments were PCR amplified (Supplemental Table 3), and the products were sequenced using an ABI PRISM 3700 DNA analyzer. DNA sequences were aligned using DNASTAR 8.0 (Lasergene), and the alignments were inspected and edited by hand with GENEDOC (Nicholas et al., 1997).
The two 5.6-kb ICA2 genomic fragments from Don-0 and Ler were cloned in the pCAMBIA 3300 binary vector (CAMBIA) using standard molecular biology techniques. These fragments were PCR amplified using Phusion high-fidelity DNA polymerase (New England Biolabs), cloned in BamHI cloning site, and checked by sequencing (Supplemental Table 3).
The ICA2 genomic constructs were transferred by electroporation into Agrobacterium tumefaciens strain AGL0u (Lazo et al., 1991), and ica1-2 or ica2-1 Arabidopsis plants were transformed by the floral dip method (Clough and Bent, 1998). T1 transformants were screened based on glufosinate resistance, and lines carrying single insertions were selected based on segregation of the resistance trait in T2 family plants. Six to eight independent homozygous T3 lines were selected for each construct and genetic background; their transgene and endogenous ICA1 and ICA2 alleles were verified by PCR prior to phenotypic analyses (Supplemental Table 3).
Phylogenetic Analysis
Intraspecific relationships among genomic and protein sequences from the 1001 Genomes Project (1001 Genomes Consortium, 2016) were determined by neighbor-joining analysis using MEGA7 (Tamura et al., 2011) and applying 10,000 bootstrap permutations to determine statistical significances. The haplotype network of ICA2 proteins was constructed using NETWORK 5 (http://www.fluxus-engineering.com) using the median-joining method.
Most ICA/Thg1 sequences from different plant species were obtained from 54 whole-genome sequences that are fully assembled and annotated at Phytozome v12 (phytozome.jgi.doe.gov). Additional sequences from the genome projects of Camelina sativa, Arabis alpina, Brassica napus, and Rapahanus sativus, as well as nonplant model organisms, were obtained from GenBank (Supplemental Table 4). Physical distances between ICA genes of the same species were determined based on their physical positions in genome contigs. Microsynteny analyses were performed using Web-based SynMap software as implemented in the Comparative Genomics CoGe platform (genomevolution.org/coge; Lyons and Freeling, 2008; Haug-Baltzell et al., 2017). The alignment of proteins was obtained with CLUSTAL (https://www.ebi.ac.uk/Tools/msa/clustalo/) and edited by hand with GENEDOC (Nicholas et al., 1997; Supplemental File). The phylogenetic relationships among ICA proteins were determined by constructing an ML tree using MEGA7 (Tamura et al., 2011) and applying 10,000 bootstrap permutations for statistical significances.
Gene Expression Analysis
For expression analyses of ICA genes, plants were grown as described for phenotypic analyses but in pots containing ∼50 plants, and plants from two blocks were mixed before RNA isolation. RNA from 14- to 18-d-old plants grown at different temperatures was isolated from whole rosette tissue using TRIzol reagent (Invitrogen) according to the manufacturer’s protocol. Potential DNA contamination was removed by DNase digestion, and subsequent RNA purification was performed using a High Pure RNA isolation kit (Roche). cDNA was synthesized from 3 µg of total RNA using avian myeloblastosis virus reverse transcriptase (Invitrogen) and dT15 oligonucleotides. ICA1 and ICA2 expression was analyzed by quantitative RT-PCR using one and two pairs of oligonucleotides, respectively (Supplemental Table 3). To avoid amplification differences caused by DNA polymorphisms, primers were designed in gene regions differentiated between ICA1 and ICA2 but carrying no polymorphism among Col, Ler, and Don-0. The ubiquitin-conjugating enzyme 21 gene (At5g25760), whose expression levels are highly steady among organs and environments (Czechowski et al., 2005), was used as an endogenous control for RNA sample standardization (Supplemental Table 3). All genes were amplified with Power SYBR Green Mix in a 7300 real-time PCR system (Applied Biosystems) and quantified using the standard curve method. Means and ses were derived from three biological replicates (RNA isolated from plants grown in different pots) for parental accessions or from three technical replicates (quantitative RT-PCR wells from the same cDNA sample) for transgenic lines.
Climatic and Statistical Analyses
To test the association between ICA2 polymorphisms and climate, we extracted the 19 BIO climatic variables of the original accession locations from the world climate database (http://www.worldclim.org). General linear models were applied using climatic measurements as dependent variables and single-nucleotide polymorphisms as independent explanatory factors. To reduce the effect of genetic structure on climatic associations, the ancestry membership proportions of each accession in nine genetic groups (1001 Genomes Consortium, 2016) were included as covariates in general linear model analyses. Since the ICA2-Ler haplogroup was mainly present in genetic groups 3, 5, and 6, tests were performed using all 687 accessions with the two major ICA2 haplogroups, or only the 159 accessions from those three genetic groups.
The phenotypic effects of transgene, transgenic line, and temperature were tested statistically via mixed model analyses of variance, including transgene, temperature, and temperature × transgene interaction as the fixed effects and line (nested within transgenes) as the random effect factor (Supplemental Table 1). Tukey’s honest significant difference test was used to compare the differences among the two types of transgenic lines grown at two different temperatures within the interaction term. These analyses were performed with the statistical packages SPSS 24 or Statistica 8.
Accession Numbers
Sequence data from this article can be found in the GenBank/EMBL libraries under the following accession numbers: MK056185 and MK056186.
Supplemental Data
Supplemental Figure 1. Analysis of the ica2-1 insertional mutant.
Supplemental Figure 2. Search for the ica1-2 ica2-1 double mutant.
Supplemental Figure 3. Differential complementation by natural alleles of ICA2.
Supplemental Figure 4. Flow cytometry analyses of nuclear DNA content at 21°C.
Supplemental Figure 5. Microsynteny relationships within ICA1/ICA2 genomic regions in Camelineae species.
Supplemental Table 1. ANOVA tables.
Supplemental Table 2. Markers used for ICA2 fine mapping.
Supplemental Table 3. Oligonucleotides used for ICA2 analyses.
Supplemental Table 4. GenBank codes of ICA/Thg1 proteins.
Supplemental File. Alignment of ICA2 proteins from plants and model organisms.
Dive Curated Terms
The following phenotypic, genotypic, and functional terms are of significance to the work described in this paper:
Acknowledgments
We thank Mercedes Ramiro for technical assistance. This work has been funded by the Agencia Estatal de Investigación of Spain and the Fondo Europeo de Desarrollo Regional (Unión Europea) (grant BIO2016-75754-P to C.A.-B.) and by the Australian Research Council (Discovery grant DP0983875 and ARC-Future Fellowship FT100100377 to S.B.).
AUTHOR CONTRIBUTIONS
C.A.-B., B.M.-V., I.A., and S.B. designed the study. B.M.-V., I.A., A.M.-M., and W.Z. performed the experiments. C.A.-B., B.M.-V., I.A., S.B., and W.Z. analyzed and interpreted the data. B.M.-V. and C.A.-B. wrote the article, with contributions from all other authors.
Footnotes
- Received February 26, 2019.
- Revised March 29, 2019.
- Accepted April 12, 2019.
- Published April 16, 2019.