Molecular Evolutionary Analysis of the Alfin-Like Protein Family in Arabidopsis lyrata, Arabidopsis thaliana, and Thellungiella halophila

In previous studies, the Alfin1 gene, a transcription factor, enhanced salt tolerance in alfalfa, primarily through altering gene expression levels in the root. Here, we examined the molecular evolution of the Alfin-like (AL) proteins in two Arabidopsis species (A. lyrata and A. thaliana) and a salt-tolerant close relative Thellungiella halophila. These AL-like proteins could be divided into four groups and the two known DUF3594 and PHD-finger domains had co-evolved within each group of genes, irrespective of species, due to gene duplication events in the common ancestor of all three species while gene loss was observed only in T. halophila. To detect whether natural selection acted in the evolution of AL genes, we calculated synonymous substitution ratios (dn/ds) and codon usage statistics, finding positive selection operated on four branches and significant differences in biased codon usage in the AL family between T. halophila and A. lyrata or A. thaliana. Distinctively, only the AL7 branch was under positive selection on the PHD-finger domain and the three members on the branch showed the smallest difference when codon bias was evaluated among the seven clusters. Functional analysis based on transgenic overexpression lines and T-DNA insertion mutants indicated that salt-stress-induced AtAL7 could play a negative role in salt tolerance of A. thaliana, suggesting that adaptive evolution occurred in the members of AL gene family.


Introduction
Transcriptional control of gene expression underlies different metabolic and developmental processes in plants, including responses to environmental stimuli or stressors [1][2][3]. In the regulation process, trans-acting factors sustain, enhance, or repress the expression of genes encoded by the genome. This responsive control over gene expression by these small pieces of genetic material creates a powerful and dynamic mechanism for higher terrestrial plants to adapt to their environment, in addition to more fundamental and genetic responses [4]. Indeed, the evolution of regulating gene may be more powerful and effective than direct structural gene evolution [5][6][7].
Transcription factors can be organized into families based on their characteristic domains, and currently, roughly 84 transcription factor families have been discovered [8]. The Alfin-like protein family was discovered to be a transcription factor in alfalfa (Medicago sativa) in the form of a 7S storage protein [9,10]. The members of this family have two sequences of approximately 130 and 50 conserved amino acid residues at their N-termini and Ctermini, respectively, named DUF3594 domain and PHD-finger motif. Originally discovered in a homeodomain protein from A. thaliana [11], the latter motif mediates binding capability to specific nuclear protein partners [12,13]. Besides its protein-protein interaction role, PHD-finger motif also binds to the core consensus cis-acting element (C/A) CAC in the promoter of the target gene [14]. Although the former DUF3594 domain is functionally uncharacterized, the highly conserved nature of DUF3594 domain and PHD-finger motif in different species indicates that Alfin-like proteins probably have fundamental biological functions in plants.
Recently, the genes of Alfin-like proteins have been found in many other land plants, such as Arabidopsis, rice, and Glycine max [2,[15][16][17][18]. None of AL (Alfin-like protein) protein containing DUF3594 domain were reported in animals, fungi, and prokaryotes [12]. Except for Alfin1 in alfalfa, the role of these factors in the mediation of biological processes in plants remains unclear. After isolating Alfin1 cDNA from alfalfa salt-tolerant cells [15], gel retardation assays showed that Alfin1 protein binds to DNA in a sequence-specific manner [14]. Further, the 35S::Alfin1 transgenic alfalfa has improved salinity tolerance whereas calli expressing Alfin1 in the antisense orientation were more sensitive to salt, demonstrating that Alfin1 functions in salt tolerance in alfalfa [15]. Additionally, the overexpressing lines of Alfin1 displayed enhanced plant root growth under normal and saline conditions while the antisense transgenic plants grew poorly, suggesting that Alfin1 expression is essential for normal alfalfa development [19]. Promoter specificity and efficiency of utilization tests indicated Alfin1 protein can bind to the MsPRP2 promoter directly and enhance its gene expression [19]. In addition, since the orthologs of Alfin1 had been shown to bind to the promoters of H3K4me2 and H3K4me3, the proteins containing the PHD domain might be involved in the process of chromatin regulation both in plants and animals [13,16,20]. The evolution of transcription factor genes appears frequently to involve gene duplication and diversification [7]. The slight modification of these trans-acting factor genes could in turn regulate large suites of structural genes, giving rise to complex traits. Duplication of regulatory genes has been investigated as a major evolutionary event providing raw material for genetic variations and adaptation [3,[21][22][23][24]. In order to understand the evolutionary events of AL transcription factor genes, we first collected and identified twenty AL family members in three species: A. lyrata (7), A. thaliana (7), and Thellungiella halophile (6). T. halophila, a close relative of the other two Arabidopsis species [25], exhibits higher resistance to cold, drought, and salinity [26]. The comparison of AL genes, transcription factors known to be related to salt-tolerance, found in closely-related species with clear differences in their physiological salt tolerance should provide insight into the molecular mechanism of adaptive evolution in response to strong environmental selection. A phylogenetic tree was constructed to evaluate the evolutionary relationships of all observed AL genes and evidence for gene duplication and loss events was examined. Further, we computed synonymous substitution ratios (dn/ds) and codon usage statistics to detect positive selection and where it might have occurred on the phylogenetic tree and in which of the two known domains. Finally, overexpression lines and T-DNA insertion mutants of A. thaliana for AL7 gene found to be under positive selection on the PHD finger domain were tested for salt tolerance to understand the physiological effects of the evolutionarily significant gene.

Sequence Alignment and Phylogenetic Analysis
Full-length amino acid sequences were aligned by the Clustal X 1.83 program [27]. The sequence alignment was further adjusted manually using BioEdit software (http://www.mbio.ncsu.edu/ bioedit/bioedit.html ). The amino acid substitution model was calculated by the ModelGenerator v0.84 and the optimal model of ''JTT+G'' was selected [28]. Phylogenetic relationships were reconstructed using a maximum-likelihood (ML) method in PHYML software with JTT amino acid substitution model [29]. One thousand bootstrap replicates were performed in each analysis to obtain the confidence support. The Alfin1 gene from Medicago sativa was used as an out-group.

Co-evolution Analysis
The hypothesis that the co-evolved relationships might exist among the domains of proteins during the evolutionary process was tested by Goh and colleagues [30]. Here, we used Goh's method to test the correlation between every domain-domain pair for the AL family. After the alignments of the DUF3594 domain, PHD-finger domain, and inter-domain, pairwise evolutionary distances for the alignments were calculated using MEGA version 5.0 program [31]. Afterward, we calculated the Pearson correlation coefficients (r) of linear and Mantel analysis between the distance matrices of all potential interacting domains using the SPSS version 13.0 software and PAST Version 2.17 (http://folk. uio.no/ohammer/past/) program (Euclidean, Permutation N: 5000).

Estimates of Amino Acid Substitution Rates
The nonsynonymous to synonymous substitution rate ratio v (d N /d S ) is commonly used as an indicator of the selective strength in coding sequences [32]: v.1 indicates positive selection, v,1 indicates purifying selection, and v = 1 indicates neutral evolution.  To determine whether positive selection had acted at specific sites in the AL proteins, two codon-based likelihood methods were run using the CODEML package of PAML version 4.2 [33]: site models [34] and branch-site models [35]. For the site models method, six models were explored: the one-ratio model (M 0 ), the discrete model (M 3 ), the nearly neutral model (M 1a ), the positiveselection model (M 2a ), the beta model (M 7 ), and the beta and v model (M 8 ). To evaluate variation in selective pressure over a phylogeny, the branch-site models were used to estimate v under different assumptions. The branch models allow the v values to vary among branches in the phylogeny and are useful for detecting positive selection acting on particular lineages. To verify which of the models best fitted the data, Likelihood Ratio Tests (LRT) were performed by comparing twice the difference in log likelihood values between pairs of the models using a x 2 distribution, with the degrees of freedom equal to the differences in the number of parameters between the models [36]. The Bayes Empirical Bayes (BEB) analysis procedure was used for identifying sites under positive selection with significant LRTs. Each branch group was labeled as foreground in turn as well.

Synonymous Codon Usage Estimation and Identification of Gene Conversion Events
Codon usage bias, where certain codons are used preferentially, was calculated by the effective number of codons (ENC) [37]. The frequency of G+C at the third synonymous variable codon position (GC3) and codon adaptation index (CAI) were also used to estimate the codon bias. All three analyses were performed on the online platform of EMBOSS explorer (http://emboss. bioinformatics.nl/cgi-bin/emboss). A graphical comparison of ENC and GC3 was used to control for the possible codon bias because it is often associated with GC3 [37]. Additionally, the program GeneConv was used to analyze the possible gene conversion events (http://www.math.wustl.edu/sawyer/ geneconv/).

Plant Materials and Treatments
A. thaliana accessions Columbia (Col)-0 (from Arabidopsis Biological Resource Centre at Ohio State University) seeds were germinated. The aboveground parts of three-week-old A. thaliana grown at 22uC were harvested and soaked in petri dishes with 25% (w/v) PEG8000 for 0, 1, 2, 4, 12 h; 0 mM, 100 mM, 200 mM, 300 mM NaCl solutions for 4 h; and 300 mM NaCl for 0, 1, 2, 4, 12 h as described in previous studies [38,39]. All of these plant materials were frozen rapidly in liquid nitrogen and stored at -80uC. For the tolerance assay, three-day-old vector, wild type, mutants and transgenic Arabidopsis seedlings were transferred to Murashige & Skoog media (MS) plates supplemented with 150 mM or 200 mM NaCl for 5 days at 2262uC with 16 h of light and 8 h of darkness [40]. Root length data were analyzed using SPSS version 13.0 software. We first calculated a tolerance index of primary root, resulting in a set of normally distributed data. The index was calculated as: (primary root length at stress condition -primary root length at normal condition)/primary root length at normal condition. We then used one way ANOVA to examine the type of seedlings and salt treatment on the tolerance index of primary root, using Bonferroni multiple comparison tests.

Northern Blot and RT-qPCR
Total RNA was isolated by phenol/chloroform extraction and LiCl precipitation. For the Northern blot analysis, 20 mg of total RNA was mixed with two times the volume of RNA denaturing sample buffer (Embitec) and was incubated in a 68uC water bath for 15 minutes and ice bath for 5 minutes. Then, the mixture was separated on formaldehyde (18.6% v/v) agarose gels (1.5% m/v), stained with ethidium bromide, and blotted to nylon membrane. The nylon membranes were hybridized with [a-32 P]dATPlabelled AtAL7 specific probes in PerfectHyb plus hybridization buffer (Sigma-Aldrich) at 68uC for 16 hours. The membrane was then washed for 10 minutes twice with 26SSC (16SSC is 0.15 M NaCl and 0.015 M sodium citrate) and 1% SDS and for 10 minutes with 0.16SSC and 1% SDS at 68uC. For transformant probes were PCR amplified from A. thaliana reverse transcription products with the following primers: 59-GCGGATCCATG-   For RT-qPCR, total RNA extracted by the use of RNAiso Plus kit (Takara). According to the method described by Michael [42], the first-strand cDNA was synthesized from 1.5 mg DNase-treated RNA in a 20 mL action volume using M-MuLV reverse transcriptase (Fermentas, now Thermo Scientific, http://www. thermoscientificbio.com) with oligo(dT)18 primer. Relative transcripts levels were determined using the iCycler IQ Real-time PCR Detection System (Bio-Rad, USA) according to the manual QuantiTect SYBR Green PCR kit and analyzed by icycler realtime detection system software (version 3.0). ACTIN2 was used as a control. Gene-specific primers used to detect transcripts are listed in Table S1 in File S1.

Co-evolution of Separate Domains Among Four Groups of AL Proteins
The phylogenetic reconstruction of the twenty AL proteins observed in the three species clearly separated into four distinct groups: I, II, III and IV (Fig. 1A), using ML estimation and Alfin1 of alfalfa as an out-group with an in-group consisting of seven, seven, and six AL protein sequences of A. thaliana, A. lyrata, and T. halophila, respectively. All genes of the members on the tree had five exons separated by four introns (Fig. 1B). All the clades except AL3 clade had three orthologs from A. thaliana, A. lyrata, and T. halophila, indicating that the orthologs of each clade may have originated from one gene in an ancestral species.
Additionally, all of the AL proteins possessed both the DUF3594 and PHD-finger domains ( Figure S1 in File S1), and separate phylogenetic reconstruction of the two domain recovered similar evolutionary histories as the entire AL sequence (Fig. 2). The phylogenetic reconstruction for the PHD finger sequences were less resolved (Fig. 2B) and the only direct conflict between the two trees is not strongly supported. This evolutionary congruence indicates that little or no recombination has occurred within and among the four groups of AL genes. Further, the pairwise distances between sequences of the DUF3594 domain were strongly correlated with the pairwise distances among PHD-finger domain sequences and their inter-domain, using either direct linear correlation or a Mantel test (r values reported respectively; r = 0.745 and 0.7933 for the DUF3594 domain and inter-domain; r = 0.7850 and 0.7038 for the PHD-finger domain and interdomain, and r = 0.8522 and 0.8043 for the DUF3594 and PHDfinger domains; p,0.001 for all correlations). Notably, the correlation coefficients of both tests were higher between DUF3594 and PHD-finger sequences than for comparisons with the inter-domain sequence. The two domains in these genes are clearly evolving together and have similar histories.

Evolutionary Patterns of AL Genes in A. thaliana
Eight pairs of paralogous genes (three for A. thaliana, three for A. lyrata, and two for T. halophila) were identified at the tips of the phylogenetic tree (Fig. 1A). We tested the segmental duplication events and compared the flanking 10 protein-coding genes of both the AL gene of A. thaliana and its paralogs. There were three pairs of genes flanking AtAL1 on chromosome 5 and AtAL2 on chromosome 3 which showed high conservation ( Table 1, Table  S2 in File S1), suggesting that these two AL genes were formed through segmental duplication in A. thaliana. For other pairs of AL paralogous genes in A. thaliana, there was no evidence supporting their origin from duplicated blocks. These results revealed that several members of the AtAL family arose through segmental duplications. Also, we searched the orthologs of the three pairs of flanking genes existing beside AtAL1 and AtAL2 in A. lyrata and T. halophila (Table 1), and the highly conserved hits were detected, indicating that similar segmental duplication event could have occurred in the ancestor of the three species.
We used GeneConv software to investigate the possible gene conversion events that might have occurred during the evolution of AL genes, but no such event was detected. Peculiarly, the phylogenetic tree Group I contained 8 members, but only AtAL3 and AlAL3 have no ortholog in T. halophila (Fig. 1). We detected 20 genes located in the flanking region of AtAL3 on chromosome 3 in the NCBI database and found 16 pseudogenes (Table S3 in File S1). It has also been reported previously that AtAL3 lacked the key conserved Tyr residue on its PHD-finger domain [16]. All of these evidences suggest that a gene loss event had occurred in the AL gene family during the divergence between A. thaliana and T. halophila or after the divergence of T. halophila from Arabidopsis lineage.

Testing for Selection in the AL Genes of A. lyrata, A. thaliana, and T. halophila
In order to detect the evolutionary driving forces in the divergence of the AL gene family, we used the BEB method to predicted positively selected codon sites on AL genes of T. halophila, A. thaliana, and A. lyrata. Both site-specific models and branch-site models were compared to evaluate the evolutionary forces for the former assuming variable selective pressures among sites and the latter examining the selective pattern among branches in the phylogeny. As shown in Table 2, the site-specific models failed to detect any site under positive selection, but showed signs of selective sweep on most of the AL gene sites (89.23% sites with v,1). Moreover, we implemented the branch-site model to detect whether the positive selection acted on some sites of specific clades in AL gene phylogeny. As summarized in Table 3, four amino acids were identified by BEB analysis as candidates for positively selected sites with Bayesian posterior probability .0.95. With one exception in the PHD-finger domain of AL7 (alignment position 233K in Figure S1 in File S1), the amino acids were located in the Table 4. Summary of the gene length, GC content, and codon usages P-value among three species' AL gene.   Figure S1 in File S1).

Comparisons of Nucleotide Variation and Synonymous Codon Usage in the AL Gene Family of Three Species
The observed differences in GC content at the three positions and the effective number of codons (ENC) among the AL members of A. lyrata, A. thaliana, and T. halophila indicated a narrow range of GC3 and codon usage patterns. Mean GC3 content and ENC were significantly different between the members of A. thaliana and T. halophila, with values of 0.428 and 53.7 for AtALs and 0.476 and 55.9 for ThALs, respectively (Table S4 in File S1). Further, we detected that the GC3 difference primarily came from the DUF3594 domain rather than the PHD finger and interdomain, and a significant difference existed between the domains of A. lyrata and T. halophila. In addition, the possible duplication fragments of A. lyrata and T. halophila also presented significant GC3 content differences (Table 4). GC3 and ENC values charted for all AL genes were significantly different from the expected ENC curve which represents the null hypothesis that GC3 bias was entirely due to mutation rather than selection, indicating that selection was likely driving biased codon usage. Overall, the AL genes are tightly clustered in a narrow range of GC3 content and ENC value. The only visible trend was that the points for ThALs were more loosely clustered than any other AL genes of the two species and lay nearest to the expected value (Fig. 3).

Overexpression of AtAL7 Suppressed Plant Root Growth Under Normal and Saline Conditions
Previous study reported that Alfin1 functions in salt tolerance of alfalfa [19]. Here, we examined the induced expression of AL genes in A. thaliana under treatments with NaCl solution by quantitative real-time PCR (RT-qPCR). As shown in Figure 3, AtAL5 (Fig. 4A) and AtAL7 (Fig. 4B) expression were strongly upregulated by 300 mmol/L NaCl solution treatment, and the levels of AtAL7 transcripts increased with prolonged treatment time (Fig. 4B) as well as increased NaCl concentration (Fig. 4C). Northern blot results also indicated that AtAL7 was induced by salt stress and inhibited by osmotic stress (Fig. 4D). It was also shown that only AL7 from the seven AL gene clades was under positive selection on the PHD-finger domain ( Figure S1 in File S1), and the three members on this branch showed the smallest difference in codon bias, as shown in the graph among the seven clusters (Fig. 3). We therefore selected the AtAL7 gene for further functional analysis.
To determine the physiological role of AtAL7 in transgenic A. thaliana under salt stress, we generated transgenic A. thaliana plants overexpressing the AtAL7 gene under the CaMV 35S promoter. Among the nine randomly selected primary T1 transformants from the pool of forty two transformants (Fig. 4E), we selected three individual overexpressing lines of AtAL7 using Northern blot analysis for collecting the T3 generation seeds. Then we further compared the root morphology of the vector, 35S-AL7-2, 35S-AL7-4, and 35S-AL7-7 plants grown on MS medium with 0 mM, 150 mM, and 200 mM NaCl ( Figure S2 in File S1). Three 35S-AL7 transgenic lines exhibited shorter root lengths than vector plants growing on all culture media (Fig. 5). Relative primary root lengths of the 35S-AL7 transgenic lines were significantly shorter than the vector plant seedlings on MS agar plates with 200 mM NaCl, that suggesting over-expression of AtAL7 in A. thaliana could reduce plant tolerance to salt stress.

The T-DNA Insertion Mutants of AtAL7 Enhanced Plant Root Growth Under Normal and Saline Conditions
To confirm the negative role of AtAL7 in plant salt tolerance, we identified two T-DNA mutants ( Figure S3 in File S1). Both the al7-1 mutant (Salk_127650) and al7-2 mutant (Salk_127657) contain a T-DNA insertion in the first exon at the 59 untranslated region of the AtAL7 gene. Additionally, we identified another T-DNA mutant of AtAL3 gene as a control. The al3 mutant (Salk_139843c) contains a T-DNA insertion in the fourth intron of the AtAL3 gene. Homozygous mutant plants were identified by PCR with AtAL7 or AtAL3 specific primers. We further compared the root morphology of the al3 mutant, al7-1 mutant, al7-2 mutant, and wild type plants grown on MS medium with 0 mM and 150 mM NaCl ( Figure S4 in File S1). The T-DNA insertion mutants of AtAL7 exhibited longer root lengths than al3 mutant and wild type plants growing on all culture media (Fig. 6), suggesting that AtAL7 could play a negative role in A. thaliana resistance to the salt stress.

Discussion
It had long been known that AL proteins play important roles in regulating signals of plant salt tolerance [14,19,43]. The AL transcription factor is a two-domain protein with the ability to bind to the highly methylated forms of histones and functions in plant salt tolerance [16,44]. Previous work has shown that the coevolution might have taken place among the protein family members which have more than two domains [45][46][47]. Here the two phylogenetic trees, one for each domain (Fig. 2 A and B), exhibited all three distinct groups and each group contained the same members in the PHD-finger and DUF3594 domain trees, implying a co-evolutionary relationship between the two domains in AL trans-acting factors, similar to that with the tubby like proteins [48]. Correlations between both domains also supported their co-evolution according to the method by Goh et al. [30] and Mantel test. Moreover, results of branch site model analysis detected more positive selection sites on the DUF3594 domain than the PHD-finger and linking region (Table 3), and the significant difference in codon usage bias between T. halophila and A. lyrata or A. thaliana occurred on the DUF3594 domain rather than the PHD-finger domain (Table 4). These findings suggest that functional divergences of AL proteins primarily came from the DUF3594 domain.
DUF3594 domain, approximately 140 amino acids in length, is functionally uncharacterized in eukaryotes, while PHD-finger domain is thought to facilitate protein-protein interaction with triand dimethylation of histone H3 at lysine 4 (H3K4me3/2) [13,16,49]. In plants, Alfin1 containing a canonical PHD finger was reported to bind to promoter elements of MsPRP2 gene [14]. Another study suggested that Alfin1 is a transcription co-activator but not a transcription activator [16]. Here, we detected three positively selected sites on DUF3594 domain and one site on PHD-finger domain, which implied that possible adaptive evolution occurred on these two domains. It is known that AtAL3 lacks the key conserved Tyr residue on its PHD-finger domain and does not bind to H3K4me3 [16]. In our gene function analysis, however, the mutant plants of AtAL3 displayed slightly enhanced salt tolerance as the mutants of AtAL7 (Fig. 6), suggesting that the possible adaptive evolution tend to occurred on DUF3594 domain rather than PHD-finger.
From the evolutionary viewpoint, gene duplication, gene transfer and gene losses play key roles in the evolution of gene families and accelerate the turnover of gene birth and death of the family members' evolution [50,51]. There are three types of gene duplication: transposition events, segmental duplication, and tandem duplication [52]. In our analysis, we found that AtAL1 and AtAL2, with three pairs of high conserved genes in its flanking region (Table 3), had expanded through segmental duplication in the three species. In contrast, gene loss has been put forward as a common response to changes from duplicated genome segments in A. thaliana [53], and was also observed in the triplicated genome segments of Brassica oleracea [54]. Here, we identified that only AtAL3 and AlAL3 have no orthologous genes in T. halophila (Fig. 1), multiple genes beside AtAL3 were pseudogenes, and a previous study reported that AtAL3 lacked the key conserved Tyr residue on its PHD-finger domain [16], implying the AL gene loss event could have occurred in the ancestral genome of the Thellungiella lineage.
Multiple studies on the duplication processes in the molecular evolution of plant regulatory genes have argued that trans-acting factors often have increased the rates of non-synonymous substitutions compared with structural genes [6,7,55,56]. In several cases it is clear that neutrally evolving regions play important roles in protein function [3,[57][58][59]. In this study we used site-specific models and branch-site models to detect positive selection among the AL genes of three species. The latter models predicted one site as positively selected for AL1, AL2, AL6, and AL7 of seven branches respectively ( Table 4), suggesting that positive selection has operated on AL genes in the three species during their evolution. Moreover, we also analyzed the codon usage bias which reflects a balance between mutational biases and natural selection for translational optimization [60][61][62]. ENC values and Mean GC3 statistics indicate significant differences between AL genes of A. thaliana and T. halophila (Table 3), and differences also may exist between the AL genes of A. lyrata and T. halophila (Table 3). In Figure 3, the gene cluster of T. halophila was obviously separated from the clusters of A. thaliana and A. lyrata, suggesting the differentiation of biased codon usage between T. halophila and the other two species had already occurred.

Supporting Information
File S1 Figure S1 in File S1. Amino acid sequence alignment for 20 AL proteins by ML methods with bootstrapping analysis (1000 reiterations). The DUF3594 domain and PHD-finger are indicated by yellow and blue boxed letters. The positively selected codon sites are indicated by red arrows. The amino acids in red box display the altered key site as Lee et al [16]. Figure S2 in File S1. Phenotype response of 35S-AtAL7 A. thaliana lines to normal condition (left), 150 mM NaCl (middle), and 200 mM NaCl (right). Seedlings of vector (upper left), 35S-AtAL7-02 (upper right), 35S-AtAL7-04 (bottom left), and 35S-AtAL7-07 (bottom right) transgenic lines were germinated on a 1/2MS agar plate for 3 days, then transferred to another MS agar plate supplemented with 150 mM NaCl (middle) and 200 mM NaCl (right) for 5 days. Figure S3 in File S1. Diagram of AtAL3 and AtAL7 and their T-DNA insertion mutants. Figure S4 in File S1. Phenotype response of A. thaliana AL mutants to normal condition (left) and 150 mM NaCl (right). Seedlings of wild type (upper left), al7-1 (upper right), al7-2 (bottom left), and al3 (bottom right) mutants were germinated on a 1/2MS agar plate for 3 days, then transferred to another 1/ 2MS agar plate supplemented with 150 mM NaCl (right) and without (left) for 4 days. (DOC)