Identification of the GRF genes and miR396s in lettuce
To identify the GRF gene family in lettuce, the GRF genes from Arabidopsis thalina, rice and tomato were firstly used as the query sequences for BLASTP searching in lettuce genome database. Thus, thirteen candidate GRF genes were obtained in lettuce. GRF protein contains two conserved functional domains, QLQ and WRC. Therefore, QLQ (PF08880) and WRC (PF08879) were used for the Hidden Markov Model (HMM) search, and another two candidate LsaGRF genes were identified. Finally, all fifteen candidate LsaGRF genes were applied for BLASTP searching, and no further hits were got. Thus, there are totally 15 LsaGRF genes identified in lettuce genome. Based on their respective location on the chromosomes, we designated them as LsaGRF1 to LsaGRF15 respectively (Fig. 1A). LsaGRFs were distributed on each chromosome of lettuce, except for chromosome 1 and 7, and the most (five LsaGRFs) occurred on chromosome 6 (Fig. 1A). The basic information of LsaGRF1-15, including gene ID, chromosomal location, length of gene and protein, PI value and exon numbers, was listed in Table 1. Most LsaGRFs had 3 or 4 exons, except for LsaGRF4, 7 and 9 (Table 1). In addition, the predicted isoelectric point (pI) values of LsaGRF preoteins were between 6 and 9, except that the pI values of LsaGRF4 and LsaGRF7 were higher than LsaGRF9 (Table 1). The amino acid sequence alignment of 15 LasGRF was showed in Fig. S1. All of the 15 LsaGRFs contained the conserved WRC domain, but only 13 LsaGRFs had complete QLQ domain. There is no QLQ domain in LsaGRF4, and LsaGRF7 has only an incomplete QLQ domain (Fig. 1B).
The GRF genes are known as the target of miR396s. To date, two MIR396s in Arabidopsis (Ath-MIR396s) and seven MIR396s in rice (Osa-MIR396s) are identified. There are five MIR396s, MIR396a to MIR396e, in lettuce genome [36]. The evolutionary relationship between these five Lsa-MIR396s was shown in Fig. S2A based on their stem-loop sequences. The Lsa-MIR396s showed much closer evolutionary relationship with Ath-MIR396s than with Osa-MI396s (Fig. S2B). The target sequence of miR396s locates at the end of the WRC domain (Fig. 1C). Although the stem-loop structures were totally distinct, five Lsa-miR396s were highly conserved in the mature region with only two nucleotides difference (Fig. 1C). Five Lsa-MIR396s were located on chromosome 1, 5, 7 and 8 respectively, among which chromosome 5 contained two Lsa-MIR396s, Lsa-MIR396a and Lsa-MIR396d (Fig. 1A). The identified GRF genes and miR396s in major species were listed and compared with those in lettuce (Table S1). The LsaGRF family was the second largest, and the number of miR396s was comparable to that in rice and tomato.
Table 1
Characteristics of the GRFs in lettuce
Name
|
Accession No.
|
Chr
|
CDS (bp)
|
Exon No.
|
Length (aa)
|
MW (KDa)
|
pI
|
LsaGRF1
|
Lsat_1_v5_gn_2_135541.1
|
Chr02
|
990
|
3
|
329
|
36.06
|
8.76
|
LsaGRF2
|
Lsat_1_v5_gn_3_14660.1
|
Chr03
|
1227
|
4
|
408
|
44.51
|
6.24
|
LsaGRF3
|
Lsat_1_v5_gn_3_85521.1
|
Chr03
|
999
|
4
|
332
|
37.74
|
8.73
|
LsaGRF4
|
Lsat_1_v5_gn_3_133500.1
|
Chr03
|
702
|
2
|
233
|
27.17
|
9.63
|
LsaGRF5
|
Lsat_1_v5_gn_4_92941.1
|
Chr04
|
951
|
3
|
316
|
36.58
|
8.81
|
LsaGRF6
|
Lsat_1_v5_gn_4_159961.1
|
Chr04
|
1551
|
4
|
516
|
56.20
|
7.70
|
LsaGRF7
|
Lsat_1_v5_gn_5_54781.3
|
Chr05
|
603
|
6
|
200
|
22.70
|
9.27
|
LsaGRF8
|
Lsat_1_v5_gn_5_141200.1
|
Chr05
|
990
|
3
|
329
|
37.54
|
8.24
|
LsaGRF9
|
Lsat_1_v5_gn_6_10680.1
|
Chr06
|
459
|
2
|
152
|
16.89
|
8.66
|
LsaGRF10
|
Lsat_1_v5_gn_6_17460.1
|
Chr06
|
897
|
3
|
298
|
33.12
|
6.44
|
LsaGRF11
|
Lsat_1_v5_gn_6_70601.1
|
Chr06
|
1113
|
4
|
370
|
40.08
|
8.23
|
LsaGRF12
|
Lsat_1_v5_gn_6_75441.1
|
Chr06
|
1014
|
3
|
337
|
37.60
|
8.74
|
LsaGRF13
|
Lsat_1_v5_gn_6_81681.1
|
Chr06
|
1029
|
4
|
342
|
36.78
|
7.09
|
LsaGRF14
|
Lsat_1_v5_gn_8_86361.1
|
Chr08
|
1071
|
3
|
356
|
40.63
|
8.97
|
LsaGRF15
|
Lsat_1_v5_gn_9_78540.1
|
Chr09
|
1233
|
4
|
410
|
44.10
|
7.67
|
Phylogenetic analysis, gene structures and motif divergence of LsaGRFs
To explore the evolutionary relationship of GRF gene family in different species, phylogenetic analysis on intron-exon and motif characteristics of LsaGRFs were performed. Phylogenetic analysis of GRF family in lettuce was firstly assessed and visualized using a Neighbor-Joining phylogenetic tree (Fig. 2A). All of the 15 LsaGRFs were divided into two groups (including 9 and 6 LsaGRFs respectively), each of which contained two small subgroups in the phylogenetic tree. Gene structure and motifs were considered to have a divergence during gene evolution. Therefore, the gene structures and motifs were listed in phylogenetic tree’s order (Fig. 2B). Most LsaGRF genes contain three or four exons. The LsaGRF genes containing the same number of exons were in the same group. For examples, LsaGRF4 and LsaGRF9 both have two exons. LsaGRF11, 1, 2 and 13 have four exons and the rests in another group have three exons expect LsaGRF7 (Fig. 2B). In conserved motifs analysis, motif 1 (yellow rectangle) and motif 2 (purple rectangle) were present to be the WRC and QLQ protein domain. As shown in Fig. 2B, the location of LsaGRF genes in phylogenetic tree were determined by the position and numbers of QLQ and WRC domains. All these analyses showed that the evolutionary relationship of LsaGRF genes was highly consistent with the gene structures and motif divergence of GRF genes in lettuce.
The functions of GRF genes in Arabidopsis and rice were extensively studied. The evolutionary relationship of GRF genes in lettuce, Arabidopsis and rice were helpful for putative function prediction of GRF genes in lettuce. There are totally 12 and 9 GRF genes in rice and Arabidopsis, respectively. The Neighbor-Joining phylogenetic tree of GRF genes from lettuce, rice and Arabidopsis showed that there were six groups according to the tree (Fig. 3). It was revealed that the evolutionary relationship of GRF genes in lettuce, rice and Arabidopsis were divergence. There were two groups, group I and VII, harboring GRF genes from lettuce, rice and Arabidopsis, which indicated that these LsaGRFs had putative orthologous genes in both rice and Arabidopsis. In group II and IV, there were just LsaGRFs and AtGRFs, but no GRF genes from rice, while there was no GRF gene from Arabidopsis in group VI. Group III contained only two LsaGRFs, LsaGRF4 and LsaGRF9, but no AtGRFs or OsGRFs, while group V contained three OsGRFs and one AtGRF gene, but no LsaGRF gene. These results probably suggested that the orthologous genes of some LsaGRFs had species specificity and some OsGRFs and AtGRFs had no ortholog in lettuce. Among the nine AtGRF genes, six GRF genes, AtGRF1, AtGRF2, AtGRF3, AtGRF4, AtGRF5 and AtGRF9, were proved to function in leaf development [3, 4, 12]. Interestingly, the role of AtGRF5 could not be taken over by other members of AtGRFs, though there were partly overlapping functions between AtGRFs [3, 12]. Therefore, we chose the putative homolog gene of AtGRF5, LsaGRF5, based on the evolutionary relationship derived from phylogenic tree for further functional analysis (Fig. 3).
Phylogenetic tree was constructed for 15 Lactuca sativa, 9 Arabidopsis thaliana and 12 Oryza sativa GRF proteins. There were 7 phylogenetic clusters designated as I–VII. LsaGRFs were written with red fronts. The cluster III that contained GRFs only from lettuce was highlighted with green and the cluster V without LsaGRFs was shaded in red. The LsaGRF5, which was selected for further investigation, were highlighted in yellow. The scale bar represents 0.1 amino acid changes per site.
Functional characterization of LsaGRF5
To characterize the putative functions of LsaGRF5, the expression profile of LsaGRF5 was detected. From the quantitative real-time PCR (qRT-PCR) results, the expression levels of LsaGRF5 in roots and leaves were relatively low, while those in the bud were significantly high (Fig. 4A). However, in mature flowers with mature pollens and pistils, the expression level of LsaGRF5 was much lower comparing that in the bud (Fig. 4A), indicating that LsaGRF5 probably function in flower development. Meanwhile, we also detected the expressions of Lsa-miR396a, putatively regulating the expression of GRF genes, in these tissues. We found that Lsa-miR396a were relatively highly expressed in stem, cotyledon and mature flower, while significantly much lower in buds (Fig. 4A). The tissues with high Lsa-miR396a expression, e.g. mature flowers and stems, showed relatively low expression of LsaGRF5. And vice versa the tissues with high LsaGRF5 expression, e.g. buds, showed relatively low expression of Lsa-miR396a., indicating that LsaGRF5 might be regulated by Lsa-miR396a.
The LsaGRF proteins are putative TFs. We chose the LsaGRF5 and performed its subcellular location observation and transactivation assay. We isolated the protoplast cell of lettuce and transformed the vector containing 35S:LsaGRF5-GFP and 35S:Ghd7-mCherry, which was reported to locate in nuclear, into lettuce protoplast cells. The empty vector was used for control. As shown in Fig. 4B, green fluorescence of GFP and red fluorescence of mCherry were totally overlapped in the protoplast cell transformed by 35S:LsaGRF5-GFP, indicating that the LsaGRF5-GFP and Ghd7-mCherry have the same nuclear localization. While the protoplast cell transformed by empty vector exhibited ubiquitous green fluorescence, excepting the overlapped region with the red nuclear fluorescence of Ghd7-mCherry (Fig. 4B). Therefore, LsaGRF5 located in the nucleus.
To identify which part of the LsaGRF5 protein had the transcriptional activity, we divided the LsaGRF5 into two parts based on the conserved protein domains. One part is the N-terminal of LsaGRF5, GRF51 − 154, containing QLQ and WRC domains, and the other, GRF5155 − 317 (Fig. 4C). Full-length and two partial LsaGRF5s were constructed into yeast expression vector, pGBD-T7. The empty vector pGBD-T7 and pGBD-T7-OsMYB103L which was proved to have transcriptional activity were designed as the negative and positive control, respectively [37]. These recombinant plasmids were transformed into yeast strain Y2HGold. They showed similar growth states without Tryptophan (Trp) under different diluted concentration (Fig. 4D), indicating that the recombinant plasmids were indeed transformed into the yeast cells and the transformation made few influences on the yeast growth. The yeast cells expressed full-length GRF5 and GRF5155 − 317 could grow with AbA (Aureobasidin A) and turn blue with X-α-galactoside, which were the same as the positive control (Fig. 4D). These results suggested that the C-terminal contributed to the transcriptional activity of LsaGRF5, while the N-terminal containing QLQ and WRC domains did not.
LsaGRF5 is a miR396a target gene in lettuce
The GRF gene family is known as the target of miR396 [21, 22]. To verify this in lettuce, we firstly predicted the complementarity between Lsa-miR396 and LsaGRFs. Lsa-miR396a shared nearly perfect complementarity with 14 LsaGRFs except LsaGRF9 (Fig. S3). The free energies of duplex structures were all lower than − 30 kcal/mol except for LsaGRF9 (-30.6 kcal/mol) (Fig. S3). It means that all LsaGRF genes except for LsaGRF9 probably were the targets of Lsa-miR396a. We chose LsaGRF5 for further verification, and performed the 5’ RNA ligase-mediated (RLM) rapid amplification of cDNA ends (RACE) assay. The results showed that the 1 ~ 10 bp of the target sequence in LsaGRF5 did not exist in the sequencing results, which means the transcript of LsaGRF5 was cleaved at base 10 of the miR396 target site (Fig. 5). Therefore, LsaGRF5 was probably the target of Lsa-miR396a. Function analysis in vivo could further clarify the regulatory relationship between LsaGRF5 and Lsa-miR396.
The phenotypes of LsaGRF5 and Lsa-miR396 overexpression lines
To investigate the function of LsaGRF5, we constructed the overexpression lines in lettuce. LsaGRF5 driven by CaMV 35S promoter was transformed into the lettuce cultivar of Romaine type ‘YIDALI’ (YDL). Eleven independent transgenic lines were obtained. Transformation verification was carried out through a pair of primers located on 35S promoter and LsaGRF5, respectively. The results showed that five out of eleven were positive transgenic lines with the same band as the positive control (Fig. 6A). qRT-PCR assay revealed that the expression of LsaGRF5 increased 5 ~ 15 folds in these five lines (Fig. 6B). Two lines, LsaGRF5-OE5 and LsaGRF5-OE11, with high LsaGRF5 expressions were used for further phenotypic analysis. The leaves of LsaGRF5-OE5 and LsaGRF5-OE11 were significantly bigger than these of YDL transformed by empty vector (Fig. 6C). The bigger leaves also existed in LsaGRF5 overexpressed in Arabidopsis (Fig. S4). From these results, LsaGRF5 could enhance the leaf growth, and this function was conserved in Arabidopsis.
To figure out the functional relevance between Lsa-miR396a and LsaGRF5, we also overexpressed Lsa-MIR396a in YDL. Three positive transgenic lines were obtained by PCR of genomic DNA (Fig. 6A). The transcriptional levels of Lsa-miR396a were about 8 ~ 15 times higher in overexpression lines compared with negative control (Fig. 6B). Two overexpression lines with high expression level of Lsa-miR396a showed smaller leaves, the opposite phenotype of LsaGRF5-OE (Fig. 6C). We detected the expression level of LsaGRF5 in Lsa-miR396-OE lines, and the results revealed that they were suppressed in Lsa-miR396a-OE lines (Fig. 6D). These results suggested that LsaGRF5 was an enhancer factor in leaf growth, while Lsa-miR396a was a repressor, and Lsa-miR396a regulated LsaGRF5 in functioning.