An increase in crop yield is essential to reassure food security to meet the accelerating global demand. Several genetic modifications can increase organ size, which in turn might boost crop yield. Still, only in a few cases their performance has been evaluated under stress conditions. MicroRNA miR396 repress the expression of GROWTH-REGULATING FACTOR (GRF) genes that codes for transcription factors that promote organ growth. Here, we show that both Arabidopsis thaliana At-GRF2 and At-GRF3 genes resistant to miR396 activity (rGRF2 and rGRF3) increased organ size, but only rGRF3 can produce this effect without causing morphological defects. Furthermore, introduction of At-rGRF3 in Brassica oleracea can increase organ size, and when At-rGRF3 homologs from soybean and rice are introduced in Arabidopsis, leaf size is also increased. This suggests that regulation of GRF3 activity by miR396 is important for organ growth in a broad range of species. Plants harboring rGRF3 have larger leaves also under drought stress, a condition that stimulates miR396 accumulation. These plants also showed an increase in the resistance to virulent bacteria, suggesting that the size increment promoted by rGRF3 occurs without an obvious cost on plant defenses. Our findings indicate that rGRF3 can increase plant organ size under both normal and stress conditions and is a valuable tool for biotechnological applications.
The growth of plant organs is tightly controlled by their developmental program and the interaction with the environment. Leaves initiate as rod-like structures protruding from the shoot apical meristem, pass through different developmental stages and become a flat organ specialized in photosynthesis1,2. Multiple regulatory gene networks are known to participate in the morphogenesis of a leaf, although their precise role and interactions are unknown in many cases.
Crop yield is a highly complex trait influenced by both external and internal factors. Intrinsic Yield Genes (IYG) have been defined as those genes that produce larger organs, such as leaves, roots or seeds, when mutated or ectopically expressed3. In this sense, the precise modification of IYG might increase crop yield and therefore they constitute a potential source of biotechnological applications. The GROWTH-REGULATING FACTORs (GRFs) genes code for a family of plant-specific transcription factors characterized by the presence of the WRC and QLQ protein domains, which have been involved in DNA-binding and protein-protein interaction, respectively4,5,6,7,8,9. In Arabidopsis thaliana, there are nine GRF coding genes (GRF1-9). Seven out of them harbor a target site for microRNA miR396. At early stages of leaf development, miR396 is expressed in the distal part of the leaf, restricting the expression of the GRFs to the proximal part, which is coincidental with the proliferative region of the organ10,11,12,13. At later stages of leaf development and after the stop of cell proliferation, miR396 is expressed throughout the organ repressing GRF expression in maturing organs10,11,12.
The miR396-GRF module is present in a broad range of plants including angiosperms and gymnosperms11,14,15. In certain cases, the ectopic expression of the GRFs is sufficient to increase leaf size. In Arabidopsis, overexpression from the 35S promoter of At-GRF516,17, Brassica napus Bn-GRF218 and Brassica rapa Br-GRF819 promotes a moderate increase of leaf size.
Modified GRFs have been generated with synonymous mutations in the miRNA target site to avoid the post-transcriptional repression mediated by miR39612,20. The observation that plants harboring these miR396-resistant versions of At-rGRF2 (rGRF2) or At-rGRF3 (rGRF3) have larger leaves with respect to wild type plants, indicates that miR396 normally restricts organ size through the repression of the genes coding for GRF transcription factors. In good agreement with these results, plants overexpressing miR39612,21 or single grf517 and multiple grf1 grf2 grf35 knock outs have smaller organs.
However, increased levels of the GRFs not always results in larger organs in Arabidopsis, as overexpression of Oryza sativa Os-GRF1 caused pleiotropic defects, including curled leaves, delayed flowering and defects in carpel development4. Results in crops have also been variable. Overexpression of Zm-rGRF1 increased maize leaf size, while it also caused additional detrimental phenotypes such as large macrohairs covering the glumes and the ear rachis that reduced fertility15, while overexpression of Zm-GRF10, which lacks a transactivation domain, reduced maize leaf size22. Furthermore, high levels of At-GRF7 and At-GRF9 caused no major increase of Arabidopsis leaf size23,24,25. Interestingly, At-GRF7 has been implicated in the response of plants to osmotic stress23, while At-GRF9 has been claimed to be a growth repressor26. Furthermore, in certain organs and conditions, the GRFs can affect both cell number and size27,28.
The capacity of certain GRFs to increase leaf size per se suggests that they can act as IYG increasing plant organ size, and therefore they could be a valuable tool for biotechnological applications. Still, not all the GRFs have a positive impact on organ size, and some of them have even deleterious effects. Here, we characterized different members of the GRF family in Arabidopsis and found that the At-GRF3 gene decoupled from miR396 regulation consistently increase organ size in Arabidopsis thaliana, an ability that likely depends on the protein sequence of the transcription factor. We also show that a miR396-resistant GRF3 can increase leaf size, root length and seed size in transgenic Brassica oleracea. Furthermore, we found that plants expressing the rGRF3 transgene still have an increase in leaf size under mild drought stress and show enhanced resistance to certain plant pathogens. We conclude that GRF transcription factors similar to At-GRF3 can be used to increase plant organ size in Brassicaceae species without an obvious deleterious impact in plant fitness.
Broad distribution of GRF regulation by miR396 in angiosperms
There are nine GRFs in Arabidopsis (Fig. 1a,b), and seven of them have a target site for miR396 located in the region that codes for the carboxyl end of the WRC domain (Fig. 1b,c). The miRNA target site is identical in the different GRFs with an exception at position eight where there is a C in GRFs 1-4, an A in GRFs 7-8, and a U in GRF9 (Fig. 1c). This variable base is located in a bulged position with respect to the miRNA so that the interaction with the miRNA is quite similar for all the transcription factors, however, the encoded amino acid sequence in the carboxyl terminal side of WRC domain varied (Fig. 1c).
Interestingly, the Arabidopsis GRF gene structures revealed differences in the exon-intron organization and in the distribution of the protein domains that define the family (Fig. 1b). We analyzed the occurrence of the miRNA target site in GRFs of different angiosperms (Fig. 1d and Supplementary Table S1). In many species, such as Populus trichocarpa (poplar), Glycine max (soybean) and Medicago truncatula, all the GRFs have a miR396-binding site (Fig. 1d), while in others like Arabidopsis and rice a few genes lack this sequence. Interestingly, there is a sequence remotely resembling a miRNA target site in Arabidopsis GRF5, one of the two GRFs that lack of miR396 regulation in this species (Fig. 1c). These findings indicate a broad distribution of GRF regulation by miR396 and suggest that this might be the default state of these transcription factors in angiosperms.
Differential GRF activity among family members
To study the loss of function of MIR396, we prepared and characterized plants expressing a miR396 target mimic (MIM396) that consists of a noncoding RNA that binds to the miRNA and blocks its function and/or promotes its degradation29,30. We prepared a MIM396 harboring 9 sites that can act as a sponge for miR396 (9X-MIM396) (Fig. 2a). First, we analyzed ca. 20 primary transgenic plants transformed with the 9X-MIM396. We observed that this construct consistently caused an increase in leaf area that reached 70% in the most extreme cases (Fig. 2b,c). The area of the first leaf was increased when independent homozygous transgenic lines were analyzed (Fig. 2f). Further molecular analysis of 9X-MIM396 #3 line revealed a reduction of 70% in miR396 expression (Fig. 2d) and a 40% increase of GRF2 and GRF3 transcript levels (Fig. 2e), in agreement with the observed phenotypes (Fig. 2b,c and f).
Previous results have shown that the ectopic expression of GRF caused different effects, e.g., while in Arabidopsis rGRF2 or rGRF3 expressed from their own promoter or 35S:GRF5 caused an increase of leaf size12,16,20, overexpression of GRF7 and GRF9 did not23,24,25. Furthermore, overexpression of certain GRFs from other species might also inhibit growth4. The combined observation that most of the GRFs are regulated by miR396 in angiosperms and that the developmental phenotype of 9X-MIM396 was an increase of leaf size, showed that the miR396-GRF module controls organ size in vivo. These results prompted us to focus on rGRF2 and rGRF3 in more detail, especially in the light of the potential biotechnological application to increase the size of leaves and other organs.
First, we analyzed approximately 30 primary transgenic plants harboring rGRF2 or rGRF3 (Fig. 3a,b) expressed from their own promoters. Although both constructs were able to increase leaf area, the effect caused by rGRF3 was significantly higher than the effect obtained with rGRF2. We then selected rGRF2 and rGRF3 homozygous T3 transgenic lines and determined the GRFs transcript levels (Fig. 3d and Supplementary Fig. 1). We found that a 2-fold increase in GRF3 expression was sufficient to change leaf area in 70%, while an increase of more than 25-fold in GRF2 transcripts is required to cause a similar increase in organ size (Fig. 3d,e). In both cases, the increase in organ size was due to a similar raise in the number of cells, with no appreciable change in cell size (Fig. 3e,f and g).
Further analysis of rGRF2 plants revealed that the leaves had morphological defects, including long and twisted petioles with downward curled leaves (Fig. 3c). On the other hand, no obvious morphological changes were observed in rGRF3 plants besides the increase in leaf area. We considered that the high levels of GRF2 required to increase leaf size, caused also additional morphological defects in plant development. To confirm that the promoter was not the cause of the observed differences between rGRF2 and rGRF3, we expressed both rGRF2 and rGRF3 from the GRF3 promoter and observed that pGRF3:rGRF3 caused a larger increase in leaf area than pGRF3:rGRF2 (Fig. 3h). Therefore, both expression levels and GRF protein sequences should be considered to efficiently enhance plant organ size. As shown here, although both rGRFs can be used to increase organ size, rGRF3 is a more active and specific enhancer of plant organ size.
Robust increase in leaf size by expression of Arabidopsis GRF3-like sequences decoupled of miR396 regulation
Given the potential use of rGRF3 as a tool to promote plant organ size, we decided to express Arabidopsis GRF3-like sequences from selected crops in Arabidopsis. We analyzed the databases for GRF transcription factor sequences from rice6 and soybean8 and selected those with the highest similarities to At-GRF3 (Supplementary Table S2). Then, we expressed soybean and rice Arabidopsis GRF3-like coding sequences in Arabidopsis under the At-GRF3 promoter. As both genes have a miR396 target site, we introduced synonymous mutations to avoid the recognition by the small RNA. We observed an increase in leaf size caused by both soybean and rice rGRF3-like transcription factors (Fig. 4a,b).
To evaluate the effect of rGRF3 on leaf size in another species, we expressed the transcription factor in Brassica oleracea transgenic plants (Fig. 4e). Analysis of two independent transgenic lines revealed a significant increase in leaf area of 20 and 32% (Fig. 4c,d). Further characterization of the transgenic leaves revealed that the increase in leaf size was caused by a higher number of cells and not by an effect in cell size (Fig. 4c). The evaluation of root growth in B. oleracea rGRF3 #10 plants showed that the primary root of these plants elongated at a higher rate than control plants (Fig. 4h,i). Furthermore, these plants also had an increase of 10% in seed size (Fig. 4f,g). Taken together, these results show that AtGRF3-like sequences from various species when decoupled from miR396 regulation robustly increase the size of several organs, including leaves, roots and seeds.
Role of the miR396-GRF node under stress conditions
Drought stress is a fairly complex situation triggering different response pathways according to the magnitude of the stress and the developmental stage of the plant31. In particular, water limiting conditions repress cell proliferation and expansion in developing organs and/or induce a complex array of tolerance and survival responses in mature organs31,32,33. Under field conditions limited water availability usually reduces plant growth, biomass accumulation and, therefore, seed yield34. We evaluated the response of plants to moderated drought stress consisting of a 55% reduction in soil water content using the automated phenotyping platform WIWAM34. Under these conditions, control plants transformed with the empty vector had a reduction in rosette area of 40% (Fig. 5a) and a reduction of 35% in leaf 1 area (Fig. 5f).
We looked at the expression of MIR396B using the transcriptional reporter MIR396B:GUS that allowed us to monitor the expression of the most abundant miR396-coding gene in leaves. At 11 days after sowing (DAS) in well wattered pots MIR396B is expressed at low levels in the proximal end of leaf 3, that is still in a proliferative state (Fig. 5b). The highest expression of MIR396B:GUS is found in the distal zone of the leaf in which cells exited the cell cycle. When plants were subjected to mild drought stress leaf size was reduced and MIR396B:GUS expression was detected in a larger distal leaf area with a more intense staining (Fig. 5b,c). Accordingly, small RNA blots revealed an increase of 70% in the mature miR396 levels in developing leaves from stressed plants (Fig. 5d). In contrast, At-GRF3 transcript levels decrease by approximately 60% (Fig. 5e). Altogehter, these results suggest that the miR396 network responds to drougth stress. We then evaluated the role of the miR396-GRF3 regulatory node during drought stress. To do this, we first analyzed the response to mild drought stress of 35S:miR396 plants, which have small leaves due to the overexpression of the miRNA and the repression of the GRFs12 (Fig. 5f). When drougth stress is imposed to these plants, no further reduction in organ size could be observed (Fig. 5f), indicating that transgenic miR396 overexpression masks the effects derived from the endogenous induced MIR396B.
When the stress was applied to rGRF3 lines (Fig. 6), leaf size was reduced with respect to rGRF3 plants grown under control conditions. However, at the end of the experiment, 19 days after sowing, rGRF3 leaves were significantly larger than leaves from the control plants growing under mild drought (Fig. 6a,b). Notably, we did not find differences in the plastochron length caused by higher GRF3 expression levels or the mild drought conditions assayed. Therefore, the differences in rosette size observed are mainly due to changes in leaf organ size rather than in changes in rosette architechture (Supplementary Fig. 2).
Having tested mild drought as an abiotic stress, we turned to pathogen infections to evaluate the susceptibility of rGRF3 plants to biotic stress. This is of particular interest, as diverting energy to organ growth might compromise plant defenses to pathogen infections. To do this, adult plants were infiltrated with the hemibiotrophic virulent bacteria Pseudomonas syringae pv. tomato DC3000 (Pst) and pathogen content was quantified at 2, 3 and 5 days post infection (dpi). Interestingly, Pst titers were lower in rGRF3 than in wild type plants, and these differences were maintained also at late stages of infection (9 times difference at 5 dpi). This indicated a positive effect of GRF3 on the immune pathways responsible to counteract pathogen proliferation (Supplementary Fig. 3). Altogether, these data indicate that rGRF3 can be a valuable tool that might ameliorate the effects of drought stress on organ growth and increase leaf size under both normal and stressfull conditions.
Here, we show that At-GRF3-like sequences from various species like soybean and rice can systematically increase leaf, seed and root size in Brassicaceae species, i.e. Arabidopsis thaliana and Brassica oleracea, when decoupled from miR396 regulation.
Previous results using GRFs to boost plant growth and yield have generated inconsistent results. On the one hand, MIM396 mimics increased tomato organ size35, and natural occurring alleles of Os-GRF4, a miR396-targeted GRF from rice with homology to At-GRF3, increases leaf size, grain weight and rice yield27,28,36,37. On the other hand, overexpression of Os-GRF1 in Arabidopsis4 or Zm-GRF1 in maize15 lead to developmental defects that affected plant fertility and seed production. Our results suggest that it is relevant to choose the proper GRF coding sequence for genetic engineering in the context of an adequate expression system in order to increase plant organ size and eventually crop yield. Also, they suggest that the controlled expression of At-GRF3-like genes have the highest chance to boost organ growth with no associated developmental defects.
Crop plants under field conditions often face drought stress that reduce plant growth and yield. An integral part of drought responses in plants is growth repression to limit shoot size, reduce evaporation surface and match sink-tissue demands to the scarce resources available due to stomatal and metabolic restrictions31. This occurs due to premature transitions from cell proliferation to cell expansion combined with a reduced magnitude of cell expansion33,38.
In leaves from dicot and monocot species, miR396 is expressed in low levels in meristems and induced at later developmental stages to promote the transition from cell proliferation to cell elongation and maturation11,12,39,40. We found that miR396 was prematurely induced while its target At-GRF3 was repressed under mild drought. These results, together with the finding that plants expressing a miR396-resistant At-GRF3 were more tolerant to the stress, indicate that miR396 might be part of the genetic mechanisms that limit leaf size during water limitation.
Microarray and RNA-seq studies have shown that MIR396 can respond to other abiotic stresses, such as aluminum41, cadmium42, salt43 and alkali stress44 (Ding et al., 2009; Gao et al., 2010). It has been recently shown, both in Arabidopsis and maize, that miR396 levels are prematurely induced by UV-B in developing leaves. This induction causes the repression of the GRFs expression and a decrease of cell proliferation, leading to smaller leaves with fewer cells45,46.
Plant expressing GRF3 decoupled of miRNA regulation reduced their size upon mild drought stress treatment. Still, rGRF3 leaves were significantly larger than those from the control plants grown under drought stress. Therefore, At-rGRF3 is able to increase plant organ size under both optimal and adverse conditions.
A particular role has been established for At-GRF7 in Arabidopsis in the context of stress responses. Mutants in this gene have higher expression levels of stress-responsive genes in developing organs, including DREB2A, a master regulator of water limitation responses. At- GRF7 has been shown to act as a transcriptional repressor of DREB2A and other stress-responsive genes to avoid the detrimental effects of stress responses on cell proliferation and organ growth23.
Recently, it has been shown that MIM396 plants, with low miR396 activity, develop fast and strong immune responses and increased resistance to nectrotrophic and hemibiotrophic fungi, apparently by triggering defense priming47. Our results suggest that At-GRF3 could mediate similar responses as well. rGRF3 plants, in which miR396-mediated repression of GRF3 is avoided, presented increased resistance to virulent Pst suggesting a positive effect of this transcription factor on defenses against hemibiotrophs. Interestingly, our results indicate that rGFR3 improves immune responses at the same time that it boosts plant growth. This contrasts with many examples illustrating the costs of increasing defenses, where activation of immune pathways under non-stress conditions produces a fruitless diversion of energy that causes a negative effect on plant growth and fitness48. In turn, other plants with enhanced resistance and no fitness cost have also been described, including MIM396 itself47 and overexpressors of NPR149. Then, it is expected that rGRF3 plants may achieve a suitable balance to improve both stress response and growth under adverse conditions, however, we cannot exclude that GRF3 has different roles in these processes.
In summary, our results indicate that it might be possible to increase plant size through the genetic engineering of Arabidopsis GRF3-like transcription factors, even under stress conditions. It’s worth to mention, that the gene manipulations required are limited to only a few base pair substitutions in the properly chosen GRF, making possible to use modern gene editing technologies for the fast introduction of these traits in crop species with desired characteristics.
Plant materials and growth conditions
Arabidopsis of the Col-0 accession was used in all the experiments. See Supplementary Table S4 for a description of the plasmids used to generate the transgenic lines characterized in this study. Brassica olaracea cultivation and transformation was performed as described in (Moloney et al., 1989). For leaf analysis plants were grown on soil while for root analysis, surface-sterilized seeds were sown on solid medium containing 1X Murashige and Skoog salt mixture, 1% sucrose, 2.3 mM 2-(Nmorpholino) ethanesulfonic acid (pH 5.8) and 1% agar. In all cases plants were grown at 16 h light/8 h dark photoperiod at 23 °C and a light intensity of 100 μmol quanta m−2 s−1. Leaf and seed size and root length were measured using FIJI50. In all cases, photographs of rosettes, leaves, seeds or roots were taken next to a ruler that was used as a reference to convert pixels to the corresponding metric unit. Later, organ size or length was measured in pixels in Fiji and, finally, these values were converted to metric units using the conversion factor obtained.
RNA was extracted using TRIzol reagent (Invitrogen) and 0.5–1.0 μg of total RNA was treated with RQ1 RNase-free DNase (Promega, http://www.promega.com). Next, first-strand cDNA synthesis was carried out using SuperScript III Reverse Transcriptase (Invitrogen) with the appropriate primers. PCR reactions were performed in a Mastercycler ep realplex thermal cycler (Eppendorf, http://www.eppendorf.com) using SYBRGreen I (Roche, http://www.roche.com) to monitor double-stranded (ds)DNA synthesis. Quantitative (q)PCR of each gene was carried out in at least three biological replicates, with technical duplicates for each biological replicate. After normalizing using PP2A (PROTEIN PHOSPHATASE 2A) for Arabidopsis51 or UBC (UBIQUITIN CONJUGATING ENZYME) for Brassica oleracea52 transcript levels relative to the selected control genotype or condition was determined for each sample. Primer sequences are given in Supplementary Table S3.
MiR396 accumulation was estimated by small RNA blots performed as described previously12. Briefly, 5 μg of total RNA was resolved on 17% polyacrylamide denaturing gels containing 7 M urea. Blots were hybridized using a radioactively labelled locked nucleic acid (LNA, Exiqon, Denmark) oligonucleotide probe designed against miR396 or a probe against U6 snRNA as a loading and blotting control (5′-CTCGATTTATGCGTGTCATCCTTGC). Phosphorimaging were obtained with a Typhoon (GE) and the signal intensity of ach band were determined using the GelQuant. NET software (biochemlabsolutions.com). The values of the bands detected with the miR396 probe were first normalized to the signal obtained with the U6 snRNA probe and then relativized to the selected control genotype or condition.
To visualize GUS reporter activity, transgenic plants were subjected to GUS staining and the staining intensity along the length of the leaf was estimated as described previously53.
To obtain paradermal views of palisade cells, leaves were fixed with formalin-acetic acid-alcohol and cleared with chloral hydrate solution (200 g chloral hydrate, 20 g glycerol, and 50 ml H2O) as described17. Palisade leaf cells were observed using differential interference contrast microscopy in a Olympus BH2 microscope. The density of palisade cells per unit area was determined, and the area of the leaf blade was divided by this value to calculate the total number of palisade cells in the sub epidermal layer. To determine the cell area, 20 palisade cells were measured in each leaf.
Plants were germinated on soil-filled 96-well plates after stratification at 4 °C in the dark. Seedlings were transferred to pots at 4 days after sowing (DAS). Before transfer, the relative water content of the pots was set at 1.19 g water g−1 dry soil for the mild drought treatment while the control condition was set at 2.19 g water g−1 dry soil. Once the seedlings were transferred to these pots they were placed on the automated phenotyping platform WIWAM34. The water content of the soil was kept constant until 8 DAS, after which it was lowered daily to 1.02 g water g−1 dry soil for the mild-drought-treated plants. This phenotyping platform is designed to monitor vegetative development as images of the rosette of each plant are taken daily until 19 DAS. Leaf 1 area was measured from these images to monitor organ growth during the experiment. Also, the plastochron length was estimated by registering the time when each consecutive leaf reached at least 1 mm in length.
Susceptibility to Pseudomonas syringae pv. tomato DC3000 was determined using 4 weeks old plants. Bacterial suspensions (5.105 cfu/ml) were infiltrated into adult leaves using a syringe. Pathogen growth was determined as previously described54, using a set of six leaf-disks (4 mm2) of independent plants per sample.
Multiple sequence alignments and phylogenetic trees were obtained using the Clustal Omega55 and PHYLIP56 softwares using the neighbor joining method. Bootstrap analysis was conducted to measure node robustness using 1000 replicates.
The datasets generated during and analyzed during the current study are available from the corresponding author on request.
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
MEA is member of the Research Career of Consejo Nacional de Investigaciones Científicas (CONICET) and is supported by ANPCyT (PICT 2014-3255). FN is a technician of CONICET. MB, MFE, JMD, AR and CG were supported by fellowships from CONICET. The majority of these studies were supported by grants to JP and RR (ANPCyT).