一作解读 | 小麦VRN1、FUL2和FUL3在小穗发育和穗部形态建成过程中起关键并冗余作用
Chengxia Li(#), Huiqiong Lin(#), Andrew Chen, Meiyee Lau, Judy Jernstedt and Jorge Dubcovsky(*)
(*):jdubcovsky@ucdavis.edu. Phone: 530 752 5159
The grass family (Poaceae) has approximately 10,000 species, including important food crops such as rice, maize, sorghum, barley and wheat.The flowers of these species are organized in a unique and diagnostic structure called spikelet, which is a compact inflorescence developing within the larger inflorescence. Spikelet is the basic unit of the grass inflorescence. Grass inflorescences have been described as a progressive acquisition of different meristem identities that begins with the transition of the vegetative shoot apical meristem (SAM) to an inflorescence meristem (IM). In wheat, the transition from vegetative SAM to IM is marked by the formation of a double-ridge structure, in which the lower leaf ridges are suppressed and the upper ridges acquire spikelet meristem (SM) identity andform spikelets. The number of spikelets per spike in wheat is determined by the number of lateral meristems formed before the transition of the IM into a SM toform the terminal spikelet. The growth of each wheatspikelet is indeterminate, with each SM initiating a variable number of floralmeristems (FM). The numbers of spikelets per spike and florets per spikelet determine the maximum number of grains per spike, therefore are important components of wheat grain yield potential.
In this study, we show that wheat MADS-box genes VRN1, FUL2 and FUL3 play critical and redundant roles in spikelet and spike development, and also affect flowering time and plant height. We combined loss-of-function mutants for the two homeologs of VRN1, FUL2 and FUL3 to generate double- and triple-null mutants in tetraploid wheat. In the vrn1ful2ful3-null triple mutant, the inflorescence meristem formed a normal double-ridge structure, however, itslateral meristems then proceeded to generate vegetative tillers subtended by leaves instead of spikelets. These results suggest an essential role of these three genes in the fate of the upper spikelet ridge and the suppression of the lower leaf ridge. Inflorescence meristems of vrn1ful2ful3-null and vrn1ful2-null remained indeterminate, and single vrn1-null and ful2-null mutants showed delayed formation of the terminal spikelet and increased number of spikelets per spike. Moreover, the ful2-null mutant producedmore florets per spikelet, which together with a higher number of spikelets, resulted in a significant increase in the number of grains per spike in the field. Our results suggest that a better understanding of the mechanisms underlying wheat spikelet and spike development can inform future strategies toimprove grain yield in wheat.
Plants carrying only the ful3-nullmutation showed no significant reduction in stem length, but those carrying thevrn1-null or ful2-null mutations were 20% and 14% shorter than the control, respectively (Fig. 1A). A three-way factorial ANOVA for stem length revealed highly significant effects for all three genes and significant synergistic interactions (Fig. 1C), indicating that VRN1, FUL2 and FUL3 have redundant roles in the regulation of stem elongation, and that the effect of the individual genes is larger in the absence of the otherparalogs.
Functional redundancy among VRN1, FUL2 and FUL3 was also observed for heading time. The vrn1-null mutant headed 37.5 dlater than the control (Fig. 1D), but differences in heading time for the ful2-null, ful3-null and ful2ful3-null mutants in the presence of the strong Vrn-A1 allele were non-significant (Fig. 1E). For the vrn1ful2-null and vrn1ful2ful3-null mutants, it was not possible to determine heading times accurately because theyhad short stems and abnormal spikes that interfere with normal ear emergence. Instead, we determined the final number of leaves (Fig. 1B) and the timing of the transition between the vegetative and double-ridge stages (Fig. S3). The vrn1-null mutant had on average 14.4 leaves (59% > control, Fig. 1B), which was consistent with its later heading time (Fig. 1D). Similar leaf numbers were detected in vrn1ful2-null (14.3) and vrn1ful3-null (14.9), but the triple vrn1ful2ful3-null mutant had on average17.7 leaves (Fig. 1B), which was consistent with the 9 to 12 d delay in the transition between the vegetative SAM and the double-ridge stage relative to the vrn1-null control (Fig. S3).
Plants with individual vrn1-null,ful2-null and ful3-null mutations produced normal spikelets and flowers, but vrn1ful2-null or vrn1ful2ful3-null mutants had spike-like structures in which alllateral spikelets were replaced by leafy shoots (inflorescence tillers, Fig.2A-J). Removal of these inflorescence tillers revealed a thicker and shorter rachis with fewer internodes of variable length, but still retaining the characteristic alternating internode angles typical of a wild type rachis (Fig. 2B).
In vrn1ful2-null, approximately 70% of the central inflorescence tillers had leafy glumes, lemmasand paleas and abnormal floral organs, whereas the rest were fully vegetative. Floral abnormalities included leafy lodicules, reduced number of anthers, anthers fused to ovaries, and multiple ovaries (Fig. 2E-G). After the first modified floret, meristems from these inflorescence tillers developed two to five true leaves before transitioning again to an IM generating lateral VMs(Fig. 2E). The presence of both floral organs and leaves suggests that the originating meristem had an intermediate identity between VM and SM before transitioning to an IM. In the vrn1ful2-null double mutant the inflorescence tillers were subtended by bracts (Fig. 2C-D).
In vrn1ful2ful3-null, the lateral meristems generated inflorescence tillers that had no floralorgans, and that were subtended by leaves in the basal positions and bracts in more distal positions (Fig. 2H-J). The presence of well-developed axillary tillers in these basal inflorescence leaves (Fig. 2H, L19 and L20) marked the border of the spike-like structure, because no axillary tillers or developing buds were detected in the true leaves located below this border (Fig. 2H,L11-L18).
Scanning Electron-Microscopy (SEM) images of the early developing inflorescences in the vrn1ful2-null and vrn1ful2ful3-null mutants revealed elongated double-ridge structures similar to those in Kronos (Fig. 3 A) or vrn1-null (Fig. 3 C). Suppression of the lower leaf ridge was complete in Kronos (Fig. 3A) and in vrn1-null (Fig. 3D, red arrows), but was incomplete in vrn1ful2-null(Fig. 3B, E; yellow arrows), and even weaker in vrn1ful2ful3-null (Fig. 3C, F: green arrows). As a result of this change, inflorescence tillers were subtended by bracts in vrn1ful2-null (Fig. 2C-D) and by leaves in vrn1ful2ful3-null (Fig. 2H-I). The upper ridges (Fig. 3A-C, dots) transitioned into normal SMs in vrn1-null, with glume and lemma primordia (Fig. 3D, G), but looked like typical vegetative meristems in vrn1ful2-null and vrn1ful2ful3-null (Fig. 3E-F, H-I).
Normal wheat spikes are determinate, with the distal IM transitioning into a terminal spikelet after producing arelatively stable number of lateral meristems (Fig. 4A). In vrn1ful2-null, by contrast, the IM was indeterminate (Fig. 4B) and continued to produce lateral meristems while growing conditions were favorable and eventually died without producing anyterminal structure. In the ful2-null background, one functional copy of VRN1 inthe heterozygous state was sufficient to generate a determinate spike (Fig.S6D, ful2-null/vrn-A1-null vrn-B1), and the same was true for a single functional copy of FUL2 in a vrn1-null background (Fig. S6K, vrn1-null/ful2-A Ful2-B).
The individual vrn1-nulland ful2-null homozygous mutantsshowed a larger number of spikelets per spike than the control. This increasewas 58% in the vrn1-null mutant (P< 0.0001, Fig. 4C) and 10% in the ful2-null mutant (P =0.0014, Fig 4D). Although no significant increases in the number of spikelets per spike were detected in the ful3-null mutant (P = 0.4096, Fig. 4E), two independent transgenic lines overexpressing FUL3 (Ubi::FUL3) showed an average reduction of 1.12 spikelet per spike relative to their non-transgenic sister lines (P = 0.0132 and P < 0.0001, Fig. S8A), which indicates that FUL3 can still play a role on the timing of the transition from IM to terminal spikelet.
In addition to the higher number of spikelets per spike, the ful2-null mutant produced a higher number of florets per spikelet than the Kronos control, an effect that was not observed for vrn1-null (Fig. 2A) or ful3-null(Fig. S10A). In spite of some heterogeneity in the distribution of spikelets with extra florets among spikes, the differences between the control and the ful2-null mutants were significant at all spike positions (Fig. S10B).
Based on its positive effect on the number of florets perspikelet and spikelets per spike (and its small effect on heading time), we selected the ful2-null mutant for evaluation in a replicated field experiment. Relative to the control, the ful2-null mutant produced 20% more spikelets per spike (P = 0.0002) and 9% more grains per spikelet (P= 0.05), which resulted in a 31% increase in the number of grains per spike (P = 0.0002, Fig. 4F). Although part of the positive effect on grain yield was offset by a 19% reduction in averagekernel weight (P = 0.0012), we observed a slight net increase of 6% in total grain weight per spike (P = 0.09, Fig. 4F). This negative correlation between grain number and grain weight suggests that in this particular genotype by environment combination grain yield was more limited by the “source” (produced and transported starch) than by the “sink” (number and size of grains).