Roles of LOFSEP genes in barley inflorescence development

Abstract

The grasses contain many of the world’s important staple crop species, such as rice, maize, wheat and barley. While the spikelets and florets of grass species are similar, variation in their arrangement leads to diverse inflorescence morphologies. The regulation of inflorescence development in grasses is an important component in reproduction and yield and thus of great interest for fundamental biology as well as crop improvement. The MADS-box gene family is of central importance in inflorescence development, as described in the ABCDE model of floral organogenesis. Within the MADS-box genes, the SEPALLATA, or E-class genes, can be divided into a SEP3 and LOFSEP subclade. In grasses the LOFSEP genes have adopted additional roles outside of the ABCDE model in inflorescence architecture. Based on expression data, and mutant phenotypes where available, it is likely that the LOFSEP genes MADS1, MADS5 and MADS34 perform the E-class function for the lemma and palea directly, and indirectly influence the development of the inner floral organs through other MADS-box genes. MADS34 is expressed the earliest in inflorescence development and plays a role in inflorescence branching. While MADS1 and MADS34 seem to act oppositely in spikelet initiation, they appear to act cooperatively in lemma development. To further the understanding of the role of MADS-box genes in grass inflorescence development, different grass species have to be compared. First 34 MIKCc MADS-box genes were identified in barley and found to be very conserved based on comparison to homologs in other grasses and the low numbers of SNPs. To broaden available data for grasses and get specific data for barley a detailed expression profiling experiment for all MIKCc MADS-box genes was performed in barley. The ABCDE-class genes were generally found to be expressed at the time and place predicted by applying the ABCDE model to barley, and collected into co-expression sets related to the sequential formation of the floral organs, confirming the general applicability of the model for barley. One of the core tenets of the eudicot ABCDE model is the antagonistic nature of A-class and C-class genes, which supress each other’s expression, however expression of A-class and C-class genes overlapped significantly in barley florets. This marked deviance from the classic ABCDE model may be indicative that a different approach to the clear divide between outer and inner floral organs has evolved in grasses, likely involving HvMADS32, a MADS-box class unique to grasses. To focus on the functional analysis of the LOFSEP genes in barley, knockout mutants for HvMADS1, HvMADS5 and HvMADS34 were generated using a CRISPR/Cas9 system optimised for plants. Homozygous mutants were detected in T0 and confirmed in T1 and T2 generations, while no off-target mutations were found. Editing rates of over 90% were observed, indicating the CRISPR/Cas9 system used here is very effective in barley. No inflorescence phenotype was observed for hvmads5 and hvmads34, indicating their functions are likely to be covered completely by other genes acting redundantly. Double and triple barley LOFSEP mutants will likely uncover inflorescence phenotypes as observed in rice. In contrast the hvmads1 mutant has shorter awns, reduced fertility and grain weight and additional tiller formation. These mutants behaved the same in the spring barley Golden Promise and in WI4330, a barley cultivar more recalcitrant to transformation. The additional tillers do not result in higher yield, because fertility and grain weight are reduced. Reduced grain weight could be the result of the smaller photosynthetic contribution from the shorter awns and competition for plant resources by additional tillers during the grain filling phase. Under high ambient temperature the hvmads1 mutant has the additional phenotype of a branching inflorescence. The branch-like structures appear at the location of the central spikelets and show that in barley the spikelet is likely to be the first lateral meristem of a transient branch meristem, not a terminal spikelet. The hvmads1 inflorescence morphology adopts characteristics resembling the rice and maize panicle, and the likely branched inflorescence shape for the last common ancestor barley and rice. Expression analysis was performed by RNAseq of the early inflorescence at two developmental stages, grown at low and high ambient temperature and comparing the hvmads1 mutant with the wild type. Expression of several genes, including MADS-box genes, indicated that floret development was delayed in the panicle-like inflorescence of the mutant grown at high ambient temperature. Increased expression of heat shock factors suggests a possible sensitisation to heat and additionally auxin signalling may be slightly altered. Surprisingly HvMADS34 expression is not significantly increased with the branching phenotype, although the peak expression of OsMADS34 is in the branch meristems of rice. HvODDSOC2, a MADS-box genes from a clade unique to grasses and previously associated with outgrowth of lateral tissue in the barley inflorescence is correlated with the branching phenotype and may be of regulatory importance. The genes found to be correlated to the branching phenotype are likely to be predominantly consequences of a developmental change. This leaves the important question of how the combination of the hvmads1 mutant and high ambient temperature triggers the branching inflorescence morphology. To address this question promoter affinity studies and ChIP-seq experiments are suggested. The redundancy among the LOFSEP genes in barley can partially be explained by mechanisms such as stabilisation by additional individual functions and stabilisation by developmental error. However, this is not sufficient for HvMADS34, which is expressed alone at an early stage of development, where no related genes can provide redundancy, while the hvmads34 mutant has no visible inflorescence phenotype. Early expression of HvMADS34 may be vestigial from a time when an ancestor had a branching inflorescence. When combined these results provide the basics of a generic toolbox for the adaptation of grass inflorescence morphology. Further research building on the branching barley presented here may lead to a higher yielding barley ‘panicle’, which may be translatable to the closely related wheat. Additionally, more research into the adaptations of the ABCDE model in grasses may provide both evolutionary insights and likely additions of complexes like the ‘floral’ quartets beyond the floral organs themselves, strengthening a grass specific model.