APETALA2 control of barley internode elongation

ABSTRACT Many plants dramatically elongate their stems during flowering, yet how this response is coordinated with the reproductive phase is unclear. We demonstrate that microRNA (miRNA) control of APETALA2 (AP2) is required for rapid, complete elongation of stem internodes in barley, especially of the final ‘peduncle’ internode directly underneath the inflorescence. Disrupted miR172 targeting of AP2 in the Zeo1.b barley mutant caused lower mitotic activity, delayed growth dynamics and premature lignification in the peduncle leading to fewer and shorter cells. Stage- and tissue-specific comparative transcriptomics between Zeo1.b and its parent cultivar showed reduced expression of proliferation-associated genes, ectopic expression of maturation-related genes and persistent, elevated expression of genes associated with jasmonate and stress responses. We further show that applying methyl jasmonate (MeJA) phenocopied the stem elongation of Zeo1.b, and that Zeo1.b itself was hypersensitive to inhibition by MeJA but less responsive to promotion by gibberellin. Taken together, we propose that miR172-mediated restriction of AP2 may modulate the jasmonate pathway to facilitate gibberellin-promoted stem growth during flowering.


Cloning, transformation and visulisation of GFP-HvAP2
The HvAP2 CDS was cloned into pENTR1A flanked by recombination sites (attL and attR). HvAP2 was fused to GFP by Gateway mediated recombination into the destination vector pK7WGF2. The GFP-HvAP2 construct was confirmed by sequencing and transformed into Agrobacterium tumefaciens (Agrobacterium) strain GV3101 by electroporation. Colony PCR confirmed the transformation of GFP-HvAP2 in independent colonies.
Agrobacterium containing GFP-HvAP2 as well as a positive control line containing EGFP-STICP 14.2 (Stam et al., 2013), and a negative control line without vector were infiltrated into N. benth leaves. Prior to infiltration, Agrobacterium lines were grown overnight in liquid broth culture, pelleted and then diluted in infiltration buffer (OD = 0.1) before syringe infiltration into N.benth leaves. Approximately 0.25 ml of Agrobacterium suspension was injected into four points on each abaxial leaf surface. After 48 hours further growing time, leaves were imaged with a 450-490 nm excitation filter over a Mercury-vapor lamp. Bright field images were also acquired for the same location as fluorescence images.

Assigning GO terms to sequences from the barley 61k chip
Peptide sequences for Arabidopsis were obtained from The Arabidopsis Information Resource (TAIR) at https://www.arabidopsis.org/ (Lamesch, et al., 2012) and peptide sequences for rice were obtained from the Rice Annotation Project Database (RAP-DB) at http://rapdb.dna.affrc.go.jp/  and from the Michigan State University (MSU) Rice Genome Annotation Project database at http://rice.plantbiology.msu.edu/ . Both sources were chosen for rice as later mapping of identifiers to Gene Ontology (GO) terms using g:Profiler (Reimand, et al., 2016) required RAP IDs rather than MSU IDs. BLASTX searching of the barley sequences against the rice and Arabidopsis peptides was carried out to identify top-ranked hits for each, but with otherwise default parameters (E=10) to allow for downstream filtering of results. Matches were filtered based on percentage identity over percentage of the query (barley) sequence aligned: 50% identity over 50% query sequence for rice matches and 40% identity over 50% query sequence for Arabidopsis matches. In cases where no RAP-DB match was identified using BLAST but an MSU match was identified, then those MSU IDs could be directly converted into RAP-IDs using the RAP-DB ID converter. RAP-DB IDs for those mapping unambiguously were then included in the set. Further RAP-DB and TAIR matches were also obtained as part of a pilot GO Slim analysis carried out at the start of this study. Whilst many of these did not pass the criteria of percentage identity / percentage query length, those matches that had BLAST E-value scores of 1e-5 or lower were also added to the sets for g:Profiler analysis. Accession numbers from RAP-DB and TAIR were then supplied to g:Profiler at http://biit.cs.ut.ee/gprofiler/ (Reimand, et al., 2016) to identify the sets of GO terms associated with each. Default parameters were used, but the option to return only significant terms was deselected as this would cause g:Profiler to attempt an enrichment analysis, but at this stage of the analysis all matching terms were required. GO terms arising from the RAP-DB ID mapping were chosen in preference to those arising from TAIR ID mapping since barley is more closely related to rice than it is to Arabidopsis, meaning that in situations where GO terms were returned for a barley sequence from both RAP-DB and TAIR mappings, no attempt was made to merge the two lists so as to avoid possible conflicting terms. Mappings arising from TAIR would only be used where there was no corresponding information from RAP-DB. The resulting set of GO terms can then be used for downstream enrichment analysis.
Of the 61,487 barley sequences 58,055 matched to Arabidopsis peptides from the TAIR set (E=10), with 24,639 matching with at least 40% identity over at least 50% of the query sequence length; 57,666 matched to rice peptides from the MSU set (E=10) and, of these 29,798 matched with at least 50% identity over at least 50% of the query (barley) sequence length; from the RAP-DB set 58,716 sequences were matched (E=10) with 27,798 matching with at least 50% identity over 50% query sequence length. Additional direct mapping of MSU IDs and inclusion of matches from the pilot GO Slim analysis resulted in final sets of 31,134 barley sequences with corresponding RAP-DB IDs and 26,039 barley sequences with corresponding TAIR IDs. These were then supplied to g:Profiler to identify the GO terms associated with each.
Not all RAP-DB or TAIR IDs returned lists of GO terms. GO terms arising from RAP-DB mappings were assigned to 24,077 barley sequences and terms arising from TAIR mappings were assigned to 24,295 barley sequences. Merging these lists produced a final set of 29,787 barley sequences with 8,222 associated GO terms arising from either RAP-DB or TAIR mappings and with a maximum term depth level of 15. Figure S19, generated using Venny (Oliveros, 2007(Oliveros, -2015 shows the distribution of barley genes with GO terms arising from RAP-DB or TAIR mappings.

Metabolic Pathway Reconstruction along the internode
To relate the probe sequences of the barley microarray to the latest HORVU gene models, the probes were aligned to the full set of HORVU transcripts sequences using the blastn command line tool (version blast+ 2.28, Camacho et al. 2009, Altschul et al. 1990). The output from this was filtered to retain only those hits with 100% query coverage and an identity value of >=95%, leaving a single HSP per query. Duplicate combinations of query x gene were then removed to eliminate multiple transcripts from the same gene that had been hit by a given query. Of the 61487 microarray probes, 42466 were associated with a HORVU transcript ID. The 36995 unique HORVU transcript IDs had 30479 unique HORVU gene IDs, which were used for the analysis of the hormone pathway genes. BarleyCyc 6.0 database (https://www.plantcyc.org/databases/barleycyc/6.0) was used to identify genes encoding enzymes from the gibberellin, brassinosteroid and jasmonic acid pathways along with their corresponding HORVu accession number. These HORVus were then filtered for those expressed in the Bowman peduncle.

Identification of potential APETALA2 binding sites
The 500bp promoters of differentially expressed genes in Zeo1.b compared to Bowman or metabolic pathway genes were retrieved from Ensembl (Frankish et al. 2017) Plants Genes 42, Hordeum vulgare genes (IBSC v2) using the Biomart tool (https://plants.ensembl.org/biomart/martview/a4076bb018d3718a542b11cf7d46d092) via HORVu accession number. Where multiple barley accessions corresponded to HORVu accessions, all possible barley accessions were considered. These sequences were analysed using the PlantTFDB 4.0 (Jin et al. 2017) for potential binding sites of known barley transcription factors Potential APETALA2 binding sites were identified in these sequences using the Find Individual Motif Occurrences within the MEME Suite 5.2 using a false discovery cut off of q > 0.05. These motifs were then filtered for those containing consensus AP2-binding motif 'AACAAA' or 'TTTGTT' identified in Dinh et al (2012).  Debenardti JM, Lin H, Chuck G, Faris JD and J Dubcovsky (2017) microRNA172 plays a crucial role in wheat spike morphogenesis and grain threshability. Development. 144(11): 1966-1975Development: doi:10.1242