{"paper_id":"22ee34c2-d987-45df-9aa0-beda63674f78","body_text":"1 \nRegulation of spikelet number during wheat spike development \nChengxia Li1, 2, Kun. Li1, 2, Chaozhong Zhang1, 2, Jorge Dubcovsky1, 2* \n1 University of California, Davis, CA 95616, USA \n2 Howard Hughes Medical Institute, Chevy Chase, MD 20815, USA \nAuthors ORCIDs and emails \nChengxia Li: ORCID 0000-0003-4634-8451. Email: chxli@ucdavis.edu  \nKun Li : ORCID 0000-0001-8308-1560. Email: kzli@ucdavis.edu  \nC\nhaozhong Zhang: ORCID 0000-0002-0247-3771. Email: cazhang@ucdavis.edu \nJorge Dubcovsky: ORCID 0000-0002-7571-4345. Email: jdubcovsky@ucdavis.edu \n* C orresponding au thor\nJorge Dubcovsky. Email: jdubcovsky@ucdavis.edu\nUniversity of California, Davis, CA 95616, USA\nSubmitted to: Current Opinion in Plant Biology \nThis article is mainly a review but also includes unpublished data. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698073doi: bioRxiv preprint \n\n2 \n \nAbstract \nWheat produces unbranched inflorescences (spikes) composed of smaller inflorescences \n(spikelets) as their fundamental building units. The spikelet number per spike (SNS) is a major \ndeterminant of grain yield and the gene networks that regulate this trait are the focus of this \nreview. Spikelet development starts with the transition of the shoot apical meristem into an \ninflorescence meristem (IM) that produces lateral spikelet meristems (SMs). The rate at which \nSMs are produced and the timing of the IM transition to a terminal spikelet (IM→TS) determine \nthe final SNS. These two traits are regulated by genes expressed in the IM (e.g. meristem identity \ngenes), as well as by the amount of FLOWERING LOCUS T1 (florigen) transported from leaves \nto developing spikes. The SNS can be also increased by the production of spikes with \nsupernumerary spikelets (SS) or branch-like structures that resemble small spikes. Mutations that \ndelay the acquisition of SM identity or that promote a reversion from SM to IM identity can \ninduce the formation of SS. Initial efforts to incorporate these mutations into commercial wheat \nvarieties have faced trade-offs in fertility and grain weight, which will require additional research \nand breeding efforts. Meanwhile, genes and allele combinations that increase SNS without \naffecting the number of spikelets per node have been identified and are being deployed in wheat \nbreeding programs. Recent spatial transcriptomics, single-cell analyses, and multi-omics studies \nof wheat spike development are accelerating the discovery of new genes affecting SNS and \nenhancing our ability to engineer more productive wheat spikes. \n \n  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698073doi: bioRxiv preprint \n\n3 \n \nIntroduction \nInflorescences bear flowers and fruits, and their architectures affect yield in crop species. In the \ngrasses, flowers are replaced by small inflorescences called spikelets (Fig. 1a-c), resulting in \ncompound inflorescences or synflorescences [1]. Both spikelets and flowers are self-contained \ndevelopmental modules that are expressed at different positions of the inflorescence and are \ncommonly regulated by MADS-box genes of the SQUAMOSA clade [2,3]. Given these shared \ndevelopmental characteristics, in this review we extend the simpler nomenclature used for \ninflorescences bearing flowers to the grass inflorescences bearing spikelets.  \nIn many grasses, spikelets are produced on inflorescences called panicles that have branches of \ndecreasing complexity from the base to the top (e.g. rice) [1]. In the Triticeae, however, genuine \nbranches are absent, and spikelets sit directly on the rachis forming a spike (Fig. 1a). Spikelet \ndevelopment proceeds faster at the central part of the spike resulting in the characteristic \nlanceolate shape of the wheat spike (Fig. 1a, d-e). By the time the central spikelets begin forming \nfloret meristems (FMs), the inflorescence meristem (IM) transitions into a spikelet meristem \n(SM) and forms a terminal spikelet (henceforth IM→TS). The terminal spikelet has different \norientation from the lateral spikelets (Fig. 1a). \nThe timing of formation and the identity of meristems shape the inflorescence architecture. In \nwheat, both vegetative shoot apical meristems (vSAM) and IM produce lateral meristems in an \nalternate-distichous pattern, but these meristems differ in activity and identity. In the vSAM, \nbract meristems develop rapidly into leaves whereas their axillary meristems (AxM) are \ntemporarily repressed before initiating tillers. By contrast, in lateral meristems generated by the \nIM, the lower ridges (bract meristems) are repressed, while the upper ridges (equivalent to AxM) \nrapidly develop into spikelets (Fig. 1d).  \nThis review focuses on gene networks that regulate wheat spike development, from the initial \nvSAM→ IM transition to the IM→TS transition (Fig. 1d). It includes genes that affect the rate at \nwhich SMs are produced, the timing of the IM→TS transition, and the number of spikelets in \neach node. Due to space limitations, the regulation of grain number per spikelet is not covered, \nalthough its agronomic importance is recognized. This review integrates recent advances in \nspatial transcriptomics and single-cell expression analyses and discusses the potential \ncontributions of this knowledge to the improvement of wheat productivity. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698073doi: bioRxiv preprint \n\n4 \n \n \nFig.1. (a) Wheat spike showing a terminal spikelet with a different orientation relative to the lateral spikelets. (b) \nWheat spikelet showing two basal sterile bracts (glumes) and multiple florets. (c) Each floret includes a bract \n(lemma) subtending the floral organs: the palea, two lodicules, three stamens, and a pistil. (d) Waddington scale of \nwheat spike development: W1.0= vegetative shoot apical meristem (vSAM), W1.5= initial transition of the vSAM to \nan inflorescence meristem (IM) that produces lateral spikelet meristems (SM). W2.5= double ridge stage. The lower \nridge (LR) corresponds to a repressed bract and the upper ridge (UR) to the AxM of the repressed bract. W3.0= \nglume primordium (GP). W3.5= floret primordia (FM) subtended by the lemma primordium (LP). This stage is also \ncharacterized by the transition of the IM into a SM that generates a terminal spikelet (IM→TS). (e-f) Developing \nspikes at W3.0 for vrn1 vrn2-null mutant (e) and svp1 vrt2-null mutants (f) showing a higher number of SMs. \n \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698073doi: bioRxiv preprint \n\n5 \n \n1. The critical role of MADS-box genes in the initial stages of wheat spike development \nWheat spike development starts with the vSAM→IM transition, which is driven by the \nupregulation of the VERNALIZATION1 (VRN1) gene (Fig. 2a and b). However, VRN1 \nupregulation in the vSAM is not sufficient to complete spike development and stem elongation, \nwhich requires long-day (LD) induction of FLOWERING LOCUS T1 (FT1) in the leaves. The \nFT1 protein (florigen) is transported to the developing spike, where it simultaneously up-\nregulates VRN1 and gibberellin (GA) biosynthetic genes [4] (Table S1). Both GA and VRN1 are \nrequired for the transcriptional up-regulation of SUPPRESSOR OF OVEREXPRESSION OF \nCONSTANS1 (SOC1) and LEAFY (LFY), two important genes in wheat spike development [4]. \nAnother important role of VRN1 in the leaves is to repress the LD flowering repressor VRN2, \nwhich negatively regulates FT1 [5].  \nVRN1 and its closest paralogues FUL2 and FUL3 are MADS-box genes of the SQUAMOSA \nclade that are widely expressed during early spike developing (Fig. 2a-b). Spikes of plants \ncarrying knockout mutations in all VRN1 homeologs (vrn1-null) are normal [5], but further \ncombinations with ful2-null and ful3-null mutations (SQUAMOSA-null) result in abnormal spikes \nwhere spikelets are replaced by vegetative tillers subtended by leaves [3]. These SQUAMOSA-\nnull plants (combined with vrn2-null mutations to avoid extremely late heading) still undergo a \nvSAM→IM transition followed by a double-ridge stage, elongated stems, and AxM that develop \nrapidly in the “spike” region [3]. These results indicate that the SQUAMOSA genes play essential \nroles in suppressing the lower ridge and conferring spikelet meristem identity to the upper ridge, \nbut that they are not essential for inflorescence initiation [3].  \nThe genes responsible for the initial stages of inflorescence development in the wheat \nSQUAMOSA-null mutants are currently unknown, but in Arabidopsis MADS-box genes of the \nSOC1 and SHORT VEGETATIVE PHASE (SVP) clades contribute to this initial inflorescence \ndevelopment [6-8]. A recent study showed that the wheat SOC1-3 protein can compete with \nVEGETATIVE TO REPRODUCTIVE TRANSITION 2 (VRT2) for interactions with VRN1 \nand can also regulate heading time [9]. In wheat, there are five SOC1 paralogs [10] with different \nexpression profiles (Fig. 2c) and possible different functions based on studies of the SOC1 \northologs in rice [11]. Therefore, a more comprehensive study of the wheat SOC1 genes would \nbe required to elucidate their roles in heading time and spike development.  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698073doi: bioRxiv preprint \n\n6 \n \nThe wheat SVP transcription factors, which include SVP1, VRT2, and SVP3, physically interact \nwith the SQUAMOSA proteins [12]. Combined loss-of-function mutations in SVP1 and VRT2 \ngenes delay heading time, increase spikelet number per spike (SNS, Fig. 1e-f), and reduce plant \nheight in a similar way as the vrn1-null mutants. The similar mutant effects and physical protein \ninteractions suggest that SQUAMOSA and SVP proteins may work cooperatively to regulate \nearly spike development [12].  \nFig. 2. Dynamic changes in MADS-\nbox gene expression during wheat \nearly spike development. (a) Curves \nare based on RNA-seq data of wheat \nspike development in tetraploid \nwheat Kronos [13] and individual \npoints are the sum of transcripts per \nmillion of all homeologs and \nparalogs within each MADS-box \ngroup (Table S1). (b-f) Spatial \ntranscriptomics of MADS-box gene \nfamilies across three spike \ndevelopmental stages [14]. (b) \nSQUAMOSA. VRN1 is expressed \nearlier than the other two \nSQUAMOSA genes and all three \ngenes are expressed at relatively \nlower levels at the FMs. (c) SOC1. \nSOC1-2 expression is higher at W1.5 \nand then decreases, whereas \nexpression of other SOC1 genes \nincreases during early spike \ndevelopment. (d) SVP. Genes are \nexpressed in the IM at W1.5 and at \nthe base of the spike at W2.5 and \nW3.5. SVP1 is also expressed in \ndeveloping anthers. (e) SEP1. SEP1-\n6 is the earliest expressed gene in \nthis clade, SEP1-4 is enriched in the \ndistal part of the spike and glume \nprimordia, and SEP1-2 is enriched in \nlemma primordia. (f) SEP3. SEP3-1 \nand SEP3-2 are expressed in FMs \ntogether with other floral homeotic \ngenes (Table S1).  \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698073doi: bioRxiv preprint \n\n7 \n \nThe wheat SVP genes are co-expressed with the SQUAMOSA genes in the early IM at W1.5, but \ntheir expression is restricted to the base of the spike and to the leaves at later stages (Fig. 2d). \nThe transcriptional downregulation of the SVP genes by the SQUAMOSA genes is important for \nthe normal progression of spike development, as SVP proteins can interfere with the \nSQUAMOSA-SEPALLATA1 (SEP1) protein interactions [12]. Plants carrying high-expressing \nnatural VRT2 alleles or transgenic plants overexpressing this gene exhibit more vegetative traits \nsuch as larger glumes and lemmas [12,15]. By contrast, the vegetative characteristics of the vrn1 \nful2 spikes are mitigated when combined with vrt2 mutations [12], highlighting the importance \nof the downregulation of the SVP genes for normal spike development.  \nHigh VRT2 levels are also associated with a higher number of basal rudimentary spikelets [16], \nsuggesting that VRT2 can delay the development of basal spikelets and contribute to the \nlanceolate shape of the wheat spikes [16]. This hypothesis is supported by the triangular shape of \nthe svp1 vrt2 mutant spike at W3.0 (Fig. 1f), where basal spikelet development is not delayed \nrelative to the central spikelets as in the wildtype (Fig. 1d).  \nThe SEP1 and SVP genes show opposite expression profiles (Fig. 2a). Among the SEP1 genes, \nSEP1-6 expression is initiated the earliest (Fig. 2e), and its overexpression in wheat results in \nreduced SNS [17]. Combined mutations in the rice homologs of SEP1-6 (OsMADS34) and \nSEP1-4 (OsMADS5) lead to enhanced branching in rice panicles [18], but have limited effects on \nbarley spikes [19,20]. Mutations in the SEP1-2 ortholog in rice (OsMADS1) result in floral \norgans with leaf-like characteristics [21], whereas mutations in the barley ortholog lead to \nsmaller lemmas and awns at room temperature [22] and branched spikes at high temperatures \n[19]. Given the different effects of the SEP1 mutants in rice and barley, it is not possible to infer \ntheir function in wheat without dedicated research efforts. The SEP3 genes are co-expressed in \nthe FM (Fig. 2f) with other floral homeotic genes [14] and contribute to floral organ identity \n[23]. \nMADS-box proteins form quaternary complexes of varied compositions, each with specific DNA \nbinding specificities [24]. Therefore, the early upregulation of VRN1 in the IM of wheat (Fig. 2b) \nand the gradual changes in the relative expression of other MADS-box genes (Fig. 2a) likely \naffect the composition of these quaternary complexes and the sequential morphological changes \nobserved during the early stages of wheat spike development.  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698073doi: bioRxiv preprint \n\n8 \n \n2 Regulation of wheat spike determinacy and spikelet number per spike (SNS) \nBoth wheat and barley produce spikes with a genetically regulated number of spikelets, but they \ndiffer in the fate of the IM. In the determinate wheat spike, the IM transitions into a SM, and \nforms a terminal spikelet (Fig. 1a), whereas in the indeterminate barley spike the IM is exhausted \nand degenerates. In barley, loss-of-function mutations in GRASSY TILLERS1 (GT1) delay IM \ndegeneration and increase SNS [25], whereas mutations in APETAL2-LIKE5 (ap2l5-null) result \nin a determinate spike with extra florets per spikelet [26]. In wheat, mutations in AP2L5 also lead \nto extra florets per spikelet and are associated with a 21% reduction in SNS [27]. Increased \nexpression of AP2L5 in miR172 resistant alleles increases SNS by 5-10% [28], but neither the \nmutants nor the AP2L5 overexpressing alleles affect wheat spike determinacy. \nBy contrast, wheat vrn1 ful2-null and vrn1 ful2 ful3-null mutants have indeterminate spikes, \ndemonstrating overlapping and essential roles of VRN1 and FUL2 in the IM→TS transition [3]. \nIndividual loss-of-function mutations in VRN1 or FUL2 significantly increase SNS, but the \neffects are more pronounced in vrn1-null (58%) than in ful2-null mutants (10%) [3]. Both vrn1-\nnull and svp1 vrt2-null mutants show delayed spike and spikelet development associated with a \nreduced rate of SM production and a significantly delayed IM→TS transition (Fig. 3a-b).  \nThe rate of SM production is also reduced in lfy and wapo1 mutants relative to the wildtype, but \nthe timing of the IM→TS transition is not significantly altered [29] (Fig. 3c). Loss-of-function \nmutations in LFY, WAPO1, or both result in similar reductions in SNS, suggesting that they work \ncooperatively to regulate this trait. This hypothesis is supported by the physical interaction \nbetween their encoded proteins and their co-expression in the early IM at W1.5 [29]. Natural \nalleles with positive effects on SNS have been identified for WAPO1 [30] and LFY and both \nshow evidence of positive selection and a significant genetic interaction for SNS [31].  \nThe IM→TS transition at W3.5 is characterized by the upregulation of SQUAMOSA genes VRN1 \nand FUL2 and the downregulation of both LFY [29] and the SQUAMOSA PROMOTER \nBINDING PROTEIN-LIKE gene SPL14 in the IM region [14]. Loss-of-function mutations in \nSPL14 result in a premature IM→TS transition and a 42% reduction in SNS, demonstrating an \nimportant role of this gene in spike development [14]. This hypothesis is further supported by the \ndiscovery of natural alleles for SPL14 and its close paralog SPL17 with positive effects on wheat \nSNS [32,33].  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698073doi: bioRxiv preprint \n\n9 \n \nAnother group of genes with significant effects on wheat SNS includes the PHOTOPERIOD1 \n(PPD1) gene and its downstream targets FT1 and FT2. Plants carrying the PPD1 photoperiod-\nsensitive (PS) allele have lower FT1 expression and higher SNS than those with photoperiod-\ninsensitive (PI) alleles. By contrast, wheat plants with loss-of-function mutations in PPD1 show \nreduced FT1 expression and higher SNS [34,35]. Shorter days also reduce FT1 transcript levels, \ndelay spike and spikelet development, and increase SNS (Fig. 3d) [35]. These effects are \nattenuated in the vrn2-null mutants, which exhibit higher FT1 expression and a faster rate of SM \nformation (Fig. 3d). The dominant role of florigen proteins in the regulation of SNS is also \nsupported by the low SNS observed in plants overexpressing FT1 [36], FT2 [37], or FT3 [38]. \nBased on these results, we hypothesize that the reduced rate of SM-production and delayed \nIM→TS transition in the vrn1 vrn2-null and svp1 vrt2-null mutants (Fig. 3a-b) could be \nassociated with the downregulation of FT1 observed in the leaves of these mutants [3,12].  \nFig. 3. Rate of spikelet meristem (SM) formation and IM→TS transition across mutant backgrounds and \nphotoperiods. (a) vrn1 vrn2-null mutant versus vrn2-null control. (b) svp1 vrt2-null mutant versus Kronos control. \n(c) lfy-null and wapo1-null mutants versus Kronos control [29]. (d) Effect of photoperiod on SM production rate: \n16h light /8 h darkness (blue) vs. 12h light / 12h darkness (orange) in Kronos (dotted lines) and Kronos vrn2-null \nmutant (solid line). Longer photoperiods [35] and absence of vrn2 [5] both result in increased transcript levels of \nFT1 and a faster rate of SM production. Arrows indicate the approximate time of the IM→TS transition. Rates are \nexpressed as number of SM formed per day. Data points represent the averages of three dissected apices and error \nbars indicate standard errors of the means (SEM). Data is available on Table S3. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698073doi: bioRxiv preprint \n\n10 \n \n  \nThe FT2 gene also affects SNS, and its encoded protein physically interacts with bZIPC1, a \nhomeodomain-leucine zipper transcription factor [35,39]. The two genes affect SNS in opposite \ndirections, with loss-of-function mutations in FT2 increasing SNS and those in bZIPC1 reducing \nSNS. Natural alleles with positive effects on SNS have been identified for both genes, which \nshow evidence of positive selection and significant genetic interactions for SNS [39,40]. The \ngenetic interactions for SNS discovered for FT2-bZIPC1 and WAPO1-LFY have both led to the \nidentification of optimal allele combinations that maximize SNS [31,40]. These results illustrate \nthe practical value of understanding the gene networks that regulate spike development. \nAn increase in SNS does not guarantee an increase in grain number per spike (GNS) because \nfloret degeneration can offset the gains in SNS. Fortunately, a natural allele of the GRAIN \nNUMBER INCREASE 1 gene (GNI1, or VRS1 in barley) provides a tool to increase spikelet \nfertility and reduce premature floret abortion [41]. Even when increases in SNS are translated \ninto more GNS, reduced grain weight can offset the gains in grain yield. When plants cannot \nproduce sufficient resources to fill the extra grains, a negative correlation is usually observed \nbetween GNS and grain weight. For example, the favorable WAPO1 allele for increased SNS \nshowed positive effects on grain yield only when introgressed into high-biomass genotypes and \nwhen the lines were grown under well-fertilized and well-watered conditions [30,42]. These \nexperiments suggest that balanced increases in both sink and source, together with favorable \nenvironmental conditions, are necessary to translate increases in SNS into higher grain yields.  \n \n3. Regulation of branching in the wheat spike \nAn alternative strategy to improve wheat SNS is to increase the number of spikelets produced \nper node, generating supernumerary spikelets (SS). In mutants such as ‘Miracle-Wheat’ and \n‘Compositum-Barley’ (com2), some basal spikelets are replaced by branch-like structures \nresembling smaller secondary spikes [43,44]. In these mutants, the basal branch-like structures \nresemble small spikes rather than panicle branches, suggesting that those SMs acquired IM \nidentity rather than branch meristem (BM) identity.  \nBranched-like phenotypes and SS have been associated with hypomorphic mutations in the \nAPETALA2/ETHYLENE RESPONSE FACTOR (AP2/ERF) domain transcription factor \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698073doi: bioRxiv preprint \n\n11 \n \nFRIZZY P ANICLE (FZP) [43-45]. However, in more severe FZP mutants in rice [46] and durum \nwheat [14], secondary axes emerge from the axils of the glumes forming new bracts/glumes that \nproduce tertiary axes and repeat the process. These recurrent glume structures suggest that, in \naddition to its effect on the conversion of basal SMs into secondary IMs, an important function \nof FZP is to suppress the glume’s AxM. This result is consistent with the restricted expression of \nFZP in the axils of wheat glume primordia (Fig. 4a-b). \nStudies in wheat and barley suggest that FZP transcript levels are positively regulated by two \ntranscription factors: wheat MULTI-FLORET SPIKELET1 (TaMFS1, referred to as DUO in \n[47]) and barley RAMOSA2 (HvRA2, synonymous VRS4) [44]. In wheat, MFS1 directly \npromotes the expression of FZP [47] and the expression domains of these two genes show \nsignificant overlap at early stages of spikelet development (Fig. 4b). Consistent with its effect on \nFZP, the wheat mfs-B1 mutant also produces spikes with SS [47].  \nIn barley, loss-of-function mutations in HvRA2 (VRS4) are also associated with reduced FZP \ntranscript levels. These mutants also promote lateral spikelets, occasional branch-like formation \n[44], and reduced expression of barley VRS1 and SISTER OF RAMOSA3 [48]. In the wheat \nspike, RA2 is first induced in the abaxial part of initiating SMs, then surrounds the growing SM \n(Fig. 4c), and later delimits the base of the developing spikelet, overlapping with the FZP \nexpression domain in the axil of the second glume (Fig. 4d). \nWheat MFS1 [47] and maize RA2 [49] are also positive regulators of TCP24, which corresponds \nto COMPOSITUM1 (COM1) in barley and BRANCH ANGLE DEFECTIVE 1 (BAD1) in maize \n[49] (Fig. 4e). Loss-of-function mutations in COM1 result in branched barley spikes, suggesting \nthat the active COM1 promotes SM identity and contributes to branch suppression [50,51]. The \ninteractions among MFS1, RA2, and TCP24 in wheat are also supported by significant \ncorrelations among their expression profiles in a single-cell RNA-seq (scRNA-seq) study of \nearly wheat spike development [14] and by their overlapping expression domains (Fig. 4d). \nMoreover, increased expression of all three TCP24 homeologs is observed in the wheat mfs-B1 \nmutant [47]. A similar interaction has been observed in maize, where RA2 operates upstream of \nBAD1 [49]. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698073doi: bioRxiv preprint \n\n12 \n \n \nFig. 4. Genes and gene network regulating branching. (a-d) Single-molecule fluorescence in situ hybridization of \nKronos developing spikes at W2.5 (a and c) and W3.5 (b and d). Cell walls are stained with calcofluor white. gl= \nglume, le= lemma, br= suppressed bract, SM= spikelet meristem, IM= inflorescence meristem. (a-b) Expression \nprofiles of FZP , TCP24, TB1 and MFS1 based on [14] (c-d) Expression profiles of LAX1, RA2, TCP24 and MFS1 in \nKronos spikes based on a 10X-Xenium spatial transcriptomics study. (e) Working model of a gene network \nregulating branching in wheat. Blue arrows indicate promotion, red T-shaped symbols indicate repression, and green \ndouble arrows represent protein-protein interactions. Solid lines are supported by wheat or barley studies, whereas \ndotted lines are supported by studies in the more distantly related rice or maize. Gene names and IDs are included in \nTable S1. Numbers in parenthesis indicate references: 1:[47], 2:[49], 3:[44], 4:[45], 5:[50], 6:[51], 7:[53], 8:[55], \n9:[56], 10:[57], 11:[54], 12:[56]. (f) Left: Spikes of wildtype Kronos and tb1-null CRISPR mutant showing SS. \nRight: nodes bearing two or three spikelets. Note the different orientation of the terminal and lateral spikelets, their \nsimilar development, and the higher frequency of triplets at the center of the spike.  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698073doi: bioRxiv preprint \n\n13 \n \n \nTCP24 belongs to the CYC/TB1 clade of TCP transcription factors, which also includes TCP22, \nTEOSINTE BRANCHED1 (TB1), and TB2. In barley, TB1 is known as INTERMEDIUM \nSPIKE-C (INT-C or VRS5), and its mutant contributes to the six-rowed phenotype [52]. In \nwheat, TB1 expression is detected in the SMs at W2.5 and becomes restricted to the base of the \nspikelets at W3.5. At both stages it is more abundant at the base of the spike (Fig. 4a-b). \nKnockout mutations of both TB1 homeologs in tetraploid wheat (tb1-null) result in frequent SS \n(Fig. 4f, Table S2), as reported previously [53]. These results suggest that TB1/INT-C modulates \nSM timing or identity, thereby contributing to the repression of SS in both barley and wheat. \nSurprisingly, TB1 duplications or transgenic overexpression were also associated with SS [54], \nsuggesting that balanced levels of TB1 may be required to prevent SS formation. \nIn maize, TB1 transcription is regulated by BARREN STALK1 (BA1), the ortholog of rice LAX \nP ANICLE1 (LAX1) [56]. BA1 acts downstream of auxin signaling to position boundary regions \nfor AxM formation [58]. Loss-of-function mutations in BA1 or LAX1 result in drastic reductions \nin panicle branching and lateral spikelet formation in both maize and rice [55,56], whereas \nmutations in the wheat ortholog have milder effects [59]. Wheat plants with mutations in the \nthree LAX1 homeologs have a more compact spike [59] that resembles gain-of-function \nmutations in the domestication gene Q (AP2L5) [28,60]. By contrast, overexpression of wheat \nLAX1 reduces spike threshability and spikelet density, similar to the pre-domestication q allele \n[59]. LAX1 and Q proteins physically interact with each other, providing a potential mechanism \nfor their opposite effects on these domestication traits [59]. In wheat, FZP has been reported to \ndirectly repress LAX1 transcription by binding to its regulatory regions [45]. However, the \nfunctional significance of this interaction remains unclear, given that FZP is expressed later than \nLAX1 in developing wheat spikes (Fig. 4a-c). From the early stages of wheat spike development, \nLAX1 is expressed at the adaxial boundary layer of the SMs (Fig. 4c), similar to observations in \nrice [55] and maize [56], suggesting a conserved function.  \nIn hexaploid wheat, a particular class of SS, designated as paired spikelets (PS), is characterized \nby the formation of a secondary spikelet immediately adjacent to and below (abaxial) the more \nadvanced main spikelet. At maturity, PSs appear parallel to each other and are more frequent at \nthe central nodes of the spike. PS are affected by genes in the photoperiod pathway, with shorter \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698073doi: bioRxiv preprint \n\n14 \n \nphotoperiods, mutations in PPD1, and reduced expression of FT1 all increasing PS frequency \n[57]. Since these changes are associated with reduced expression of meristem identity genes \n(VRN1, FUL2, and FUL3), the authors of [57] hypothesized that this reduction may contribute to \nthe delayed initiation of the SM, providing the time for PS formation.  \nIn summary, none of the known mutants in barley and wheat form genuine branches as observed \nin the panicles of other grasses, suggesting that these species have lost the ability to generate \nfunctional branch meristems. However, multiple mutants generate SS or branch-like structures \nby modifying the timing or identity of the SMs, providing the time for the formation of one or \nmore lateral spikelets. It remains unclear why some mutations induce higher SS frequencies at \nthe center of the spike (e.g. Tb1), whereas others result in branch-like structures with an \nelongated secondary rachis (e.g. FZP). Studying branching in the un-branched Triticeae spikes \ncan provide valuable insights into our understanding of grass inflorescence development and \nevolution. \n \n \nConclusions and perspectives \nIncreasing spikelet number through SS or branching has been an attractive strategy to increase \ngrain yield, but initial efforts have been limited by trade-offs in fertility and grain size. The GNI1 \nallele for high fertility [41] may help mitigate the reductions in SS fertility, whereas high-\nbiomass varieties with higher source resources may limit negative correlations between grain \nnumber and grain size. A recent study using a cross between a spike-branching landrace and the \nelite durum cultivar CIRNO showed that a better balance of source–sink traits can reduce yield \npenalties associated with branched phenotype [61]. However, developing higher yielding wheat \nvarieties with SS or branched spikes will require a dedicated breeding effort to fine-tune the \nmultiple epistatic interactions within the complex gene network regulating branching.  \nMeanwhile, rapid progress has been made in the identification of genes that control grain number \nby modifying the rate of SM production, the timing of the IM →TS transition, and/or the number \nof fertile florets. Favorable alleles and allele combinations are being identified and incorporated \ninto breeding programs through marker-assisted selection or by integrating functional SNPs into \nhigh-throughput marker platforms used for genomic selection. The impressive number of \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698073doi: bioRxiv preprint \n\n15 \n \nspikelets and grains produced by high yielding triticale varieties suggest that there is still room \nfor improvement of the SNS trait in wheat.  \nThe identification and functional characterization of genes regulating wheat spike development \nhave been greatly accelerated in recent years, fueled by new tools including CRISPR, sequenced \nmutant populations, improved wheat transformation efficiency, and sequenced genomes. These \nresources have been complemented by single-cell RNA-seq and spatial transcriptomic studies, \nwhich provide cellular resolution to the rapid changes in gene expression that occur during spike \ndevelopment [14,62]. In parallel, multi-omics studies are integrating expression profiles, \nchromatin accessibility, binding motifs, GWAS, and evolutionary information into \ncomprehensive networks that will help prioritize the functional characterization of novel \nregulators of wheat spike development [63,64]. The knowledge generated by these new tools can \nenhance our ability to engineer changes in spike development.  \nEach year, more than 200 trillion wheat spikes are harvested globally, providing one fifth of the \ncalories and proteins consumed by the human population. Therefore, every small improvement in \nwheat spike productivity is worth the effort.  \n \nAcknowledgements \nThe authors appreciate the comments and suggestions provided by Dr. Thorsen Schnurbusch \n(Leibniz Institute of Plant Genetics and Crop Plant Research, IPK) and Dr. Scott Boden (The \nUniversity of Adelaide) on the branching section of this review. The authors also thank Dr. \nFrancois Parcy (Universite Grenoble Alpes) for valuable discussions on inflorescence \ndevelopment. \n \nFunding statement  \nThis project was supported by the Agriculture and Food Research Initiative Competitive Grant \n2022-68013-36439 (WheatCAP) from the USDA National Institute of Food and Agriculture, and \nby the Howard Hughes Medical Institute. \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698073doi: bioRxiv preprint \n\n16 \n \nReferences \n1. Kellogg EA: Genetic control of branching patterns in grass inflorescences. Plant Cell \n2022, 34:2518-2533. \n2. Ferrándiz C, Gu Q, Martienssen R, Yanofsky MF: Redundant regulation of meristem \nidentity and plant architecture by FRUITFULL, APETALA1 and CAULIFLOWER. \nDevelopment 2000, 127:725-734. \n3.** Li C, Lin H, Chen A, Lau M, Jernstedt J, Dubcovsky J: Wheat VRN1, FUL2 and FUL3 \nplay critical and redundant roles in spikelet development and spike determinacy. \nDevelopment 2019, 146:dev175398. \nThis study demonstrates that the SQUAMOSA-clade MADS-box genes VRN1, FUL2, and FUL3 in wheat \nplay critical and redundant roles in spikelet development. These genes are essential for the suppression of \nthe lower leaf ridge and the establishment of spikelet meristem identity. Moreover, loss-of-function \nmutations in VRN1 or FUL2 lead to a significant increase in spikelet number per spike, and the \ninflorescence meristem of the vrn1 ful2 double mutant becomes indeterminate, indicating both genes are \nrequired for spike determinacy in wheat. \n4. Pearce S, Vanzetti LS, Dubcovsky J: Exogenous gibberellins induce wheat spike \ndevelopment under short days only in the presence of VERNALIZATION1. Plant \nPhysiology 2013, 163:1433-1445. \n5. Chen A, Dubcovsky J: Wheat TILLING mutants show that the vernalization gene VRN1 \ndown-regulates the flowering repressor VRN2 in leaves but is not essential for \nflowering. PLoS Genetics 2012, 8:e1003134. \n6. Lee J, Oh M, Park H, Lee I: SOC1 translocated to the nucleus by interaction with AGL24 \ndirectly regulates leafy. Plant J 2008, 55:832-843. \n7. Perez-Ruiz RV , Garcia-Ponce B, Marsch-Martinez N, Ugartechea-Chirino Y , Villajuana-\nBonequi M, de Folter S, Azpeitia E, Davila-Velderrain J, Cruz-Sanchez D, Garay-Arroyo \nA, et al.: XAANTAL2 (AGL14) is an important component of the complex gene \nregulatory network that underlies Arabidopsis shoot apical meristem transitions. \nMolecular Plant 2015, 8:796-813. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698073doi: bioRxiv preprint \n\n17 \n \n8. Liu C, Xi W, Shen L, Tan C, Yu H: Regulation of floral patterning by flowering time genes. \nDevelopmental Cell 2009, 16:711-722. \n9. Luo X, Liu B, Xie L, Wang K, Xu D, Tian X, Xie L, Li L, Ye X, He Z, et al.: The TaSOC1-\nTaVRN1 module integrates photoperiod and vernalization signals to regulate wheat \nflowering. Plant Biotechnol J 2024, 22:635-649. \n10. Schilling S, Kennedy A, Pan S, Jermiin LS, Melzer R: Genome-wide analysis of MIKC-\ntype MADS-box genes in wheat: pervasive duplications, functional conservation and \nputative neofunctionalization. New Phytologist 2020, 225:511-529. \n11. Ryu CH, Lee S, Cho LH, Kim SL, Lee YS, Choi SC, Jeong HJ, Yi J, Park SJ, Han CD, et al.: \nOsMADS50 and OsMADS56 function antagonistically in regulating long day (LD)-\ndependent flowering in rice. Plant Cell Environ 2009, 32:1412-1427. \n12.* Li K, Debernardi JM, Li C, Lin H, Zhang C, Jernstedt J, von Korff M, Zhong J, Dubcovsky \nJ: Interactions between SQUAMOSA and SVP MADS-box proteins regulate \nmeristem transitions during wheat spike development. The Plant Cell 2021, 33:3621-\n3644. \nThe SVP-clade MADS-box proteins, including SVP1, VRT2, and SVP3, physically interact with the \nSQUAMOSA proteins, and these interactions can interfere with SQUAMOSA-SEPALLATA interactions. \nLoss-of-function mutations in VRT2 and SVP1 result in delays in heading time and increases in spikelet \nnumber per spike like those observed in SQUAMOSA mutants, suggesting that these genes play \noverlapping roles in the regulation of heading time and SNS during the early reproductive phase. As spike \ndevelopment progresses, the SVP genes are repressed by the SQUAMOSA genes. This repression is \nimportant for enabling the interactions between SQUAMOSA and SEPALLATA1 MADS-box proteins, \nwhich are required for the initial stages of spikelet development. Therefore, SQUAMOSA-SVP \ninteractions are key regulators of meristem transitions in wheat spike development. \n13. VanGessel C, Hamilton J, Tabbita F, Dubcovsky J, Pearce S: Transcriptional signatures of \nwheat inflorescence development. Scientific Reports 2022, 12:17224. \n14.* Xu X, Lin H, Zhang J, Burguener G, Paraiso F, Li K, Tumelty C, Li C, Liu Y , Dubcovsky J: \nSpatial and single-cell expression analyses reveal complex expression domains in \nearly wheat spike development. Genome Biology 2025, 26:352. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698073doi: bioRxiv preprint \n\n18 \n \nThis study integrated single-molecule fluorescence in situ hybridization (smFISH) and single-cell RNA \nsequencing (scRNA-seq) during three early stages of spike development in a tetraploid spring wheat to \ngenerate an atlas of cell clusters and expression domains at high spatial and cellular resolution. The study \nused scRNA-seq information to impute the expression of 74,464 genes into the spatially anchored cells, \nextending spatiotemporal resolution to a genome-wide scale. The study provides a public website to \nvisualize the smFISH and imputed gene expression (https://dubcovskylab.ucdavis.edu/spatial-\ntranscriptomic). Functional validation of key genes, including LFY, SPL14 and FZP documented the \nvalue of the spatial transcriptomics data for inferring gene function. The study was performed in a \ntetraploid spring wheat variety. \n15. Adamski NM, Simmonds J, Brinton JF, Backhaus AE, Chen Y , Smedley M, Hayta S, Florio \nT, Crane P, Scott P, et al.: Ectopic expression of Triticum polonicum VRT-A2 underlies \nelongated glumes and grains in hexaploid wheat in a dosage-dependent manner. The \nPlant Cell 2021, 33:2296-2319. \n16. Backhaus AE, Lister A, Tomkins M, Adamski NM, Simmonds J, Macaulay I, Morris RJ, \nHaerty W, Uauy C: High expression of the MADS-box gene VRT2 increases the \nnumber of rudimentary basal spikelets in wheat. Plant Physiol 2022, 189:1536-1552. \n17. Wang YG, Yu HP, Tian CH, Sajjad M, Gao CC, Tong YP, Wang XF, Jiao YL: Transcriptome \nassociation identifies regulators of wheat spike architecture. Plant Physiology 2017, \n175:746-757. \n18. Zhu W, Yang L, Wu D, Meng Q, Deng X, Huang G, Zhang J, Chen X, Ferrandiz C, Liang W, \net al.: Rice SEPALLATA genes OsMADS5 and OsMADS34 cooperate to limit \ninflorescence branching by repressing the TERMINAL FLOWER1-like gene RCN4. \nNew Phytologist 2022, 233:1682-1700. \n19. Li G, Kuijer HNJ, Yang X, Liu H, Shen C, Shi J, Betts N, Tucker MR, Liang W, Waugh R, et \nal.: MADS1 maintains barley spike morphology at high ambient temperatures. \nNature Plants 2021, 7:1093-1107. \n20. Shen C, Yang X, Wang D, Li G, Tucker MR: Functional retrogression of LOFSEPs in \nspecifying floral organs in barley. aBIOTECH 2025, 6:1-11. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698073doi: bioRxiv preprint \n\n19 \n \n21. Wu D, Liang WQ, Zhu WW, Chen MJ, Ferrandiz C, Burton RA, Dreni L, Zhang DB: Loss of \nLOFSEP transcription factor function converts spikelet to leaf-like structures in \nrice. Plant Physiology 2018, 176:1646-1664. \n22. Zhang Y , Shen C, Li G, Shi J, Yuan Y , Ye L, Song Q, Shi J, Zhang D: MADS1-regulated \nlemma and awn development benefits barley yield. Nature Communications 2024, \n15:301. \n23. Shen C, Zhang Y , Li G, Shi J, Wang D, Zhu W, Yang X, Dreni L, Tucker MR, Zhang D: \nMADS8 is indispensable for female reproductive development at high ambient \ntemperatures in cereal crops. Plant Cell 2023, 36:65-84. \n24. Theissen G, Melzer R, Rumpler F: MADS-domain transcription factors and the floral \nquartet model of flower development: linking plant development and evolution. \nDevelopment 2016, 143:3259-3271. \n25. Shanmugaraj N, Rajaraman J, Kale S, Kamal R, Huang Y , Thirulogachandar V , Garibay-\nHernandez A, Budhagatapalli N, Tandron Moya YA, Hajirezaei MR, et al.: Multilayered \nregulation of developmentally programmed pre-anthesis tip degeneration of the \nbarley inflorescence. Plant Cell 2023, 35:3973-4001. \n26.\n* Zhong J, van Esse GW, Bi X, Lan T, Walla A, Sang Q, Franzen R, von Korff M: \nINTERMEDIUM-M encodes an HvAP2L-H5 ortholog and is required for \ninflorescence indeterminacy and spikelet determinacy in barley. Proceedings of the \nNational Academy of Sciences of the United States of America 2021, 118:e2011779118. \nThis study identifies the causal gene for the intermedium-m (int-m) and double seed1 (dub1) mutant \nphenotypes, which convert the indeterminate inflorescences of barley into wheat-like determinate \ninflorescences characterized by a terminal spikelet and spikelets with multiple florets. INT-M/DUB1 \nencodes an APETALA2-like transcription factor (HvAP2L-H5) that suppresses ectopic and precocious \nspikelet initiation and maintains inflorescence meristem activity. This study further demonstrates that \nHvAP2L-H5 inhibits the identity shift of the inflorescence meristem to a terminal spikelet meristem in \nbarley. Null mutations in AP2L-5 result in fewer spikelets per spike but additional florets per spikelet. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698073doi: bioRxiv preprint \n\n20 \n \n27. Debernardi JM, Greenwood JR, Jean Finnegan E, Jernstedt J, Dubcovsky J: APETALA 2-\nlike genes AP2L2 and Q specify lemma identity and axillary floral meristem \ndevelopment in wheat. Plant Journal 2020, 101:171-187. \n28. Greenwood JR, Finnegan EJ, Watanabe N, Trevaskis B, Swain SM: New alleles of the \nwheat domestication gene Q reveal multiple roles in growth and reproductive \ndevelopment. Development 2017, 144:1959-1965. \n29. Paraiso F, Lin H, Li C, Woods DP, Lan T, Tumelty C, Debernardi JM, Joe A, Dubcovsky J: \nLEAFY and WAPO1 jointly regulate spikelet number per spike and floret \ndevelopment in wheat. Development 2024, 151:dev202803. \n30. Kuzay S, Lin H, Li C, Chen S, Woods DP, Zhang J, Lan T, von Korff M, Dubcovsky J: \nWAPO-A1 is the causal gene of the 7AL QTL for spikelet number per spike in wheat. \nPLoS Genetics 2022, 18:e1009747. \n31. Zhang J, Burguener GF, Paraiso F, Dubcovsky J: Natural alleles of LEAFY and WAPO1 \ninteract to regulate spikelet number per spike in wheat. Theoretical and Applied \nGenetics 2024, 137:257. \n32. Chen H, Zhang X, Xu SH, Song CX, Mao HL: TaSPL17s act redundantly with TaSPL14s \nto control spike development and their elite haplotypes may improve wheat grain \nyield. Frontiers in Plant Science 2023, 14:1229827. \n33. Liu Y , Chen J, Yin C, Wang Z, Wu H, Shen K, Zhang Z, Kang L, Xu S, Bi A, et al.: A high-\nresolution genotype-phenotype map identifies the TaSPL17 controlling grain \nnumber and size in wheat. Genome Biology 2023, 24:196. \n34. Shaw LM, Turner AS, Herry L, Griffiths S, Laurie DA: Mutant alleles of Photoperiod-1 in \nwheat (Triticum aestivum L.) that confer a late flowering phenotype in long days. \nPLoS ONE 2013, 8:e79459. \n35. Gauley A, Boden SA: Stepwise increases in FT1 expression regulate seasonal progression \nof flowering in wheat (Triticum aestivum). New Phytologist 2021, 229:1163-1176. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698073doi: bioRxiv preprint \n\n21 \n \n36. Lv B, Nitcher R, Han X, Wang S, Ni F, Li K, Pearce S, Wu J, Dubcovsky J, Fu D: \nCharacterization of FLOWERING LOCUS T1 (FT1) gene in Brachypodium and \nwheat. Plos One 2014, 9: e94171. \n37. Shaw LM, Lyu B, Turner R, Li C, Chen F, Han X, Fu D, Dubcovsky J: FLOWERING \nLOCUS T2 regulates spike development and fertility in temperate cereals. Journal of \nExperimental Botany 2019, 70:193-204. \n38. Gauley A, Lacey I, Gonzalez-Suarez P, Taylor H, Hirsz D, Hayta S, Smedley M, Harwood W, \nGriffiths S, Cheng S, et al.: Fine-tuning wheat development for the winter to spring \ntransition. Plant Commun 2025, 6:101501. \n39. Glenn P, Zhang J, Brown-Guedira G, DeWitt N, Cook JP, Li K, Akhunov E, Dubcovsky J: \nIdentification and characterization of a natural polymorphism in FT-A2 associated \nwith increased number of grains per spike in wheat. Theoretical and Applied Genetics \n2022, 135:679-692. \n40. Glenn P, Woods DP, Zhang J, Gabay G, Odle N, Dubcovsky J: Wheat bZIPC1 interacts \nwith FT2 and contributes to the regulation of spikelet number per spike. Theoretical \nand Applied Genetics 2023, 136:237. \n41. Sakuma S, Golan G, Guo Z, Ogawa T, Tagiri A, Sugimoto K, Bernhardt N, Brassac J, \nMascher M, Hensel G, et al.: Unleashing floret fertility in wheat through the mutation \nof a homeobox gene. Proceedings of the National Academy of Sciences of the United \nStates of America 2019, 116:5182-5187. \nThis study reports that Grain Number Increase 1 (GNI1), a homeodomain leucine zipper class I (HD-Zip \nI) transcription factor, is an important contributor to floret fertility in both wheat and barley (where it is \nknown as VRS1). GNI1 inhibits rachilla growth and apical floret development; therefore, a single amino \nacid substitution in the HD domain of GNI1 results in a reduced‐function allele, which increases grain \nnumber compared to the ancestral wild type allele, thereby leading to higher yield in field conditions. \nHarnessing natural and induced variation in GNI1 and its orthologs that favor increased grain production \nrepresents a promising strategy for crop improvement.\n \n42. Kuzay S, Xu Y , Zhang J, Katz A, Pearce S, Su Z, Fraser M, Anderson JA, Brown-Guedira G, \nDeWitt N, et al.: Identification of a candidate gene for a QTL for spikelet number per \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698073doi: bioRxiv preprint \n\n22 \n \nspike on wheat chromosome arm 7AL by high-resolution genetic mapping. \nTheoretical and Applied Genetics 2019, 132:2689–2705. \n43. Dobrovolskaya O, Pont C, Sibout R, Martinek P, Badaeva E, Murat F, Chosson A, Watanabe \nN, Prat E, Gautier N, et al.: FRIZZY PANICLE drives supernumerary spikelets in \nbread wheat. Plant Physiology 2015, 167:189-199. \n44.** Poursarebani N, Seidensticker T, Koppolu R, Trautewig C, Gawronski P, Bini F, Govind G, \nRutten T, Sakuma S, Tagiri A, et al.: The genetic basis of composite spike form in \nbarley and 'Miracle-Wheat'. Genetics 2015, 201:155-165. \nThis study reports the identification of the wheat FRIZZY P ANICLE (FZP) gene, which encodes a \nmember of the APETALA2/Ethylene Response Factor (AP2/ERF) transcription factor family, as the \ncausal gene responsible for the branch-like structures form at the spikes of the barley COMPOSITUM2 \nmutants and in ‘Miracle Wheat’. Mutants in both barley and wheat showed amino acid changes either in \nthe AP2/ERF domain or in the conserved terminal region but not loss-of-function mutations. In barley, \nCOM2 transcripts are significantly down-regulated in the vrs4.k mutant suggesting that VRS4 \n(synonymous HvRA2) may function upstream of COM2. \n45. Du DJ, Zhang DX, Yuan J, Feng M, Li ZJ, Wang ZH, Zhang ZH, Li XT, Ke WS, Li RH, et \nal.: FRIZZY PANICLE defines a regulatory hub for simultaneously controlling \nspikelet formation and awn elongation in bread wheat. New Phytologist 2021, \n231:814-833. \n46. Komatsu M, Chujo A, Nagato Y , Shimamoto K, Kyozuka J: FRIZZY PANICLE is required \nto prevent the formation of axillary meristems and to establish floral meristem \nidentity in rice spikelets. Development 2003, 130:3841-3850. \n47.* Wang YG, Du F, Wang J, Wang K, Tian CH, Qi XQ, Lu F, Liu XG, Ye XG, Jiao YL: \nImproving bread wheat yield through modulating an unselected AP2/ERF gene. \nNature Plants 2022, 8:930-939. \nThis study reports that the DUO-B1 gene (referred to as MFS-B1 in this review using the original name \nfrom rice) encodes an AP2/ERF transcription factor that regulates spike architecture in bread wheat. \nMutations in DUO-B1 lead to a mild supernumerary spikelet (SS) phenotype and increased grain yield \nunder field conditions. This study further demonstrates that DUO-B1 regulates spike architecture by \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698073doi: bioRxiv preprint \n\n23 \n \npromoting FZP and TCP24 (TraesCS5A02G207300 and TraesCS5B02G205600) which was incorrectly \nreferred to in this study as TB1.  \n48. Koppolu R, Anwar N, Sakuma S, Tagiri A, Lundqvist U, Pourkheirandish M, Rutten T, Seiler \nC, Himmelbach A, Ariyadasa R, et al.: Six-rowed spike4 (Vrs4) controls spikelet \ndeterminacy and row-type in barley. Proceedings of the National Academy of Sciences \nof the United States of America 2013, 110:13198-13203. \n49. Bai F, Reinheimer R, Durantini D, Kellogg EA, Schmidt RJ: TCP transcription factor, \nBRANCH ANGLE DEFECTIVE 1 (BAD1), is required for normal tassel branch \nangle formation in maize. Proceedings of the National Academy of Sciences of the \nUnited States of America 2012, 109:12225-12230. \n50. Moraes TD, van Es SW, Hernández-Pinzón I, Kirschner GK, van der Wal F, da Silveira SR, \nBusscher-Lange J, Angenent GC, Moscou M, Immink RGH, et al.: The TCP \ntranscription factor HvTB2 heterodimerizes with VRS5 and controls spike \narchitecture in barley. Plant Reproduction 2022, 35:205-220. \n51. Poursarebani N, Trautewig C, Melzer M, Nussbaumer T, Lundqvist U, Rutten T, Schmutzer \nT, Brandt R, Himmelbach A, Altschmied L, et al.: COMPOSITUM 1 contributes to the \narchitectural simplification of barley inflorescence via meristem identity signals. \nNature Communications 2020, 11:5138. \n52. Ramsay L, Comadran J, Druka A, Marshall DF, Thomas WT, Macaulay M, MacKenzie K, \nSimpson C, Fuller J, Bonar N, et al.: INTERMEDIUM-C, a modifier of lateral spikelet \nfertility in barley, is an ortholog of the maize domestication gene TEOSINTE \nBRANCHED 1. Nature Genetics 2011, 43:169-172. \n53. Hale CO: Genetic optimization of seed size, plant height, and tillering traits for \nenhanced yield in Montana wheat. Edited by. Bozeman, Montana: Montanta State \nUniversity; 2025. vol PhD.] \n54. Dixon LE, Greenwood JR, Bencivenga S, Zhang P, Cockram J, Mellers G, Ramm K, \nCavanagh C, Swain SM, Boden SA: TEOSINTE BRANCHED1 regulates inflorescence \narchitecture and development in bread wheat (Triticum aestivum). Plant Cell 2018, \n30:563-581. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698073doi: bioRxiv preprint \n\n24 \n \n55. Komatsu K, Maekawa M, Ujiie S, Satake Y , Furutani I, Okamoto H, Shimamoto K, Kyozuka \nJ: LAX and SPA: Major regulators of shoot branching in rice. Proceedings of the \nNational Academy of Sciences of the United States of America 2003, 100:11765-11770. \n56. Gallavotti A, Zhao Q, Kyozuka J, Meeley RB, Ritter MK, Doebley JF, Pe ME, Schmidt RJ: \nThe role of barren stalk1 in the architecture of maize. Nature 2004, 432:630-635. \n57.** Boden SA, Cavanagh C, Cullis BR, Ramm K, Greenwood J, Jean Finnegan E, Trevaskis B, \nSwain SM: Ppd-1 is a key regulator of inflorescence architecture and paired spikelet \ndevelopment in wheat. Nature Plants 2015, 1:14016. \nThis study shows that PHOTOPERIOD1 (PPD1) is a key regulator of spike architecture in wheat. It \nfurther demonstrates that PPD1 inhibits paired spikelet formation by promoting FT1 expression. Reduced \nFT1 expression in wheat is associated with paired spikelets. Unlike the supernumerary spikelet phenotype \nobserved in “Miracle Wheat”, paired spikelets are characterized by the formation of a secondary spikelet \ndirectly adjacent to and below the primary spikelet, which is usually more advanced in its development \nthan the secondary spikelet. At maturity, the PS appear parallel and are more frequent at the center of the \nspike. The study hypothesizes that low levels of FT1 lead to reduced expression of the meristem identity \ngenes, which delay the formation of the SM and provides the time necessary for the formation of the \nlateral meristem that originates the secondary spikelet.\n \n58. Galli M, Liu Q, Moss BL, Malcomber S, Li W, Gaines C, Federici S, Roshkovan J, Meeley \nR, Nemhauser JL, et al.: Auxin signaling modules regulate maize inflorescence \narchitecture. Proceedings of the National Academy of Sciences of the United States of \nAmerica 2015, 112:13372-13377. \n59. He G, Zhang Y , Liu P, Jing Y , Zhang L, Zhu Y , Kong X, Zhao H, Zhou Y , Sun J: The \ntranscription factor TaLAX1 interacts with Q to antagonistically regulate grain \nthreshability and spike morphogenesis in bread wheat. New Phytologist 2021, \n230:988-1002. \n60.* Debernardi JM, Lin H, Chuck G, Faris JD, Dubcovsky J: microRNA172 plays a crucial \nrole in wheat spike morphogenesis and grain threshability. Development 2017, \n144:1966-1975. \nThis study shows that a nucleotide substitution in the miR172 target site of the Q allele (AP2L5) disrupts \nthe ability of miR172 to cleave the mRNA. This results in increased levels of AP2L5 contributing to more \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698073doi: bioRxiv preprint \n\n25 \n \ncompact spikes and free-threshing grains. Increased levels of AP2L5 favor the partial conversion of \nglumes to lemmas by reducing sclerenchymatic tissue, increasing awn length and reducing the axillary \nmeristem repression. By contrast, reduced levels of AP2L5 favor the partial conversion of lemmas to \nglumes by increasing sclerenchymatic tissue, reducing awn length and repressing flower development. \n61. Abbai R, Golan G, Longin CFH, Schnurbusch T: Grain yield trade-offs in spike-branching \nwheat can be mitigated by elite alleles affecting sink capacity and post-anthesis \nsource activity. Journal of Experimental Botany 2024, 75:88-102. \n62. Long KA, Lister A, Jones MRW, Adamski NM, Ellis RE, Chedid C, Carpenter SJ, Liu X, \nBackhaus AE, Goldson A, et al.: Spatial transcriptomics reveals expression gradients \nin developing wheat inflorescences at cellular resolution. bioRxiv 2024, \nhttps://doi.org/10.1101/2024.12.19.629411. \n63. Chen Y , Guo Y , Guan P, Wang Y , Wang X, Wang Z, Qin Z, Ma S, Xin M, Hu Z, et al.: A \nwheat integrative regulatory network from large-scale complementary functional \ndatasets enables trait-associated gene discovery for crop improvement. Molecular \nPlant 2023, 16:393-414. \nThis study presents a genome-wide wheat integrative gene regulatory network (wGRN) that combines an \nupdated genome annotation, large-scale gene expression profiles, transcription factor (TF) binding motifs \n(DAP-seq), chromatin accessibility data, and evolutionary conserved regulatory regions. Using these \ndatasets, the study built seven independent regulatory networks and integrated them into a single large-\nscale network that comprises ~7.2 million TF–target interactions, linking 5,947 TFs to 127,439 target \ngenes. The authors demonstrated the applicability of wGRN in uncovering previously unknown gene \nfunctions and pathways, interpreting new datasets, and prioritizing candidate genes from GWAS and map-\nbased cloning studies. By integrating the wGRN with a spike transcriptome time-series dataset, this study \nconstructed high-resolution gene networks that enabled the identification of novel regulators of SNS and \nenhanced the power of spike phenotypic trait prediction. An interactive webserver \n(http://wheat.cau.edu.cn/wGRN) is provided to explore gene regulation and discover trait-associated \ngenes. \n64. Lin X, Xu Y , Wang D, Yang Y , Zhang X, Bie X, Gui L, Chen Z, Ding Y , Mao L, et al.: \nSystematic identification of wheat spike developmental regulators by integrated \nmulti-omics, transcriptional network, GWAS, and genetic analyses. Molecular Plant \n2024, 17:438-459. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698073doi: bioRxiv preprint \n\n26 \n \nThe study used a multi-omics approach to analyze and integrate the transcriptome and epigenome profiles \nat eight early stages of the spike development in winter wheat. It combines RNA-seq, ATAC-seq, and \nhistone modification data to construct a transcriptional regulatory network (TRN) of spike development. \nFurther integration of this TRN with genome-wide association studies (GWAS) identified numerous \ntranscription factors potentially involved in spike development. Mutant analysis confirmed spike \nphenotypes for 36 of these transcription factors and identified MYB30 as a direct target of FZP . The \ndatasets generated in this study are available in the open access Wheat Spike Multi-Omic Database \n(http://39.98.48.156:8800/#/).  \n \nReferences and recommended reading Papers of particular interest, published within the period \nof review, have been highlighted as: * of special interest and ** of outstanding interest \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698073doi: bioRxiv preprint","source_license":"CC-BY-4.0","license_restricted":false}