Yidong Wang, Shanshan Wei, Yubing He, Lang Yan,Rongchen Wang, Yunde Zhao*

aNational Key Laboratory of Crop Genetic Improvement,National Center of Plant Gene Research(Wuhan),Huazhong Agricultural University,Wuhan 430070,Hubei,China

bCollege of Life Science and Technology,Huazhong Agricultural University,Wuhan 430070,Hubei,China

cSection of Cell and Developmental Biology,University of California San Diego,La Jolla,CA 92093-0116,USA

Keywords:LAX1 FZP Rice CRISPR Flower development Sterile lemma

ABSTRACT Rice florets are subtended by two sterile lemmas, whose origin and biological functions have not been studied extensively. Here we demonstrate that two putative transcription factors, LAX PANICLE1 (LAX1) and FRIZZY PANICLE (FZP), synergistically control the development of sterile lemmas. Both LAX1 and FZP are previously known for their roles in panicle and floret development. Disruption of either LAX1 or FZP greatly reduces the number of floret development. We generated new lax1 mutants (lax1-c) using CRISPR/Cas9 gene editing technology.In addition to the expected lax panicle phenotypes,we noticed that a significant number of spikelets of lax1-c developed elongated sterile lemmas. Moreover,our characterization of lax1-RNAi plants also revealed sterile lemma phenotypes similar to lax1-c mutants. We isolated a weak allele of fzp (fzp-14) in a genetic screen for lax1-1 enhancers. The fzp-14 lax1-1 double mutants completely eliminated flower development.Interestingly, the isolated fzp-14 produced spikelets with elongated sterile lemmas.Furthermore, fzp-14 was haploid-insufficient in the lax1-1 background whereas fzp-14 heterozygous plants were indistinguishable from wild type plants.The lax1-1 fzp-14+/-also developed elongated sterile lemma as observed in lax1-c, lax1-RNAi, and fzp-14, suggesting that LAX1 and FZP synergistically control sterile lemma development.

1. Introduction

The architecture of rice inflorescence is largely determined by the number of branches and spikelets [1-3]. A rice spikelet consists of one floret that is subtended by a pair of sterile lemmas(also called empty glumes)and a pair of rudimentary glumes [1-3]. Sterile lemmas are distinct from lemmas in cellular pattern and morphology.The origin and physiological functions of sterile lemmas remain ambiguous.The prevalent hypothesis is that a rice spikelet is derived from an ancestral structure that contains three florets, and that the sterile lemmas are remnants of the two reduced florets that have lost all of their inner floral organs [4]. Despite of some recent progresses [4-6], the genetic control of sterile lemma development is not fully understood.

LONG STERILELEMMA1 (G1)/ELONGATEDEMPTY GLUME(ELE)and PANICLE PHYTOMER2(PAP2)/OsMADS34 are reported to determine the identities of empty glumes[4,5,7-12].G1/ELE encodes a DUF640 domain protein and mutations in G1/ELE lead to the transformation of sterile lemmas to lemma-like organs [4,8]. PAP2/OsMADS34, a rice SEPALLATA-like (SEP-like) MADS-box gene, is another key regulator for sterile lemma identity. Disruption of PAP2/OsMADS34 leads to the development of elongated sterile lemma to form leaf-like or lemmalike organs [5,7,9-12]. Another SEP-like MADS-box gene OsMADS1 is also involved in sterile lemma development.Overexpression of OsMADS1 results in the transformation of sterile lemmas into lemma-like organs [13,14]. The defective glume1 (dg1) and aberrant spikelet and panicle1 (asp1) mutants have elongated sterile lemma and rudimentary glumes,indicating that DG1 and ASP1 play a role in the development of the glumes[15,16].EXTRA GLUME1(EG1)encodes a plastidtargeted lipase that participates in JA biosynthesis and that specifies the fate of sterile lemma.Mutations in eg1 lead to the elongation of sterile lemma into a glume-like organ[17,18].

Several APETALA2/ethylene-responsive element-binding protein (AP2/EREBP) transcription factors are also involved in the development of sterile lemma, including MULTI-FLORET SPIKELET1 (MFS1), OsINDETERMINATE SPIKELET1 (OsIDS1),SUPERNUMERARY BRACT (SNB), and FRIZZY PANICLE (FZP)[6,19-23]. In mfs1 and osids1 mutants, the sterile lemmas degenerate into rudimentary glumes [21,23]. Disruption of SNB and FZP leads to production of extra rudimentary glumes with a loss of sterile lemmas [6,20-22]. The reduced sterile lemmas in fzp-12 and fzp-13 null mutants are transformed into organs resembling rudimentary glumes in wild type plants [6]. In contrast,FZP-overexpression lines produce spikelets with elongated sterile lemma [6,19]. It was proposed that the lemma, rudimentary glume,and sterile lemma are homologous organs[6].

The spikelets of strong fzp mutants, such as fzp2, were replaced by reiteration of branches, whereas weak fzp alleles,such as fzp3, form fertile or defective spikelets at the higher order branches [19,20,24]. The reiteration of branches of strong fzp mutants can be suppressed by a weak lax1 mutant.The lax1-1 fzp2 double mutants lacked lateral spikelets as observed in lax1-1, whereas primary branches elongated in a zigzag manner [24]. Rice LAX1 was previously identified as a basic Helix-loop-helix transcriptional factor with roles in axillary meristem (AM) initiation and maintenance [24,25]. In rice, disruption of LAX1 abolishes AM initiation and leads to fewer branches, a phenotype that closely resembles the pinlike inflorescences in Arabidopsis [24,25]. The weak lax1 alleles, such as lax1-1, show defects in axillary meristem initiation, resulting in failure to develop any lateral spikelets.However, lax1-1 develops normal terminal spikelets. Compared to weak lax1 mutants that have normal terminal spikelets, strong lax1 alleles show a range of abnormalities in the terminal spikelets,including incomplete florets or even reiteration of extra glumes [24-26], suggesting that LAX1 also functions in determining rice floral organs.

Here, we report the characterization of new lax1 (lax1-c)mutants generated by CRISPR/Cas9 gene editing technology and the lax1-RNAi knock down lines. Both the lax1-c and lax1-RNAi lines developed spikelets with elongated sterile lemmas.We also identified a new fzp weak allele,fzp-14,in a genetic screen for enhancers of lax1-1. The fzp-14 mutant also developed spikelets with elongated sterile lemmas. We show that fzp-14 is haploid-insufficient in the lax1-1 background in terms of sterile lemma development and the elongated sterile lemma in lax1-1 fzp-14+/-resembles those observed in lax1-c, lax1-RNAi lines, and fzp-14. We propose that LAX1 and FZP play synergistic roles in rice sterile lemma development.

2. Materials and methods

2.1. Plant materials

Oryza sativa Geng (japonica) variety Shiokari and Zhonghua11(abbreviated as ZH11) were used as wild-type. fzp-14 was screened from an EMS mutagenized population of lax1-1 (in shiokari background).Two CRISPR/Cas9 lines lax1-c1 and lax1-c2 in ZH11 background were generated for genetic analysis.All of the plants used in this study were grown in a paddy field at Huazhong Agricultural University (HZAU, Wuhan, China)under natural conditions in summer or in the greenhouse with a 12 h light and 12 h dark cycle.

2.2. DNA constructs and rice transformation

For genetic complementation tests of fzp-14,a 9.1 kb genomic DNA fragment covering the FZP open reading frame(ORF)and the related 5′ and 3′-flanking sequences, was amplified from the bacterial artificial chromosome(BAC,ID:a0034A07)of the Nipponbare BAC library that was available at HZAU.The DNA fragment was cloned into the pCAMBIA2300 between EcoR I and Pst I sites by Gibson assembly [27]. The constructs of CRISPR/Cas9 was made as previously described [28]. The U6-gRNA unit for single target gene was cloned into the Kpn I site in pCXUN-Cas9 by Gibson assembly [27]. Transgenic lines were produced by Agrobacterium tumefaciens (EHA105)-mediated callus transformation [29].

2.3. SEM analysis

Florets of fzp-14 and young floret of WT,fzp-14,lax1-RNAi were sampled and fixed overnight in 2.5% glutaraldehyde at 4 °C.Samples were dehydrated through a series of ethanol solutions(50%to 100%)and then were critical-point dried,coated with platinum powder, and observed under the scanning electron microscope.

2.4. Statistical analyses

Three largest panicles of each plant of WT, lax1-1, lax1-1 fzp-14, lax1-1 fzp-14+/-were obtained for measuring panicle length, the number of primary or secondary branches, and seeds per panicle. The data were presented as mean ± SD(Standard deviation). The percentages of different types of spikelets of fzp-14 and lax1-1 fzp-14+/-were calculated from the panicles analyzed above. The percentages of different types of spikelets of lax1-c and lax1-RNAi were calculated from panicles of lax1-c and lax1-RNAi in ZH11 background.

2.5. RNA extraction and qRT-PCR

Total RNA was isolated from young panicles (＜5 mm) of lax1-RNAi and wild type using TRIzol reagent (Thermo Fisher Scientific, Waltham, USA). First-strand cDNA was synthesized from 3 μg total RNA using Superscript III reverse transcriptase Kit(Thermo Fisher Scientific, Waltham, USA) according to the manufacturer's protocol.The quantitative real-time PCR experiments were performed using the FastStart Universal SYBR Green Master(ROX)kit(Roche,Basel,Switzerland)and the StepOnePlus Real-Time PCR System (Applied Biosystems, Foster City, USA).The results were analyzed by the comparative Ct method and normalized by the UBIQUITIN expression level with QuantStudio Real-Time PCR Software v1.3 (Applied Biosystems, Foster City,USA).

2.6. Accession numbers

Sequence data from this article can be found in the Rice Genome Annotation Project Database (MSU) under the following accession numbers: LAX1 (LOC_Os01g61480), and FZP(LOC_Os07g47330).

3. Results

3.1. LAX1 is involved in specifying rice sterile lemmas

To investigate the roles of LAX1 in rice spikelet development,we generated new alleles of lax1(lax1-c,c stands for CRISPR)in Zhonghua 11 (ZH11), a Geng variety (Fig. 1-A). Two independent lax1-c alleles showed the expected inflorescence phenotypes similar to those previously reported in lax1-1(Fig.1-A,B,Fig. S1) [25]. Both lax1-c1 and lax1-c2, which contained a 2-base-pair (bp) and 1 bp deletion, respectively (Fig. 1-A),showed the same inflorescence morphology and did not develop any lateral spikelets (Fig. 1-B, Fig. S1). Detailed analysis of lax1-c1 morphology was shown in the Fig. 1.About half(n = 306)of the lax1-c spikelets were very similar to those of lax1-1. However, many spikelets in lax1-c plants had enhanced developmental defects than lax1-1(Fig. 1-C-J). We classified the abnormal lax1-c spikelets into three types:Type I(20.06%,n = 306)spikelets had elongated sterile lemma;type II(7.77%,n = 306)spikelets developed two opposite lemmas that enclosed an elongated lodicule and had indeterminate number of stamens;type III(17.8%,n = 306)displayed degenerated palea and reduced development of inner organs. We also analyzed the inflorescence phenotypes of lax1-RNAi lines,which were described previously [30]. The lax1-RNAi line showed the typical lax1-1 mutant phenotypes (Fig. S2-A, B).The lax1-RNAi plants also developed spikelets with elongated sterile lemmas (30.6%, n = 281) (Fig. S2-C-L). Our results suggested that LAX1 plays a critical role in developing the spikelet and floral organ identity.

In wild-type spikelets, the sterile lemmas had regularly arranged flat cells that rarely exhibited trichomes,resulting in an overall smooth appearance (Fig. 1-K, L, Fig. S2-D, E). The lemma and palea had rugose appearance with parallelly arranged cells with a convex structure called tubercles. Trichomes were also formed on the abaxial surface (Fig. 1-M, N and Fig. S2-F, G). The lemma and palea of spikelets in lax-c1 and lax1-RNAi were similar to that of WT (Fig. 1-Q, R, Fig. S2-K, L). While the elongated sterile lemmas of spikelets of lax-c1 and lax1-RNAi showed rough appearance that were highly similar to that of the lemma and palea (Fig. 1-O, P and Fig. S2-I, J). Our results indicated that epidermal cells of the sterile lemmas in the lax1-RNAi plants had an identity similar to those in lemma/palea, suggesting that LAX1 has a role in specifying rice sterile lemma.

3.2. Identification of a genetic enhancer of lax1-1

To further investigate the mechanisms by which LAX1 regulates inflorescence and flower development, we conducted a forward genetic screen for mutations that could enhance or suppress the phenotypes of lax1-1. One of the mutants, enhancer of lax1-1 (el1), greatly enhanced the lax1-1 phenotypes. The lax1-1 el1 double mutants completely abolished the development of flowers, resulted in a ‘pin' like panicle, while lax1-1 was able to produce normal terminal spikelets [24,25] (Fig. 2-A, C, D). The panicle length and the number of primary branches and secondary branches of the lax1-1 el1 double mutants were similar to those of lax1-1(Fig.2-A, E, F). The lax1-1 mutant had fertile terminal spikelets similar to those of wild type (WT), except that lax1-1 had elongated pedicel(Fig.2-B,C).The panicle of lax1-1 el1 lost all spikelets but had bract-like organs and formed elongated branches in a zigzag fashion (Fig. 2-D). Among 184 progenies from selfing lax1-1 EL1+/-, 44 plants displayed the ‘pin' like panicle, suggesting that el1 contained a single recessive mutation.

3.3. Enhancer of lax1-1 (el1) is a weak fzp allele

To map the EL1 gene,we crossed lax1 EL1+/-plants to Dular,a Xian (indica) variety with a wide-compatibility. We harvested seeds from individual F1plants and analyzed the F2population that segregated plants with the‘pin'like panicle.Among the 5660 F2plants analyzed, 3317 plants behaved like wild type. The number of progenies with lax1-1 and ‘pin' like phenotype was 963 and 418, respectively. Moreover, 962 plants exhibited a novel phenotype, whose panicles had dramatically increased number of secondary branches and did not produce any lateral spikelets (Fig. S3-A). The segregation ratios were 9:3:3:1 for WT:lax1-1:novel plants:lax1-1 el1,which matched the expected Mendelian segregation. The novel phenotype was conferred by the single el1 mutation.

Occasionally, strong el1 phenotypes were observed in the mapping population, of which the panicle continuously formed branches instead of florets, which resembled the strong frizzy panicle (fzp) mutants as previously reported (Fig.S3-B, C) [20,24]. To determine whether EL1 is FZP, we used DNA markers to map the location of EL1. We found that el1 was linked to the SSR maker RM1306, which was close to FZP(Fig.S4).DNA sequencing analysis of el1 revealed that a C to T substitution at the 184th base pair in FZP coding region occurred in the el1 mutant, resulting in the substitution of Arginine (R62) with a Tryptophan (W62) (Fig. 3-A). The mutation was located adjacent to the N terminal of the ERF DNA binding domain, which probably influenced the binding activity of this domain to the downstream regulated genes(Fig. 3-A). We transformed a 9.1-kb genomic DNA fragment,which covers the FZP open reading frame (ORF) and the related 5′ and 3′ flanking sequences, into the el1 mutant to further confirm whether EL1 is FZP.Indeed,the genomic DNA fragment was able to rescue the defects of el1(Fig.3-B-D).

Fig.1-LAX1 is involved in sterile lemma development.(A)A schematic structure of LAX1.The target sequence of CRISPR/Cas9 for editing LAX1 is indicated.The PAM site(CGG)is highlighted in red and the Cas9 cleavage site is marked with an arrow.The edited sequence of the two independent alleles of lax1-c in the ZH11 background are aligned with that of WT.Note that both lax1-c1 and lax1-c2 have a small deletion,which presumably causes frame-shift.(B)Inflorescences of Wild type(WT,left)and lax1-c1(right).((C)The overall morphology of a WT spikelet.(D)The inner organs of a WT spikelet.(E-J)Three types of lax1-c1 spikelets. The overview of the spikelets is shown in(E, G and I)and the inner parts are shown in(F, H and J).The Roman numerals in parentheses indicate the different types of spikelets.(-N)Scanning electron microscopy(SEM)graphs of the outer surface of a spikelet(K),sterile lemma(L),lemma(M),and palea(N)of a WT spikelet.(O-R)Scanning electron microscopy(SEM)analysis of the outer surface of a spikelet(O),elongated sterile lemma(P),lemma(Q),and palea(R)of a spikelet from the lax1-c1 plant.Le,lemma;Pa,palea;SL,sterile lemma;LSL,elongated sterile lemma.Bar = 1 mm in(C-J),Bar = 500 μm in(K and O),and Bar = 100 μm in(L-N and P-R).

Fig.2- Identification of el1 as a genetic enhancer of lax1-1.(A)The lax1-1 el1 double mutants fail to make flowers.Inflorescences of wild type(WT,left),lax1-1(middle),and lax1-1 el1 double mutants(right)are shown.Note that the double mutants developed pin-like inflorescences.(B-D)Floral morphology of WT(B),lax1-1(C),and lax1-1 el1(D).Note that the double mutants did not have a normal floral structure. Rather,the double mutants only had an enlarged bract.The red arrowhead in D indicates the absence of spikelets in the lax1-1 el1 inflorescences.Rudimental glume/glume like organs and elongated spikelet pedicels are indicated with an arrow and arrow head,respectively.(E and F)The effects of el1 on inflorescence branching.The analysis of the primary branch number(E)and the secondary branch number(F)of lax1-1(n = 40)and lax1-1 el1(n = 27) are shown.The data are presented as mean ± standard deviation.Asterisks (**and*)indicate that the differences between lax1-1 el1 and lax1-1 are statistically significant at P ＜0.001 and P ＜0.05,respectively,according to the Student's t-test.Le, lemma;Pa, palea;SL,sterile lemma.Bar = 2 cm in(A),Bar = 1 mm in (B-D).

In summary,our results demonstrated that EL1 is FZP.The weak allele of fzp identified in this work was designated as fzp-14, following the names of previously reported fzp alleles[6,19,20,31].

3.4. Abnormal morphology of the fzp-14 inflorescence

Fig.3-The el1 phenotypes were caused by a mutation in FZP.(A)A mutation in the FZP gene is responsible for el1 phenotypes.The newly identified fzp allele contained a nucleotide substitution that caused an Arginine to Tryptophan change.(B) A schematic diagram of the complementation vector with the FZP genome fragment.The green and red lines refer to the FZP genome fragment and border sequences of the vector,respectively.F1,R1,F2, and R2 are the primers for genotyping.(C)The genotyping result of progenies segregated from the complemented fzp-14 lines.1-12 are individual plants that have WT like(1-6)phenotype or el1(7-12)phenotype.CK is the positive control,which is the vector with the FZP genome fragment.(D)Inflorescences of fzp-14/pFZP-FZP(left)and fzp-14(right).The defects in fzp-14 inflorescences were completely rescued by the introduction of the FZP genomic fragment.The fzp-14/pFZP-FZP plants developed wild type like spikelet.

We backcrossed the lax1-1 fzp-14+/-to Shiokari, where the lax1-1 was generated. We conducted phenotypic characterization of the F2plants. There was no obvious difference on primary branch number between fzp-14 and WT,even though the panicle of fzp-14 was slightly longer than WT (Fig. 4-A, T,U).However,fzp-14 showed dramatically increased secondary branches compared to WT (Fig. 4-A, V), which were also seen in the lower FZP expression lines described previously[32,33].The mutant fzp-14 even developed tertiary or high order branches which were rarely seen in WT (Fig. 4-A). The development of lateral spikelet was completely abolished in fzp-14,a phenotype that was also observed in lax1-1.The fzp-14 plants had slightly reduced number of terminal spikelets than WT, but had diverse developmental defects (Fig. 4-B-K,W). All of these terminal spikelets lacked the rudimentary glumes. We classified fzp-14 spikelets into four types: Type I(36.46%, n = 1670) had florets with one or two normal sterile lemmas; type II (30.72%, n = 1670) developed one elongated sterile lemma; type III (7.60%, n = 1670) displayed two elongated sterile lemmas; type IV (25.21%, n = 1670) produced degenerated paleas(Fig.4-D-K).All of the fzp-14 florets except for type IV were able to produce seeds. The elongated sterile lemma in type II and III spikelets also acquired lemma/palea characteristics (Fig. 4-L-S). These results indicated that FZP played a crucial role in controlling rice inflorescence architecture by suppressing the secondary branch development and by specifying floral organ identities,especially in determining sterile lemma.

3.5. LAX1 and FZP work synergistically in sterile lemma development

Intriguingly, fzp-14 is haploid-insufficient in the lax1-1 background in terms of sterile lemma development.The lax1-1 fzp-14+/-plants showed enhanced phenotypes compared to lax1-1 in terms of sterile lemma development. The overall phenotype of lax1-1 fzp-14+/-was similar to that of lax1-1, but they differed significantly in spikelets (Fig. 5-A-D, Table 1). The lax1-1 fzp-14+/-spikelets could be divided into three types:type I (33.64%, n = 437) resembled lax1-1 spikelets, type II(16.48%,n = 437)showed elongated sterile lemma,and type III(49.88%, n = 437) had degenerated palea (Fig. 5-B-D). Sterile lemmas and hull in lax1-1 were similar to those in WT(Fig.4-L-O, Fig. 5-E-H). The lemma and palea of spikelets in lax1-1 fzp-14+/-were also similar to those in lax1-1 and WT(Fig.5-K,L).However,a significant number of spikelets in lax1-1 fzp-14+/-(type II spikelets)developed elongated sterile lemmas,which showed rough appearance that was highly similar to that of the lemma and palea(Fig.5-J-L).Similar phenotypes were also observed in lax1-c1,lax1-RNAi and fzp-14(Fig.1-O,P,Fig.4-P,Q,Fig. S2-I, J). These results indicated that fzp-14 enhanced the defects of lax1-1 in a dose-dependent manner. LAX1 and FZP worked synergistically in sterile lemma development.However,we did not observe the dose-dependency in lax1-1+/-fzp-14. The spikelets phenotype of fzp-14 was not enhanced when combined with lax1+/-.

Fig.4- Phenotypic analysis of WT and fzp-14 flowers.(A)A comparison of WT inflorescence(left)with that of fzp-14(right).(B-C)A WT spikelet.(B)shows the overall morphology of a WT spikelet and(C)displays the inner organs of a WT spikelet.(DK)Four types of fzp-14 spikelets.The overview of the spikelets is shown in(D,F,H and J)and the inner parts are shown in(E,G,I and K).The Roman numerals in parentheses indicate the different types of spikelets.(L-O)Scanning electron microscopy(SEM)graphs of the outer surface of a spikelet(L),sterile lemma(M),lemma(N),and palea(O)of a WT spikelet.(P-S)SEM analysis of the outer surface of a spikelet(P), elongated sterile lemma(Q),lemma(R),and palea(S)of a type III spikelet from the fzp-14 plant.(T-W)The effects of fzp-14 on inflorescence architecture.The analysis of panicle length(T),primary branch number(U),secondary branch number(V)and spikelet number(W)of WT(n = 51),and fzp-14(n = 37)are shown.The data are presented as mean ± standard deviation.Asterisks(**and*)indicate that the differences between fzp-14 and its corresponding wild type are statistically significant at P ＜0.001 and P ＜0.05,respectively,according to the Student's t-test.Rudimental glume/glume like organ is indicated with an arrow.Le,lemma;Pa,palea;SL,sterile lemma;LSL,elongated sterile lemma.Bar = 2 cm in(A),Bar =1 mm in(B-K),Bar = 500 μm in(L and P),Bar = 100 μm in(M-O and Q-S).

Fig.5-Phenotypic analysis of lax1-1 and lax1-1 fzp-14+/-flowers.(A)A comparison of inflorescences of lax1-1(left)and lax1-1 fzp-14+/-(right).(B-D)The overview of three types of lax1 fzp-14+/-flowers.Type I spikelet has similar morphology to that of lax1-1.The Roman numerals in parentheses indicate the different types of spikelets.(E-H)SEM graphs of the outer surface of a spikelet(E),sterile lemma(F),lemma(G)and palea(H)of a WT spikelet.(I-L)SEM analysis of the outer surface of a spikelet(I),elongated sterile lemma(J),lemma(K)and palea(L)of a type II spikelet from the lax1-1 fzp-14+/-plant.Rudimental glume/glume like organ and elongated spikelet pedicel are indicated with an arrow and arrow head,respectively.Le,lemma;Pa,palea;SL,sterile lemma;LSL,elongated sterile lemma.Bar = 2 cm in(A),Bar = 1 mm in(B-D),Bar = 500 μm in(E and I),Bar = 100 μm in(F-H and J-L).

4. Discussion

We demonstrated that disruption of either LAX1 or FZP leads to the defects in rice floral organs, especially in the development of sterile lemmas. Our genetic analysis of lax1 and fzp mutants demonstrated that the two genes play synergistic roles in the development of rice sterile lemma.

LAX1 is well known for its roles in axillary meristem initiation and maintenance [24-26]. Our analysis of new lax1 alleles and lax1-RNAi lines(Fig.1-A,E,F,O,P,Fig.S2-I,J)demonstrated that LAX1 also has a function in preventing sterile lemmas from elongating. Our lax1-c and lax1-RNAi lines developed elongated sterile lemmas, which actually were similar to normal lemmas morphologically.Interestingly,lax1-1,a relatively weak allele,did not develop the elongated sterile lemmas. The phenotypicdifferences among the lax1 mutants/RNAi lines in sterile lemma development may be caused by differences in genetic backgrounds or strength of the alleles.

Table 1--Phenotypic characterization of lax1-1 FZP+/+and lax1-1 fzp-14+/- plants.

The roles of FZP in sterile lemma development are very confusing. The spikelets of fzp-14 had elongated sterile lemmas, which acquired lemma/palea identity (Fig. 4-F-I, P,Q). However, the sterile lemmas in fzp-12 and fzp-13 null mutants are reduced and are transformed into organs resembling rudimentary glumes in wild type plants [6], suggesting that FZP is required for sterile lemma development, but a reduction of FZP activity actually stimulates sterile lemma elongation.More intriguingly,overexpression of FZP also led to the development of spikelets with elongated sterile lemma[6,19], suggesting that a dosage effect of FZP on sterile lemma development.

Our results and previous studies have clearly demonstrated that both LAX1 and FZP play a role in sterile lemma development. A more interesting observation is that fzp-14 is haploid-insufficient in the lax1-1 background (Fig. 5-C, I, J).The lax1-1 fzp-14+/-developed elongated sterile lemmas that were also observed in lax1-c(Fig.1-E,F,O,P),lax1-RNAi(Fig.S2-H-J), and fzp-14 (Fig. 4-F-I, P, Q). Haplo-insufficiency usually suggests the two genes work in the same genetic pathway.Although we still do not understand the molecular mechanisms of LAX1 and FZP, it is clear that LAX1 and FZP synergistically control sterile lemma development.

Acknowledgments

We thank Professor Junko Kyozuka from Tohoku University for kindly providing the lax1-1 mutant and the related wild type variety Shiokari, and Lei Wang from Huazhong Agricultural University for providing the lax1-RNAi line. This work was supported by the National Transgenic Research Program of China (2016ZX08010002).

Appendix A.Supplementary data

Supplementary data for this article can be found online at https://doi.org/10.1016/j.cj.2019.06.006.