Arabidopsis TCP4 transcription factor inhibits high temperature-induced homeotic conversion of ovules

Abnormal high temperature (HT) caused by global warming threatens plant survival and food security, but the effects of HT on plant organ identity are elusive. Here, we show that Class II TEOSINTE BRANCHED 1/CYCLOIDEA/ PCF (TCP) transcription factors redundantly protect ovule identity under HT. The duodecuple tcp2/3/4/5/10/13/17/24/1/12/18/16 (tcpDUO) mutant displays HT-induced ovule conversion into carpelloid structures. Expression of TCP4 in tcpDUO complements the ovule identity conversion. TCP4 interacts with AGAMOUS (AG), SEPALLATA3 (SEP3), and the homeodomain transcription factor BELL1 (BEL1) to strengthen the association of BEL1 with AG-SEP3. The tcpDUO mutant synergistically interacts with bel1 and the ovule identity gene seedstick (STK) mutant stk in tcpDUO bel1 and tcpDUO stk. Our findings reveal the critical roles of Class II TCPs in maintaining ovule identity under HT and shed light on the molecular mechanisms by which ovule identity is determined by the integration of internal factors and environmental temperature.

Global warming and the increased frequency, intensity, and duration of high temperature (HT) severely threaten the worldwide food security, as well as the diversity and distribution of plants and animals 1 .As sessile organisms, plants undergo morphological changes in a process called plant thermomorphogenesis under ambient HT.The changes associated with plant thermomorphogenesis include longer hypocotyls, leaf petioles and roots, as well as early flowering, male sterility and accelerated fruit dehiscence [2][3][4][5][6] .In contrast to plants, animals can move to avoid adverse living conditions such as HT, and they do not undergo morphological changes like plants under HT.However, embryo development in immobile eggs may be highly affected by HT, and this effect of HT is particularly severe in some fish and reptile species 7 .It is well established that differences in hatching temperature lead to the homeotic conversion of the sexual organs of embryos 7 .For example, the eggs of the turtle Trachemys scripta develop into males at 26 °C hatching temperature, but they develop into females at 32 °C, due to sexual homeotic transformation mediated by temperature-sensitive epigenetic regulation 7 .Although plant thermomorphogenesis has been studied extensively in recent years, the effects of HT on the homeotic transformation of reproductive organs in plants remain largely unknown.
Ovules are the site of double fertilization and the precursors of seeds in angiosperms, and they are essential for plant reproduction and dispersal 8,9 .Ovules are initiated within the carpels, which are the fourth-whorl organs of flowers 10 .An angiosperm ovule typically consists of a proximal funiculus connecting to the placenta at the adaxial margin of the carpel and a distal nucellus covered by the inner and outer integuments initiated from the central chalaza 9 .The nucellus is the site at which female gametophytes are formed by megasporogenesis and megagametogenesis.In Arabidopsis, megasporogenesis begins with a subepidermal cell specified to develop into a megaspore mother cell (MMC) in the distal nucellus.The MMC undergoes meiosis to give rise to the functional megaspore (FM) which further develops into the female gametophyte 9 .The funiculus coordinately elongates, and the inner and outer integuments grow asymmetrically to envelop the nucellus, leaving a micropyle for pollen tube entry to complete the double fertilization 9 .
The specification and coordinated development of the parts of the ovule are tightly regulated and are very critical for the formation of a functional female gametophyte and thus plant reproduction.According to the floral quartet model, the ovules are specified by MADS-box quartets consisting of SEPALLATA (SEP) proteins, AGA-MOUS (AG), and AG homologs such as SEEDSTICK (STK) and SHAT-TERPROOF1 (SHP1) or SHP2, while the carpels outside ovules are specified by quartet complexes comprising only SEPs and AG 11,12 .The co-existence of SEPs, AG, STK and SHPs suggests that the AG-SEPs-STK/SHPs quartets specifying ovule identity co-exist in the ovules with AG-SEPs quartets specifying carpel identity, suggesting that regulatory mechanisms exert tight control over the activity of AG-SEPs-STK/SHPs quartets and AG-SEPs quartets to promote ovule identity, but not carpel identity, in ovules.Genetic analysis has shown that the activity of AG-SEPs-STK/SHPs quartets and AG-SEPs quartets, and the balance between them, are essential for the specification of ovules [13][14][15] .It has been shown that ag single mutants, sep1 sep2 sep3 triple mutants and sep1 sep2 sep3 sep4 quadruple mutants do not form carpels or ovules because they lack both types of quartets [16][17][18] .Overexpression of AG causes the transformation of ovules into carpelloid structures, possibly due to increased activity of AG-SEPs quartets in ovules 19,20 .In the haploid-insufficient SEP1/sep1 sep2 sep3 triple mutants, ovules are also converted into carpelloid structures, suggesting that SEPs might be inclined to form AG-SEPs quartets instead of AG-SEPs-STK/SHPs quartets when SEP proteins are deficient in ovules 21 .STK is specifically expressed in ovules 22,23 .Disruption of STK causes longer ovule funiculi 22 , while shp1 shp2 double mutants generate normal ovules 24 .However, stk shp1 shp2 triple mutants produce carpelloid structures instead of ovules owing to a lack of AG-SEPs-STK/SHPs quartets 22 .Few factors have been identified to control the balance between the activity of AG-SEPs-STK/SHPs quartets and AG-SEPs quartets in ovules.The homeodomain transcription factor BELL1 (BEL1) has been demonstrated to regulate the identity of ovules in a temperature-dependent manner 25 .The bel1 produced ovules that could be converted into carpelloid structures at the normal growth temperature (22 °C), but the phenotype is less severe at the low temperature (16 °C) 25 .BEL1 interferes with the function of AG-SEPs quartets by directly interacting with the AG-SEPs dimer 13,14 .However, the molecular mechanisms by which ovule identity are affected by temperature, as well as those by which BEL1 stabilizes ovule identity against the effects of temperature differences, are still elusive.
TEOSINTE BRANCHED 1/CYCLOIDEA/PCF (TCP) proteins belong to a plant-specific transcription factor family that is present in all plants 26 .These proteins contain a conserved TCP domain that is responsible for interacting with DNA or proteins 27 .The TCP family proteins in Arabidopsis are classified into the Class I and Class II subfamilies according to sequence differences in the TCP domain 26,28,29 .The Class I subfamily consists of thirteen TCP members, while the Class II subfamily contains eleven TCPs 26,28,29 .The Class II subfamily is further divided into three TB1/CYC-like TCPs (TCP1, TCP12, and TCP18) and eight CINCINNATA (CIN)-like TCPs (TCP2, TCP3, TCP4, TCP5, TCP10, TCP13, TCP17 and TCP24) 26,27 .Class II TCPs have highly redundant and dosage-dependent functions in the control of plant development 29 .They play critical roles in plant developmental plasticity and diverse processes, including leaf development [30][31][32][33][34] , shoot branching 35 , flower morphogenesis 36 , trichome formation 37,38 , thermomorphogenesis 39,40 , and photomorphogenesis 41 .However, Class II TCPs have not yet been shown to play a role in controlling ovule identity.

TCP transcription factors play highly redundant and critical roles during ovule development
We previously reported that overexpression of TCPs disrupts MMC production during ovule development 42 .However, the tcp3/4/5/10 quadruple mutant displayed only weak ovule developmental phenotypes; therefore, we hypothesized that the high redundancy of Class II TCP transcription factors could mask the important roles of TCPs in ovules.We first investigated ovule development in a septuple tcp2/3/4/ 5/10/13/17 (tcpSEP) mutant generated by genetic crosses using T-DNA insertion mutants as described previously 38 .The tcpSEP mutants generated ovules and seeds with no significant differences from those of wild-type control plants, except that the siliques of tcpSEP were shorter (Fig. 1a), and the seeds were more crowded than that in wild-type control (Fig. 1b, c).This phenotype resembled that observed in tcp3/4/ 5/10 42 .
The eleven Class II TCP genes and TCP16, which is classified as a Class I TCP member but contains a TCP domain more similar to that of Class II TCP proteins 43 , are rather evenly distributed in all five chromosomes (Supplementary Fig. 1a).In order to finalize the possible roles of the eleven Class II TCP genes and TCP16 in controlling ovule development, we decided to generate a tcp multiple mutant in which all these TCPs were mutated.We identified SALK_023116, GABI_655C10 and SALK_091920, with a T-DNA insertion in the exon of tcp12, tcp16, and tcp18, respectively, and we generated tcp1 and tcp24 mutants carrying one-bp insertion in the appropriate exon using CRISPR/Cas9 technology (Supplementary Fig. 1b).We finally generated a duodecuple null mutant tcp2/3/4/5/ 10/13/17/24/1/12/18/16 (tcpDUO) by genetic crosses.The tcpDUO generated waving leaves and displayed late flowering as that observed in tcpSEP when compared to the wild-type control (Supplementary Fig. 2a, b).However, the length of pistils in tcpDUO was significantly shorter than that in wild-type control (Supplementary Fig. 2c, d).To determine whether the pistils of tcpDUO could affect the growth of pollen tube, we pollinated the pistils of tcpDUO or wild-type control with wild-type pollen.The staining with aniline blue showed that pollen germinated on the stigma and pollen tubes grew through styles in pistils from both tcpDUO and wild-type control (Supplementary Fig. 2e, f).After pollination, the siliques of tcpDUO mutant did not fully elongated and were even shorter than that of wild-type control, and displayed low fertility (about 18.01% of 1259 ovules from 30 siliques) (Fig. 1a-d, i, j).To determine whether defects in male or female gametophytes were the cause of the observed low fertility, we first performed a reciprocal cross between tcpDUO and the wild-type control (Table 1).When pollen from tcpDUO or wild-type control plants was used to pollinate the pistils of wild-type plants, more than 98% of ovules developed into seeds, while only about 18% ovules from 10 siliques of tcpDUO plants formed seeds following pollination with pollen from tcpDUO or wild-type control plants (Table 1), suggesting that wild-type pollen did not rescue infertility in the tcpDUO siliques.These results indicated that the female gametophytes of tcpDUO, but not the male gametophytes, were defective, and maternally derived tcp mutations caused the aborted ovules or seeds in tcpDUO plants.
To confirm that TCP transcription factors play key roles in the development of ovules and maternal fertility, we selected TCP4 as a representative gene and generated TCP4pro-TCP4-GFP, in which TCP4 (fused with GFP) was driven by its own promoter for complementation.
The expression of TCP4-GFP in tcpDUO by transforming TCP4pro-TCP4-GFP significantly increased the length of siliques and decreased the ratio of aborted ovules or seeds (Fig. 1e-h, j).These results indicate that TCP4 and the other members of the Class II TCP subfamily transcription factors perform essential and redundant functions in the development of ovules and seeds.

TCP transcription factors regulate the growth of ovule funiculi, integuments, and megagametogenesis
To reveal further details regarding ovule defects in tcpDUO, we first observed mature ovules using scanning electron microscopy (SEM).The SEM results showed that the funiculi from tcpDUO ovules contained more cells and were clearly longer than those from wild-type control ovules (Fig. 2a-d).Furthermore, while the outer integument grew over the inner integument and nucellus in the mature wild-type ovule, leaving a small, clear opening, the outer and inner integuments from the tcpDUO ovule were observed to grow excessively (Fig. 2a-g).We observed 54 ovules produced excessive growth in the outer and/or inner integuments from 57 ovules of tcpDUO.Among the 54 abnormal ovules, 46 (85.2%) ovules exhibited excessive growth in both the outer  and inner integuments (Fig. 2f), while 8 (14.8%) ovules showed only excessive growth in inner integuments, causing defective ovules with a large micropyle and an outer integument that failed to cover portions of the inner integument (Fig. 2c, d, g).
We then observed megasporogenesis and megagametogenesis during the development of tcpDUO ovules.In ovules from tcpDUO or wild-type plants, a hypodermal cell was differentiated into a megasporocyte, which further produced an FM and three degenerated megaspores after meiosis at the tip of the nucellus, suggesting that megasporogenesis was not significantly affected in tcpDUO (Fig. 2h-k).During megagametogenesis, ovule autofluorescence showed that a clear FM was formed in the ovules from both tcpDUO and wild-type plants at the FG1 stage (Supplementary Fig. 3a, b), as demonstrated by observations using a microscope with differential interference contrast (DIC) (Fig. 2j, k).However, at the FG2 stage, two haploid nuclei were formed by the first round of mitosis (Supplementary Fig. 3c, d), but they were immediately degenerated in most tcpDUO ovules at the FG3 stage (Supplementary Fig. 3e, f).Four haploid nuclei were obviously observed after the second round of mitosis in wild-type ovules (Supplementary Fig. 3g), whereas more than eighty percent (84.3%) of tcpDUO ovules did not undergo the second and third rounds of mitosis, leading to the presence of an abnormal nucellus without haploid nuclei (Supplementary Fig. 3h).Less than twenty percent (9.3% + 6.2%) of tcpDUO ovules contained a nucellus with nuclei in different arrangements (Supplementary Fig. 3i, j), which was consistent with the ratio of aborted seeds in tcpDUO plants (Fig. 1j).These results suggest that TCPs play pivotal roles during ovule development.

TCP transcription factors inhibit the homeotic conversion of ovules into carpelloid structures under HT
During observation of the ovule phenotype of the tcpDUO mutant, we found that ovules were converted into carpelloid structures with papillae in a very low ratio (0.2%, n = 1259 ovules) (Fig. 3a, b), suggesting that TCP transcription factors might play roles in the determination of ovule identity.Previous report indicated that the homeotic conversion ratio of ovules into carpelloid structures in bel1 mutants was affected by temperature 25 .Considering that TCPs participate in plant thermomorphogenesis 39,40 , we treated tcpDUO and wild-type control plants with HT (28 °C).We found that almost all of the ovules in the tcpDUO pistils were converted into carpelloid structures (Fig. 3d), while no ovule identity conversion was observed in the wild-type control under 28 °C treatment (Fig. 3c).The expression of TCP4-GFP by transforming TCP4pro-TCP4-GFP into tcpDUO rescued the identity conversion in tcpDUO under 28 °C treatment (Fig. 3e), indicating that TCP4 plays a critical role in maintaining ovule identity under HT.The SEM results further showed that the integuments of wild type ovules were rather normal under 28 °C (Fig. 3f).However, the integuments of tcpDUO grew excessively, and clear papilla cells covered the top of the inner and/or outer integuments (Fig. 3g-i).In some carpelloid structures, secondary ovules or carpelloid organs were observed (Fig. 3i).To investigate the developmental stage at which the ovules of tcpDUO begin to be converted into carpelloid structures, we observed the ovules from the flower developmental stage 9 to stage 12 of tcpDUO and wild type control under 28 °C.The results showed that all the ovules from wild type had no homeotic conversion (Fig. 3j-n).The early ovules of flowers at stage 9 and 10 were also rather normal in tcpDUO (Fig. 3o-q).However, the ovules at flower developmental stage 11 began to be homeotically converted into carpelloid structures (Fig. 3r).Clear carpelloid structures were observed in the flowers at stage 11 and 12 in tcpDUO under 28°C (Fig. 3r, s).These results indicate that TCP transcription factors play pivotal roles in maintaining the stability of ovule identity under HT.
To determine the contribution and redundancy of TCPs in maintaining ovule identity under HT, we crossed the tcp duodecuple mutant (tcp2/3/4/5/10/13/17/24/1/12/18/16) with the septuple mutant (tcp2/3/4/5/10/13/17) to generate the different combinations of tcp  We treated these tcp multiple mutants with 28 °C to observe the phenotype of homeotical conversion.The results showed that the disruption of all the CIN-like TCPs did not cause the conversion of ovule identity (Supplementary Fig. 4a, b).However, the compromise of all the Class II TCPs led to all the ovules being converted into carpelloid structures (Supplementary Fig. 4g, h).This indicated that CYC-like TCPs played critical and redundant roles with CIN-like TCPs in protecting the ovule identity under HT, and TCP16 was dispensable in this process.No ovule conversion events were observed in the nonuple mutant and the decuple mutant (Supplementary Fig. 4c-f), indicating that TCP24 made an important contribution in maintaining the ovule identity under HT.

TCP transcription factors directly repressed the expression of carpel-related genes to protect ovule identity
Although the effects of HT on the growth of hypocotyls, petioles, roots, and male organs have been studied extensively 1 , the mechanisms underlying the effects of HT on female organs, including the stability of ovule identity, remain largely unknown.To reveal the underlying mechanisms by which TCPs control ovule development and identity under normal temperature and HT, we emasculated the flowers of tcpDUO and wild-type control plants under normal temperature or HT, and pistils were collected 2 days after emasculation (DAE).These materials were used to perform RNA-seq analysis with three replicates.The principal component analysis (PCA) showed that the results from the three replicates from each sample were similar (Supplementary Fig. 5a).We first analyzed the expression level of 24 TCP family genes in wild-type pistils under normal temperature and HT.The RNA-seq results showed that 22 TCPs were expressed in wildtype pistils.TCP16 was expressed at a very low level in wild-type pistils, and TCP9 and TCP1 were not found to be expressed (Supplementary Fig. 6a, b).To determine the effects of HT on the expression level of TCPs, we analyzed the expression fold change of TCP genes in wild type treated with 22 °C Vs 28 °C (Supplementary Fig. 6c).The results showed that the gene expression was not changed more than 2 fold in the TCPs under HT (false discovery rate < 0.01; fold-change ≥ 2.0 or ≤ −2.0).We collected mature ovules from pistils of wild-type plants treated with 22°C or 28°C to test the expression of Class II TCPs and TCP16 using reverse transcription quantitative PCR (RT-qPCR).The results showed that almost all these TCPs except TCP1 and TCP17 were expressed in mature ovules (Supplementary Fig. 7a).We further cloned a 3043-bp-bp promoter of TCP1, a 2466-bp promoter of TCP12, a 2428bp promoter of TCP18 and a 2571-bp promoter of TCP24.We generate TCP1pro-GUS, TCP12pro-GUS, TCP18pro-GUS, and TCP24pro-GUS constructs in which these promoters were used to drive GUS reporter gene, respectively.GUS staining showed that the expression of TCP1 was hardly detected (Supplementary Fig. 7b, c), while the expression of TCP12, TCP18 and TCP24 was clearly observed in the early ovules (Supplementary Fig. 7d-j).Staining experiments with the previously reported TCP4pro-TCP4-GUS transgenic line confirmed that TCP4 protein was expressed during ovule development (Supplementary Fig. 7j-m). 38In situ hybridization showed that TCP4 was highly expressed in nucellus and integuments, consistent with the GUS staining of TCP4pro-TCP4-GUS (Supplementary Fig. 7o-r).Next, we mapped the reads of 12 TCPs to their corresponding genomic regions.In contrast to the TCP reads from wild-type plants, which mapped to the whole corresponding TCP genomic sequences, the 9 T-DNA insertion mutated tcp reads from tcpDUO did not, indicating that no complete transcripts for the 9 TCPs were generated in the tcpDUO null mutant (Supplementary Fig. 6b).In addition, no transcripts of TCP1 or TCP16 were found in wild type and tcpDUO, and the transcripts of TCP24 carrying one-bp-insertion were decreased in tcpDUO (Supplementary Figs.1b, 3b).
Because a low proportion of ovules were converted into carpelloid structures in tcpDUO under a low temperature, while almost all ovules showed homeotic conversion under HT, we speculated that the expression level of DEGs related to homeotic organ transformation in tcpDUO might increase in response to HT.To identify the DEGs responsible for homeotic ovule conversion, we first constructed a set of overlapping DEGs that were identified in both WT in comparison with tcpDUO under HT or normal temperature treatment, including 289 up-regulated genes and 578 down-regulated genes (Supplementary Fig. 5e).Cluster analysis revealed that the overlapping DEGs were clustered into 6 groups according to their expression levels in wild-type and tcpDUO plants under 22 °C or 28 °C treatment (Supplementary Data 5, Supplementary Fig. 5f).The DEGs in Cluster 1 and Cluster 2 were up-regulated in tcpDUO in comparison with wild type, but had a lower expression level under 28 °C treatment in comparison with 22 °C treatment, or a similar level under the two temperatures.These characteristics suggest that cluster 1 and cluster 2 are unlikely to mediate the ovule conversion phenotype of tcpDUO under 28 °C treatment.However, the DEGs in Cluster 3 were expressed at higher levels in tcpDUO in comparison with wild type, and the up-regulation of these DEGs was greater in magnitude under HT in comparison with that observed under normal temperature treatment, consistent with the generation of more carpelloid structures in tcpDUO under HT.This finding suggested that Cluster 3 DEGs could contribute to the efficient ovule homeotic conversion phenotype of tcpDUO under HT.The DEGs in Cluster 4-6 were all down-regulated in tcpDUO in comparison with wild type, and the DEGs in Cluster 4 all showed lower expression levels under HT in comparison with normal temperature (Supplementary Fig. 5f).These analyses suggested that the genes in Cluster 3 and 4 might be important for the efficient ovule homeotic conversion phenotype of tcpDUO under HT.Gene ontology (GO) analysis of the DEGs in Cluster 3 revealed enrichment in terms such as floral organ morphogenesis, formation and development, and carpel development (Supplementary Data 6, Fig. 4a, b), while the DEGs in Cluster 4 were enriched in the sexual reproduction pathway term (Supplementary Data 7, Fig. 4c, d).Heatmap analysis showed that many genes related to carpel development, including CRABS CLAW (CRC) (encoding a YABBY transcription factor) 44 , SPATULA (SPT) 45 , HECATE1 (HEC1), HEC2, and HEC3 (encoding basic helix-loop-helix (bHLH) transcription factors) 46 , were highly induced in tcpDUO in comparison with wild type at 22 °C, with even stronger induction under 28 °C treatment (Fig. 4b).In contrast, many genes related to the formation of female gametophytes, including egg cell-specific genes EC1.1, EC1.3, and EC1.5 47 , were severely down-regulated in tcpDUO in comparison with wild type, with even stronger repression under 28 °C treatment (Fig. 4d).In addition, the expression of INNER NO OUTER (INO) key for ovule development was significantly down-regulated in tcpDUO under HT (Supplementary Fig. 8).
Because the most striking phenotype in tcpDUO under HT was the conversion of ovules into carpelloid structures, we collected ovules and carpelloid structures from wild-type and tcpDUO plants at 2 DAE under normal temperature and HT.We used RT-qPCR to measure the relative expression levels of four genes known to be important for carpel development in these tissues.The RT-qPCR results showed that the expression levels of CRC, SPT, HEC1 and HEC2 in tcpDUO were much higher than those of wild-type plants under both 22 °C and 28 °C treatments (Fig. 4e).Indeed, the expression levels of CRC, HEC1 and HEC2 in the ovule-derived carpelloid structures from tcpDUO plants were hundreds of folds higher than those of wild-type plants under HT (Fig. 4e), consistent with the RNA-seq data and the ovule homeotic conversion phenotype of tcpDUO under HT.
We previously identified thousands of TCP4 binding sites using chromatin immunoprecipitation assays with sequencing (ChIP-seq) with 35S-MYC-mTCP4 in which MYC tag fusion with microRNA319 (miRNA319)-resistant mTCP4 was driven by a CaMV 35 S promoter 41 .The comparison of the ChIP-seq data with the DEGs of WT Vs tcpDUO showed that 308 up-regulated and 436 down-regulated genes in WT Vs tcpDUO under normal temperature were bound by TCP4 (Fig. 4f).However, 591 up-regulated and 632 down-regulated genes in WT Vs tcpDUO under HT overlapped with TCP4 binding genes (Fig. 4f), suggesting that HT may affect the expression of genes bound by TCP4.Further analysis showed that TCP4 proteins were obviously enriched in the genomic loci of CRC, SPT, and HEC1 (Fig. 4g).Gel electrophoresis mobility shift assay (EMSA) results indicated that TCP4 bound to the promoter of CRC or SPT (Fig. 4h), suggesting that TCP4 directly interacted with the promoters of CRC and SPT.
Our RNA-seq data also showed that the master regulator WUSCHEL (WUS) was also significantly up-regulated in the pistils of tcpDUO (Supplementary Fig. 9a).RT-qPCR using the ovules as materials confirmed that WUS was highly induced in tcpDUO, and HT further increased the expression of WUS in tcpDUO (Supplementary Fig. 9b), suggesting that TCPs play key roles in the suppression of WUS under normal or high temperature.This is consistent with the indeterminate growth of carpelloid structures in tcpDUO under HT.
These results demonstrate that HT promotes the expression of genes essential for carpel development, including CRC, SPT, and HEC1, and TCP transcription factors stabilize ovule identity by repressing the induction of carpel-related genes by HT.  g ChIP-seq data showed that TCP4 protein was enriched in the promoter regions of CRC, SPT, and HEC1 in vivo.h EMSA analysis indicated that TCP4 protein was directly bound to the promoter of CRC and SPT in vitro.The EMSA experiments were repeated at least three times.The results in a-d were from RNA-seq analysis and the results of (e) were from RT-qPCR analysis.Source data for e and h are provided as a Source Data file.
TCP transcription factors repress the activity of AG-SEP3 complexes by interacting with AG and SEP3 The balance between the simultaneous activity of AG-SEPs-STK/SHPs (specifying ovule identity) and AG-SEPs (specifying carpel identity) in the ovules determines ovule identity 13 .We searched for AG-SEPs-STK/ SHPs quartet genes, including AG, STK, SHP genes, and SEP genes, in our RNA-seq data.The results showed that the expression of STK was up-regulated under normal temperature, but was a little downregulated under HT in pistils (Supplementary Fig. 10a).SHP1 was significantly induced in the pistils of tcpDUO.The expression of SHP2 and SEP genes were not significantly altered in pistils (Supplementary Fig. 10a).We next tested the expression change of STK in ovules by RT-qPCR.The results showed that the treatment of HT did not change the expression of STK in wild type.However, HT significantly repressed STK in tcpDUO (Supplementary Fig. 10b).In situ hybridization assays confirmed that the expression of STK was decreased and was mainly observed in the funiculi in tcpDUO, when compared to that in wild-type control under HT (Supplementary Fig. 10d-f).These data suggest that TCPs are important for maintaining the expression level of STK in ovules under HT, and the decreased expression of STK is consistent with the homeotic conversion in tcpDUO under HT.
Considering that both AG-SEPs quartets and TCP4 directly regulate carpel development genes, including CRC and SPT [48][49][50] , we hypothesized that TCPs could interact with AG-SEPs at the protein level.We performed yeast two-hybrid assays to examine the interaction between TCPs and AG or SEP3.The results showed that both AG and SEP3 each interacted with TCP4 and TCP18 in yeast two-hybrid assays (Fig. 5a, d).Firefly luciferase complementation imaging (LCI) and co-immunoprecipitation (Co-IP) experiments confirmed that TCP4 interacted with AG and SEP3 in vivo (Fig. 5b, e, c, f).To identify the domains in AG and SEP3 responsible for their interactions with TCP4 and TCP18, we separated both AG and SEP3 into three fragments: AG-MI or SEP3-MI (containing the MADS-box and Intervening domain), AG-IK or SEP3-IK (containing the Intervening and Keratin domains), and AG-C or SEP3-C (containing the C-terminus) (Supplementary Fig. 11a, c).Yeast two-hybrid assays indicated that these truncated proteins all interacted with TCP4 and TCP18 (Supplementary Fig. 11b,  d), while no interactions were detected in the negative controls.That is, no interactions between AG and TCP16, SEP3 and TCP16, STK and TCP4, and STK and TCP18 were found (Supplementary Fig. 11b, d).These data suggested that the various domains of AG and SEP3 play roles in their interactions with TCPs.
To further determine the effect of interactions between TCP4 and AG-SEP3 complexes, we cloned the promoters of AG-SEP3 downstream genes CRC and SPT [45][46][47] .We generated the reporters 35Spro-REN-CRCpro-LUC and 35Spro-REN-SPTpro-LUC, in which the firefly Luciferase (LUC) gene was driven by the CRC or SPT promoter, respectively, and the renilla Luciferase (REN) gene was driven by a CaMV 35S promoter as an internal expression control (Fig. 5g).We co-transformed the effectors 35Spro-AG and 35Spro-SEP3, in which AG and SEP3 were driven by a CaMV 35S promoter, and a reporter (35Spro-REN-CRCpro-LUC or 35Spro-REN-SPTpro-LUC) into Arabidopsis protoplasts.The results showed that the AG-SEP3 complex activated the LUC reporter as expected, while 35Spro-mTCP4, in which miRNA319-resistant mTCP4 was driven by a CaMV 35S promoter, repressed LUC expression (Fig. 5h, i).Co-expression of 35Spro-mTCP4 with the AG-SEP3 complex significantly repressed the LUC reporter expression level, suggesting that TCP4 inhibited the activity of AG-SEP3 (Fig. 5h, i).
These results suggest that TCP proteins prevent AG-SEP3 from activating the transcription of carpel-related genes in ovules.

TCP4 promotes the association of BEL1 with AG-SEP3 by directly interacting with BEL1
BEL1 has been reported to repress the temperature-dependent homeotic conversion of ovules into carpelloid structures by interacting with the AG-SEP3 complex 13,14 .Our results showed that TCP4 also interacted with AG-SEP3 to repress its activity (Fig. 5), and the ovules of tcpDUO were converted into carpelloid structures under HT (Fig. 3), resembling those observed in a previous study of bel1 13 .However, the expression of BEL1 was not obviously altered in pistils of tcpDUO under normal temperature or HT (Supplementary Fig. 10a, c), suggesting that TCPs may not regulate BEL1 at the transcriptional level.To test whether TCPs regulate BEL1 at the protein level, we first assessed the interaction between BEL1 and TCP4 using yeast two hybrid assays, revealing that BEL1 clearly interacted with TCP4 (Fig. 6a).We confirmed the association of TCP4 with BEL1 in vivo using LCI and Co-IP experiments (Fig. 6b, c).Since TCP4 interacted with both BEL1 and AG-SEP3, we hypothesized that TCP4 could strengthen the interaction between BEL1 and AG-SEP3.To test this hypothesis, we fused the N-terminus of LUC (nLUC) with AG, while the C-terminus of LUC (cLUC) was fused with BEL1.Strong fluorescence was observed following co-expression of AG-nLUC, 35S-SEP3 and BEL1-cLUC under normal temperature, whereas no fluorescence was produced by the control combinations (Fig. 6d), consistent with previous results showing that BEL1 interacts with the AG-SEP3 complex 13 .When we co-expressed the miRNA319-resistant mTCP4-MYC with AG-nLUC, 35S-SEP3 and BEL1-cLUC, the fluorescence was significantly increased in three replications (Fig. 6d, f and Supplementary Fig. 12a, b), suggesting that TCP4 may reinforce the association of BEL1 and AG-SEP3.Our results indicated that HT is an important environmental factor promoting the homeotic conversion of ovules into carpelloid structures in tcpDUO (Fig. 3).Thus, we speculated that the interaction between BEL1 and AG-SEP3 could be compromised to release the activity of AG-SEP3 under HT.Indeed, the HT treatment significantly weakened the interaction between BEL1 and AG-SEP3 (Fig. 6e, f and Supplementary Fig. 12c, d).
To further reveal the molecular mechanism by which HT weakened the interaction between BEL1 and AG-SEP3, we investigated the stability of BEL1 under HT, because the transcript level of BEL1 was not significantly altered by HT treatment (Supplementary Fig. 10a, c).We generated 35S-BEL1-FLAG-P2A-GFP, in which expression of BEL1-FLAG and GFP (connected by a sequence encoding P2A) was driven by a CaMV 35 S promoter.P2A is a sequence from porcine teschovirus-1.The 2A peptide mediates self-cleavage, and thus results in one mRNA containing a P2A sequence to be translated into two separate proteins 51,52 .Because P2A self-cleavage caused the expressed BEL1-FLAG-P2A-GFP to be cleaved into equal amounts of BEL1-FLAG and GFP, the abundance of GFP served as the internal control for the evaluation of BEL1-FLAG stability.We transiently expressed 35S-BEL1-FLAG-P2A-GFP in tobacco leaves and treated the leaves with HT.The results showed that the abundance of both the uncleaved BEL1-FLAG-P2A-GFP product and BEL1-FLAG cleaved from it decreased after HT treatment, while the abundance of GFP cleaved from the fusion protein was not altered, indicating that BEL1 was unstable under HT (Fig. 6g).To determine whether the stability of BEL1 under HT was dependent on the 26S proteasome, we treated the samples with MG132 (a proteasome inhibitor) and HT.The degradation of BEL1 under HT was highly inhibited by MG132, suggesting that the 26S proteasome was required for the degradation of BEL1 (Fig. 6h).To provide more evidences to determine the stability of BEL1 under HT, we generated 35Spro-BEL1-FLAG in which BEL1 fusion with the sequence encoding FLAG tag was driven by a CaMV 35S promoter.We treated the flowers of 35Spro-BEL1-FLAG transgenic plants with a protein synthesis inhibitor cycloheximide (CHX).The results showed that HT treatment led to the obvious decrease of BEL1, while the proteasome inhibitor MG132 inhibited the degradation (Fig. 6i), further confirming that BEL1 was regulated by the degradation via 26 S proteasome under HT.We further co-expressed miR319-resistant mTCP4 with 35S-BEL1-FLAG-P2A-GFP and found that TCP4 might increase the stability of BEL1 under HT (Fig. 6j), consistent with the results showing that TCP4 promoted the interaction between BEL1 and AG-SEP3.
These results suggest that TCP transcription factors may promote ovule identity by directly interacting with BEL1 to prevent degradation of BEL1 by the 26 S proteasome under HT.

TCP transcription factors synergistically interact with BEL1 and STK
Our biochemical analysis indicated that TCP transcription factors and BEL1 directly interacted with each other.In addition, a previous report and our data showed that TCPs and BEL1 both repressed the activity of the AG-SEP3 complex by directly interacting with AG-SEP3 13 .Disruption of TCPs or BEL1 led to temperature-dependent aberrations in the establishment of ovule identity 25 , suggesting that TCPs and BEL1 may play synergistic roles in maintaining ovule identity.To assess this possibility, we used CRISPR/Cas9 technology to mutate BEL1 in the wild-type and tcpDUO backgrounds.We obtained the bel1-1, bel1-8, bel1-11, and bel1-15 lines, with one-bp-insertion, a 455 bp, 454 bp, or 460 bp deletion, respectively, in the first exon of BEL1 in the wild-type and tcpDUO backgrounds (Fig. 7a).The ovule phenotypes of the mutant lines were observed under normal temperature using SEM.The results showed that the wild-type control generated normal ovules (Fig. 7c), while bel1 mutant produced bell-shaped ovules absent of inner integuments as previously reported (Fig. 7e, f) 25 .Both the single mutant and tcpDUO rarely displayed homeotic ovule conversion into carpelloid structures under normal temperature (Fig. 7d-f), whereas all of the ovules of the tredecuple mutant tcpDUO bel1 turned into carpelloid structures with papilla cells and long funiculi under normal temperature (Fig. 7g-j).In the carpelloid structures of tcpDUO bel1, many ovule-like structures were initiated from the chalaza position (Fig. 7h-j).The integuments of the ovule-like structures continued to grow and were then converted into carpelloid structures with papilla cells (Fig. 7h, i).The carpelloid structures showed an indeterminate growth, because the new carpel-like structures continued to be produced in the inner of the old ones (Fig. 7h-j).The striking phenotypes of tcpDUO bel1 demonstrated that TCP transcription factors play essential roles in determining ovule identity in cooperation with BEL1.The results described above indicated that TCP transcription factors repress the activity of AG-SEPs complexes in ovules, tilting the balance toward the activity of AG-SEPs-STK/SHPs, thus promoting ovule identity.Accordingly, the decreased activity of AG-SEPs-STK/ SHPs in tcpDUO could synergistically interact with the increased activity of AG-SEPs.To test this hypothesis, we generated the tredecuple mutant tcpDUO stk, with 1651 bp and 871 bp deletions between the second and third exons of STK, in the tcpDUO and wild-type backgrounds (Fig. 7b).The SEM observations indicated that the stk mutant formed ovules with longer funiculi as previously reported 20 (Fig. 7k).However, the carpels of the tredecuple mutant tcpDUO stk were full of carpelloid structures instead of ovules (Fig. 7l-n).The carpel-like structures were connected by even longer funiculi (Fig. 7l,  n).In addition, stigmatic papilla cells were clearly observed at the tip of the carpelloid structures (Fig. 7l-n).Nevertheless, no ovule-like structures or secondary carpel-like structures were observed in tcpDUO stk (Fig. 7l-n).
The ovule defective phenotypes of tcpDUO bel1 and tcpDUO stk unequivocally demonstrate that TCP transcription factors play a critical role in maintaining the identity of ovules.

Discussion
In this study, we demonstrated that plant-specific TCP transcription factors play critical and highly redundant roles in the control of ovule development and identity.The generation of a high-order tcpDUO mutant, in which both CIN-and CYC-like TCPs were disrupted, revealed that the two clades of TCPs were redundantly pivotal for ovule development.Our observations of the conversion of ovules into carpelloid structures in tcpDUO under HT indicated that TCPs have highly redundant functions in protecting the stability of ovule identity under HT.We propose a seesaw working model for the function of TCPs in determining ovule development and identity.In wild-type ovules, AG, SEPs, STK and SHPs co-exist in ovules.The ovules are specified by AG-SEPs-STK/SHPs, while the carpels are specified by AG-SEPs.The balance between the two kinds of complexes is important for the specification of ovule identity [13][14][15] .Under normal temperature (22 °C), TCPs interacts with BEL1 and AG-SEPs to inhibit the activity of AG-SEPs in wild type.The inhibition of AG-SEPs activity leads to the relatively higher activity of AG-SEPs-STK/SHPs which promotes the expression of ovule-related genes to determine ovule identity (Fig. 8a).In wild-type ovules under HT (28°C), BEL1 degradation induced by HT releases the activity of AG-SEPs, but TCPs and the remaining BEL1 (protected by TCPs) are sufficient to inhibit the activity of AG-SEPs.TCPs also promote the expression of STK to prevent the conversion of ovules into carpelloid structures under HT (Fig. 8b).In tcpDUO ovules under 22 °C, the loss of TCP function increases the activity of AG-SEPs, but the stable BEL1 under 22°C inhibits the activity of AG-SEPs; therefore, ovule development is affected, but only a very small proportion of ovules are converted into carpelloid structures (Fig. 8c).However, in the tcpDUO ovules under HT, the loss of TCP function and HT-induced degradation of BEL1 release the activity of AG-SEPs, the expression of STK is decreased without TCPs and the predominant activity of AG-SEPs over AG-SEPs-STK/SHPs promotes the expression of carpelrelated genes and the generation of carpelloid structures (Fig. 8d).In the tredecuple mutant tcpDUO bel1, because the inhibitory influences of both TCPs and BEL1 are lost, the released activity of AG-SEPs causes all of the ovules of tcpDUO bel1 to be converted into carpelloid structures, even under normal temperature (Fig. 8e).In the tredecuple mutant tcpDUO stk, the loss of STK function reduces the activity of AG-SEPs-SHPs, while the loss of function of TCP released the activity of AG-SEPs; therefore, the relative higher activity of AG-SEPs promotes the formation of carpelloid structures instead of ovules in the carpels of tcpDUO stk under normal temperature (Fig. 8f).Although all the ovules are converted into carpelloid structures in the carpels of tcpDUO bel1 and tcpDUO stk, there are differences between them.tcpDUO bel1 produced ovule-like structures in the carpelloid structures which displayed indeterminate growth, while tcpDUO stk produced even longer funiculi.These differences are consistent with the distinct functions between BEL1 and STK.It has been reported that BEL1 plays a critical role in repressing the formation of ectopic nucelli and the homeotic conversions of ovule integuments into carpelloid structures by regulating the proper expression of PIN1 and WUS 53 .In this paper, we found that the expression of WUS was highly up-regulated in tcpDUO, indicating TCPs are also important for the regulation of WUS.However, STK was reported to suppress the elongation of funiculi and to maintain the identity of ovules 22 .Our data demonstrate that CIN-and CYClike TCPs redundantly act as critical factors in determining ovule identity, while important environmental factors such as HT can reduce the stability of ovule identity, and TCPs act with BEL1 to stabilize ovule identity and antagonize the effects of HT.
The emergence of ovules was the most important morphological innovation for the fertility, dispersal, and survival of seed plants 54,55 .The determination of ovule identity is a fundamental step in ovule development.During the past few decades, factors that play a role in determining ovule identity have been identified.Genetic analysis demonstrated the roles of D-function MADS box proteins in specifying ovule identity, including STK, SHP1, and SHP2, in the widely accepted ABCDE model of floral organ specification 22 .In addition, BEL1 homeodomain protein was also shown to play an essential role in specifying ovule identity [56][57][58] .The double mutant shp1 shp2 produces rather normal ovules, and the single mutant stk generates ovules with longer funiculi and normal identity 22,24 .However, the triple mutant stk shp1 shp2 displays severe defects in the production of ovules 22 .At the early developmental stage, the ovules of stk shp1 shp2 have no obvious differences from those of wild-type plants; megasporogenesis proceeds normally and the two integuments are initiated in the ovule of the triple mutant 13 .Nevertheless, the integuments of stk shp1 shp2 grow aberrantly and are converted into a carpelloid structure during the late developmental stage 13,22 .A similar progression was observed in the bel1 mutant line 13,25 .The process of megasporogenesis by bel1 is rather normal, and the megaspore is formed in the ovules 23,56 .
Fig. 6 | TCP4 interacted with BEL1 to inhibit the degradation of BEL1 and enhance the association of BEL1 with the AG-SEP3 complex.a TCP4 interacted with BEL1 using yeast two-hybrid assays.The transformed yeasts were grown on selection SD-Leu-Trp-His medium with 10-, 100-, and 1,000-fold dilutions.The fulllength BEL1 was used as the bait.2.5 mM 3-AT was added to inhibit the selfactivation activity of BEL1.AD, activation domain; DBD, DNA-binding domain.b LCI assays indicated that TCP4 interacted with BEL1 in vivo.c Co-IP experiments confirmed that TCP4 interacted with BEL1 in vivo.35Spro-GFP-BEL1 or the control 35Spro-GFP was co-transformed into tobacco leaves with 35Spro-mTCP4-FLAG.The extracted proteins were immunoprecipitated with GFP-trap Agarose beads and detected by western blotting with antibodies against GFP or FLAG.The experiments were repeated at least three times, with at least three independent biological replicates.d The association of BEL1 with AG-SEP3 was enhanced by the coexpression of miR319-resistant mTCP4 under 22 °C using a transient expression system in tobacco leaves.e The association of BEL1 with AG-SEP3 was obviously attenuated under HT.f The quantitative analysis of fluorescence in (e, f) and the replications in Supplementary Fig. 12.Three biologically independent experiments were performed.Data are presented as mean values ± SEM.Statistical significance was determined by one-way ANOVA.Significant differences at P < 0.05 are indicated by different lowercase letters.g The expression of BEL1-FLAG-P2A-GFP in tobacco leaves showed that BEL1-FLAG was unstable under HT.h The treatment with MG132 inhibited the degradation of BEL1-FLAG-P2A-GFP and BEL1-FLAG expressed in tobacco leaves, indicating that the degradation of BEL1 was dependent on 26 S proteasome was required for the degradation of BEL1 under HT.i The CHX or CHX + MG132 treatment of inflorescence tissues from 35Spro-BEL1-FLAG transgenic plants showed that BEL1 was regulated by degradation via 26 S proteasome.j The degradation of BEL1 might be inhibited by the co-expression of miR319resistant mTCP4 with BEL1-FLAG-P2A-GFP in tobacco leaves under HT.Source data are provided as a Source Data file.The experiments of g-j were repeated at least three times, with at least three independent biological replicates.However, the conversion of ovules into carpelloid structures is also observed in bel1 at the late developmental stage of ovules 25 .In this study, we showed that CIN-and CYC-like TCP transcription factors have highly redundant functions in the control of ovule identity.The tcpDUO mutant line displays several phenotypes similar to those of stk, stk shp1 shp2, and bel1.First, tcpDUO produces longer funiculi, similar to stk and stk shp1 shp2.Second, megasporogenesis proceeds rather normally in tcpDUO, whereas megagametogenesis proceeds abnormally, as in stk shp1 shp2 and bel1.Third, tcpDUO, stk shp1 shp2 and bel1 all produce carpelloid structures instead of ovules.These similar phenotypes suggest that TCP transcription factors function in the same pathway with STK/SHPs and BEL1 in the control of ovule development and identity.Indeed, our biochemical evidence demonstrates that TCP4 interacts with AG-SEP3 and BEL1.The interaction of AG-SEP3 with TCP4 represses the transactivation activity of the AG-SEP3 complex, while the interaction of TCP4 and BEL1 increases the stability of BEL1 and thus stabilizes the interaction between BEL1 and AG-SEP3 to enhance the suppressive effect of BEL1 on the function of AG-SEP complexes.In addition, because TCP4 directly binds to the promoter region of carpel-related genes such as CRC, SPT and HEC1, TCP4 may also act as transcription factor to inhibit these genes and to prevent the carpelloid conversion during ovule development by recruiting the transcriptional repressor TCP INTERACTOR CONTAINING EAR MOTIF PROTEIN 1 (TIE1) and the corepressor TPL/TPRs 59 .
Temperature is one of the most important environmental conditions because it affects the development, reproduction, geological distribution and survival of organisms 1,60 .The elevated temperature caused by global warming has brought severe challenges to the distribution and survival of many plants and animals 1 .HT affects not only the progress of development but also the program of development, in organisms.In plants, thermomorphogenesis has been coined to describe a series of morphological changes, including elongated petioles and hypocotyls, which occur under mildly elevated temperatures 1 .Using elongated hypocotyls under HT as an output trait, many factors have been shown to play roles in the control of plant thermomorphogenesis 1,39,61,62 .However, the mechanisms by which the program of plant organ identity are affected by HT remain largely unknown.It is well established that homeotic conversion between male and female organ identity results in the determination of sexual organ identity under low or high temperatures for some animals, including turtles 7 .For example, the eggs of the turtle T. scripta all develop into female individuals under a high hatching temperature (32 °C), whereas they develop into male individuals under a relatively low hatching temperature (26 °C) 7 .The deposition of trimethylated histone 3 at lysine residue 27 (H3K27me3) in the promoter of Doublesex And Mab-3 Related Transcription Factor 1 (DMRT1), which is crucial for male sex determination, is affected by temperature, leading to different expression levels of DMRT1 under low or high temperatures 7 .In plants, the homeotic conversion of ovules into carpelloid structures is also affected by temperature in the bel1 mutant line 25 .The relative high temperatures favor the conversion of ovules into carpelloid structures 25 .However, the molecular mechanisms by which HT reduces the stability of ovule identity remain unknown.In this study, we found that HT strongly favors the conversion of ovules into carpelloid structures in tcpDUO.At normal temperature (22 °C), only 0.2% of ovules are transformed into carpelloid structures in tcpDUO, while HT (28 °C) treatment increases the rate of carpelloid structure conversion to 100%, indicating that ovule identity was less stable under HT, as observed in bel1.Our biochemical data showed that TCP4 interacted with BEL1, and both TCP4 and BEL1 repressed the expression of carpel identity genes.Furthermore, we show that HT promotes the degradation of BEL1 via the 26 S proteasome, and TCP4 may stabilize BEL1 protein to facilitate the BEL1 inhibition of AG-SEPs in repressing the specification of carpel identity.This finding suggests that plants protect ovule identity by stabilizing a key regulator of organ identity at the protein level under HT, in contrast to the epigenetic modification of a gene locus crucial for reproductive organ identity under low or high temperatures in animals 7 .TCP transcription factors are grouped into Class I and II subfamilies based on the TCP domain, and Class II TCPs are further classified in two clades: CIN-like TCPs and CYC-like TCPs 28 .The CIN gene was first identified in snapdragon (Antirrhinum majus) by analyzing the cin mutant line, which produces wavy leaves with aberrant flatness 63 ; thus, CIN-like TCPs are mainly characterized as central regulators in the control of leaf morphology in diverse plant species 31,32 .The CYC gene is a foundational member of the TCP family and was also first identified in snapdragon 64 .Disruption of CYC and its close homolog DICHOTOMA (DICH) converts snapdragon flower symmetry from bilateral to radial symmetry.The other foundational member of the TCP family is TEOSINTE BRAN-CHED1 (TB1), which also belongs to the CYC-like TCP clade and is an important domesticated gene controlling shoot branching in maize (Zea mays) 65,66 .CYC-like TCPs have been identified mainly as regulators in the control of shoot branching and flower development.Both CIN-like TCPs and CYC-like TCPs have redundant functions in regulating plant development 26,38 .We previously reported that CINlike TCPs play redundant roles in the repression of green petals and trichome formation 38,67 .The potential redundancy of the functions of CIN-like and CYC-like TCPs has not been investigated due to the difficulty associated with generating high-order multiple tcp mutants.In this study, we succeeded in generating the duodecuple mutant tcpDUO line by combining twelve mutations that disrupt both CIN-like and CYC-like TCPs.We revealed that CIN-and CYC-like TCPs redundantly control ovule identity.The functions of Class II TCPs in determining ovule identity have been masked by high crossprotein redundancy and thus have not been identified in previous studies.In addition to the homeotic transformation of ovules into carpelloid structures in tcpDUO, this mutant line also displays some other interesting phenotypes, which could be valuable subjects for experiments aimed at identifying additional functions of Class II TCPs.Our findings reveal an important area of function for Class II TCP transcription factors and expand our understanding of the molecular mechanisms underlying the determination of ovule identity.

Plant materials and growth conditions
The Arabidopsis Columbia-0 (Col-0) ecotype was used for wild-type and mutant lines in this study.The tcpDUO duodecuple mutant was generated on the basis of the tcpSEP septuple mutant by genetic crosses.The tcp12/brc2 (SALK_023116), tcp16 (GABI_655C10), and tcp18/brc1 (SALK_091920) mutants are T-DNA insertion mutants from public databases.The tcp1 and tcp24 mutants were generated by CRISPR/Cas9 technology.Arabidopsis seeds were sterilized with 75% ethanol for 10 min and were then grown on half-strength Murashige and Skoog (1/2MS) solid medium containing 10 g/L sucrose.Seed stratification was carried out at 4 °C for 48 h.The plates were then transferred to 22 °C chambers with a long-day condition (16 h light/8 h dark, 170 mmol/m2/s light intensity).The seedlings were transferred into soil six days after germination and grown in a greenhouse under the conditions described above.
For observation of the HT-induced ovule homeotic conversion phenotype of tcpDUO, bolting plants were transferred to 28  °C chambers with a long-day condition for 6 days.Then the flowers were emasculated at developmental stage 9.After two days, the ovules, the carpelloid structures or pistils were isolated for observation.Tobacco (Nicotiana benthamiana) seeds were sown in soil, and tobacco plants were grown for the transient expression experiments in the same greenhouse.

Genotyping and gene expression analysis
For genotype analysis of the T-DNA insertion tcp mutants, including tcp2, tcp3, tcp4, tcp5, tcp10, tcp12, tcp13, tcp16, tcp17, and tcp18, the fragments from wild-type loci were amplified with the TCPx-G-F and TCPx-G-R primer pairs, while the fragments from the T-DNA insertion sites of mutants were amplified with the TCPx-T and TCPx-G-R or TCPx-G-F primer pairs.
For genotyping of the mutants generated by CRISPR/Cas9, including tcp1, tcp24, bel1, and stk, the genomic fragments were amplified by the corresponding Genotyping-F and Genotyping-R primer pairs.The mutations were identified by sequencing.
For analysis of gene expression, the tissue samples, including pistils, ovules or carpelloid structures, were dissected from the flowers of wild-type or tcpDUO mutants.The tissues were immediately put in liquid nitrogen after dissection.Total RNA was isolated with a Plant Total RNA Purification Kit (GeneMark, TR02-150), and was then reverse-transcribed with an M-MLV kit (Promega, M170A).The products were diluted and then used as the templates for RT-qPCR with three repeats.UltraSYBR Mixture (CWBIO, CW2601M) and a Thermo-Fisher QuantStudio 5 real-time system were used for RT-qPCR.The cycling conditions were 94 °C for 30 s, 60 °C for 30 s and 72 °C for 30 s.The relative expression levels of genes were calculated based on the 2 -△△CT (cycle threshold) method 68 .The gene expression levels were normalized based on the expression level of ACT7.
The primers used in this study are listed in Supplementary Table 1.
The constructs were introduced into Agrobacterium tumefaciens strain GV3101 to allow the generation of Arabidopsis transgenic lines using the floral dip method or for transient expression in tobacco leaves 38 .All primers are listed in Supplementary table 1.

Microscopy analysis
For DIC observation of cleared ovules, the pistils were dissected and the ovules were separated in ovule clearing solution (8 g chloral hydrate, 1 mL 100% glycerol and 3 mL ddH 2 O) for several seconds, after which they were observed under a Zeiss AX10 Imager M2 microscope.
For SEM observation, the samples were prepared as described in the user's manual of the FEI Helios Nanolab G3 UC.The pictures were captured by an EasyLift EX NaonoManipulator and Auto Slice and View G3 software.

Yeast two-hybrid assays
For testing the interaction between TCP proteins with MADS-box proteins and BEL1 in yeasts, the bait constructs were generated by LR reactions between pENTR-AG/SEP3/STK/BEL1 and pDEST32 (Invitrogen).The prey constructs were generated by LR reactions between pENTR-TCP4/TCP16/TCP18 and pDEST22 (Invitrogen).For the truncation analysis of SEP3 and AG, the proteins were truncated according to the conserved sequences of the MADS-box domain, intervening domain, keratin domain, and C terminus.The primers used for truncation analysis are listed in Supplementary Table 1.
The bait and prey constructs were co-transformed into Saccharomyces cerevisiae strain AH109.The blank pDEST22 was cotransformed with each prey as a negative control.Medium supplemented with SD-Leu-Trp-His with or without 2.5 mM 3-amino-1,2,4triazole (3-AT) was used for yeast selection.
Tobacco leaves were co-infiltrated with different combinations of vectors.The tobacco plants were grown under a weak light for 12 hr and transferred into a greenhouse with a long-day condition (22 °C) for 32 hr.To test the interactions between proteins under 22 °C and 28 °C treatments, the tobacco plants were further subjected to an ambient temperature of 22 °C or 28 °C for 4 hr.The infiltrated leaves were then dissected and injected with 200 μM luciferin.The leaves were kept in darkness for 10 mins.The illumination images were captured with a low-light cooled CCD imaging apparatus (NightOWL II LB983) and indiGO software.An exposure time of 3 min was used to take images.The relative luminescence unit (RLU) was calculated with indiGO software.

Co-IP and western blotting
To test the interaction between TCP4 and AG or SEP3 by Co-IP in Arabidopsis protoplasts, the 35Spro-GFP-mTCP4, 35Spro-AG-FLAG, and 35Spro-SEP3-FLAG plasmids were purified from Escherichia coli strain DH5α for transformation into Arabidopsis protoplasts.To prepare protoplasts, Arabidopsis rosette leaves were digested with enzyme solution (1.5% cellulose R10, 0.4% macerozyme R10, 0.4 M mannitol, 20 mM KCl, 20 mM MES, 10 mM CaCl 2 , and 0.1% BSA) for 5 h.The protoplasts were collected and suspended with MMG buffer (0.4 M mannitol, 15 mM MgCl 2 , and 4 mM MES) to a final concentration of 2 ~5 × 10 5 cells/mL.For each Co-IP experiment, 2 mL protoplast solution was used.Transfection was performed with 100 μg purified plasmids for each expression construct combination in PEG solution (40% PEG4000, 0.2 M mannitol, and 0.1 M CaCl 2 ).The combined volume of the protoplast solution and the plasmids was kept constant among the construct combinations.After incubation at room temperature for 10 min, 10 mL W5 buffer (154 mM NaCl, 125 mM CaCl 2 , 5 mM KCl, 2 mM MES, and 5 mM glucose) was added to terminate the transfection.The transfected protoplasts were incubated under a weak light for 14 ~16 hr at room temperature, collected, and treated with IP buffer (50 mM pH 8.0 Tris-HCl, 100 mM NaCl, 1 mM MgCl 2 , 10% glycerol, 1% NP-40, 1 mM PMSF, 1 mM DTT, 20 mM MG132, and 1× protease inhibitor cocktail).The lysis mix was incubated on ice for 10 min and then centrifuged at 12000 × g for 10 min.The supernatant was mixed with the prewashed anti-GFP agarose beads (KTSM1301, KT HEALTH), and rotated at 4 °C for 4 hr.The beads were washed five times with the IP buffer.The bound proteins were eluted from the affinity beads by heating the beads to 100 °C for 10 min in 100 μL 1× SDS loading buffer, after which the proteins were detected by western blotting.
To test the interactions between TCP4 and BEL1, 35Spro-GFP-BEL1 and 35Spro-mTCP4-FLAG constructs were co-transformed into tobacco leaves.The tobacco plants were put under a weak light for 12 h, and then transferred into a greenhouse with a long-day condition (22 °C) for 36 h.The transformed leaves were dissected, homogenized with liquid nitrogen, and suspended with IP buffer.The procedures used for protein extraction, immunoprecipitation, and immunoblotting were described above.
The constructed plasmid was transferred into A. tumefaciens strain GV3101, and 35S-BEL1-FLAG-P2A-GFP was transformed into N. benthamiana leaves.After 2 days of culture, the tobacco plants were treated at 28 °C for the specified period.Proteins were extracted, and the BEL1-FLAG protein was detected with an anti-FLAG antibody (A8592, Sigma, 1:1000 dilution).GFP was detected with an anti-GFP antibody (AE011, ABclonal, 1:1000 dilution) as a control.

In vivo degradation assay
To perform in vivo BEL1 degradation assay, the inflorescences from 35S-BEL1-FLAG transgenic plants were treated by solutions containing 50 mM CHX with or without 100 mM MG132 for a time course.The treated tissues were ground into powder in liquid nitrogen.Total proteins were extracted with extraction buffer (0.1 M HEPES at pH7.5, 5 mM EGTA, 5 mM EDTA, 5 mM NaF, 0.05 M phosphoglycerol, 1% Triton X-100, 10% glycerin, and protease inhibitor cocktail, PMSF) for 1 h and then centrifuged for 10 min.The proteins were detected by western blotting using the anti-FLAG antibody (A8592, Sigma) and an anti-Actin antibody (AC009, ABclonal).

Transient dual-luciferase reporter assay
To investigate the transactivation activity of the AG-SEP3 complex, the dual-luciferase reporter assay was used with an Arabidopsis protoplast transient expression system.The 4067-bp and 4907-bp promoters of CRC and SPT, which are direct targets of the AG-SEP3 complex, were amplified with the CRCpro-F/R and SPTpro-F/R primer pairs.The promoters were each cloned into the pENTR/D TOPO vector to generate pENTR-CRCpro and SPTpro.The reporter constructs 35Spro-REN-CRCpro-LUC and 35Spro-REN-SPTpro-LUC were generated by LR reactions between pENTR-CRCpro or pENTR-SPTpro and pGreenII0800-GW-LUC. pGreenII0800-GW-LUC was generated by ligating attR1-ccdB-attR2 amplified from pK2GW7 using the primers attR1-Xho I-F and attR2-Spe I-R, with pGreenII0800-LUC digested by Xho I and Spe I. To generate effectors, the coding regions of mTCP4, AG, and SEP3 were cloned into pGreenII62SK using the FastClone method to generate 35Spro-mTCP4, 35Spro-AG and 35Spro-SEP3.The primers used for amplifying the backbones of pGreenII62SK were 62SK-FC-F and 62SK-FC-R.
Arabidopsis protoplasts were prepared as described above.For dual-luciferase reporter assays, 100 μL protoplasts were used for each transfection experiment.For plasmid transfection, each plasmid was adjusted to 10 μg.The transfected protoplasts were centrifuged and suspended with 1 mL W5 solution and incubated under a weak light for 12 ~16 hr.The LUC/REN intensity was detected with a dual-luciferase reporter system (Promega, GLO-MAX 20/20 luminometer).

CRISPR/Cas9-induced mutants
To generate tcp1 and tcp24, a CRISPR/Cas9 system containing an egg cell-specific promoter was used as previously described 69 .To generate the tredecuple mutant tcpDUO bel1 or tcpDUO stk, a CRISPR/ Cas9 system containing a UBIQUITIN10 (UBQ10) promoter was used 52 .The spacer oligonucleotides were designed on the sequences of target genes, ligated into pAtU6 vectors, and cloned into pCAMBIA1300-UBQ10-Cas9-P2A-GFP.The constructs were transformed into wild-type or tcpDUO.The mutated fragments were first amplified using PCR and were then determined by sequencing.The oligonucleotides and primers are listed in Supplementary Table 1.

GUS staining
TCP4pro-TCP4-GUS transgenic plant generation and β-glucuronidase (GUS) staining were carried out 36 .Briefly, the pistils of TCP4pro-TCP4-GUS plants were isolated and immersed in GUS staining solution at 37°C overnight.The samples were then transferred into 75% ethanol.The ovule staining pattern was observed using a Zeiss AX10 Imager M2 microscope.

RNA-seq analysis
To perform RNA-seq analysis, the bolting wild-type or tcpDUO plants were treated at 28 °C for 6 d.Then the flowers at developmental stage 9 were emasculated, and the pistils were isolated after 2 DAE from the wild-type or tcpDUO plants treated with 22 °C or 28 °C for total RNA extraction.RNA-seq library preparation and high-throughput sequencing were performed by Novogene Inc. (Tianjin, China).The libraries were sequenced on an Illumina NovaSeq 6000 platform by generating 150 bp paired-end reads.Quality control of the raw reads was performed by fastp (0.20.1).The clean reads were mapped to the Arabidopsis reference genome (TAIR10) by STAR.Expression counts were generated by featureCounts of Subread 2.0.1withdefault parameters.The R package DEseq2 was used for gene differential expression analysis (1.32.0).Volcano plots were generated by R package ggpubr (0.4.0).Genes with Bonferroni-adjusted p-value (padj) ≤ 0.05 and a log 2 fold change ≤ −1 or ≥ 1 were considered to be significantly differentially expressed.

Electrophoretic mobility shift assay
The DNA sequence encoding the TCP4 DNA binding domain was cloned into pET-21b with a C-terminal His-tag.The vector was transformed into E. coli strain BL21.The bacteria were cultured overnight at 18°C, and proteins were induced by adding IPTG to a final concentration of 0.5 mM.The cells were collected and resuspended in buffer A (25 mM Tris and 1 M NaCl [pH 8.0]).The target protein was eluted with 200-500 mM imidazole using a Ni chelating column (GE Healthcare, USA).The crude protein was then subjected to size-exclusion chromatography (Superdex 75, GE Healthcare) for the final purification step.20-30 bp DNA fragments containing a potential TCP4 binding motif were synthesized.The binding motif was mutated to AAAAAA in the mutated DNA fragments.Double-stranded DNA was obtained by annealing equimolar concentrations in an annealing buffer (25 mM Tris and 100 mM NaCl [pH 8.0]).
To carry out EMSA, the purified TCP4 DNA binding domain proteins (about 180 ng) were mixed with double-stranded DNA (about 200 ng).The mixture was incubated at 4°C for 20 min and then loaded onto a 7% nondenaturing polyacrylamide gel and subjected to electrophoresis using 0.5×TBE buffer.Electrophoresis was performed at 120 V for 40 min.Ethidium bromide staining was used to detect DNA under Tanon2500.

In situ hybridization analyses
The in situ hybridization experiments were performed as previously described 70 .Briefly, the transcripts of TCP4 or STK were amplified with primer pairs listed in Supplementary Table 1 for probe preparation.Probes were transcribed with the templates in vitro with the DIG RNA labeling kit (Roche, Switzerland) following the user manual. 1 mL hybridization buffer was prepared by adding 87.5 μL DEPC-H 2 O, 125 μL in situ hybridization salts, 500 μL deionized formamide, 250 μL 50% dextran sulfate, 25 μL 50× Denhardt's solution, and 12.5 μL 100 mg/mL tRNA for hybridization.
For sample fixation and hybridization, the fresh pistils of wild type and tcpDUO growing at 22 °C or 28 °C were dissected and immediately fixed in FAA solution.The tissues were then embedded in paraffin (Sigma-Aldrich, USA) and sectioned to 7 μm-thick slides.The samples were tiled on poly-L-lysine-treated microscope slides (Sigma-Aldrich, USA), then de-paraffinized using HistoClear (Amersco).The samples were re-hydrated by ethanol series of 100%, 100%, 95%, 90%, 80%, 60%, and 30% (vol/vol) at room temperature.The slides attached with samples were transferred to DEPC-H 2 O, then treated with freshly prepared 2×SSC for 10 min.Samples were digested with 1 μg/mL proteinase K for 20 min at 37 °C.The digestion was stopped by being treated with 0.2% glycine for 2 min.After washing with PBS twice and dehydrating with ethanol series, samples were hybridized with probes overnight at 55 °C.The samples were washed twice with 0.2× SSC at 55 °C, then twice with NTE buffer (0.01 mol/L Tris-HCl with pH8.0, 0.001 mol/L EDTA, 0.5 mol/L NaCl) at 37 °C.The samples were digested with 5 μg/mL RNase A in NTE buffer for 30 min, followed with twice washing in NTE buffer at 37 °C and once in 0.2× SSC at 55 °C.The samples were blocked in a blocking buffer (Roche) at room temperature for 45 min.Then the hybridized transcripts were detected with anti-digoxigenin antisera conjugated to alkaline phosphatase (Roche).The signals were detected with a microscope after incubation overnight in the dark at room temperature.
BioProject ID PRJNA891376.TCP4 ChIP-seq data are available under BioProject ID PRJNA474688.Source data are provided with this paper.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made.The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material.If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.To view a copy of this licence, visit http://creativecommons.org/ licenses/by/4.0/.

Fig. 1 |
Fig. 1 | Disruption of Class II TCP transcription factors in duodecuple tcpDUO null caused low fertility.a Siliques from wild type, tcpSEP, and tcpDUO.b-d Dissected siliques and close-up views showing the fertility of wild type (b), tcpSEP (c), and the tcpDUO mutant (d).e Siliques of wild type, tcpDUO, and the complementation line TCP4pro-TCP4-GFP tcpDUO.f-h Dissected siliques and close-up views for wild type (f), tcpDUO (g), and TCP4pro-TCP4-GFP tcpDUO (h).The low fertility of tcpDUO was complemented by the transformation of TCP4pro-TCP4-GFP.i Statistical analysis of the silique lengths of wild type, tcpDUO and TCP4pro-TCP4-GFP tcpDUO.j Statistical analysis of aborted and normal seeds in wild-type, tcpDUO, and TCP4pro-TCP4-GFP tcpDUO siliques.Data are the means ± SD (n = 30) of three biological replicates.Student's two-sided t-test was used for statistical analysis.****, P < 0.0001.Source data for i and j are provided as a Source Data file.Scale bars, 1 mm in a-h, 100 μm in the insets of b-d and f-h.

Fig. 2 |
Fig. 2 | The funiculi, integuments, and megagametogenesis were abnormal in the tcpDUO ovules.a-d SEM analysis of the mature ovules from wild-type (a, b) and tcpDUO (c, d).b and c, Close-up view of ovules from a and b.The red arrows indicate the ovules with abnormal integuments.The cells of funiculi are marked with red lines in b and d to show the different number of cells consisting of funiculi in wild-type and tcpDUO.Images of a-d are representatives of at least five independent individuals.e-g Representative images of DIC micrographs of ovules from wild-type (e) and tcpDUO (f, g) ovules at the FG7 stage.Green-shaded areas are the nucellus in the wild-type ovule.Pink-shaded areas are inner integuments.Yellowshaded areas are the outer integuments.The egg cell and central cell were clearly observed in the nucellus of wild-type plants (e).Excessive growth of outer or inner integuments was observed in the ovules of tcpDUO (f, g).h-k DIC micrographs of ovules from wild-type (h, j) and tcpDUO (i, k) during megasporogenesis.Images are representatives of at least five independent individuals.Source data for f-g are provided as a Source Data file.Pink-shaded areas are the megaspore mother cells.Green-shaded areas are the functional megaspore cells.MMC, megaspore mother cell.FM, functional megaspore.DM, degenerated megaspore.Mp, micropyle.Ch, chalaza.Ec, egg cell.Cc, central cell.Nc, nucellus.Ii, inner integument.Oi, outer integuments.Scale bars, 100 μm in a and c, 10 μm in b, d, h-k.20 μm in e-g.

Fig. 3 |
Fig. 3 | The tcpDUO mutant exhibited homeotic conversion of ovules into carpels under HT.a, b The opened wild-type (a) and tcpDUO (b) siliques under 22 °C.A carpelloid structure in the silique of tcpDUO is shown in (b).c, d Opened siliques from wild-type (c), tcpDUO (d) and TCP4pro-TCP4-GFP tcpDUO (e) under 28 °C.The ovules were converted into carpelloid structures in tcpDUO under 28 °C.The transformation of TCP4pro-TCP4-GFP rescued the homeotic conversion of ovules in tcpDUO.f SEM observation of a wild-type ovule under 28 °C.g-i SEM observation of tcpDUO carpelloid structures under 28 °C.g A carpelloid structure with papilla-like cells in the outer integument.h A carpelloid structure with papillalike cells in the inner and outer integuments.i A carpelloid structure with secondary carpelloid structures.Images of a-i are representatives of at least ten independent individuals.j The inflorescences of wild type under 28 °C.The flower stages were indicated in the picture.k-n, The ovules from the flower at the developmental stage 9 (k), stage 10 (l), stage 11 (m) and stage 12 (n) of wild type.Images of k-n are representatives of at least three independent individuals.o The inflorescences of tcpDUO under 28 °C.The flower stages were indicated in the picture.p-s The ovules from the flower at the developmental stage 9 (p), stage 10 (q), stage 11 (r) and stage 12 (s) of tcpDUO.Images of p-s are representatives of at least three independent individuals.The pictures of a-e, j, and o were taken with a dissecting microscope.The pictures of f-i were obtained by SEM and those of k-n and p-s were taken using a DIC Microscope.Scale bars, 1 mm in a, 100 μm in the inset of a, b-e, 20 μm in f-i. 1 mm in j and o, 50 μm in k-n and p-s.

Fig. 4 |
Fig. 4 | Carpel-related genes were repressed by TCP transcription factors during ovule development.a GO analysis of the DEGs in cluster 3. b Heatmap analysis of the genes related to carpel development.c GO analysis of the DEGs in cluster 4. d Heatmap analysis of the genes related to the formation of female gametophytes.e Relative expression levels of carpel-related genes, including CRC, SPT, HEC1, and HEC2, in 2 DAE ovules from wild-type plants or carpel-like structures from tcpDUO under 22 °C or 28 °C.The expression levels of genes in the wild-type ovules were set to 1.0.The data are the mean (±SD) of three biological replicates.One-way ANOVA analysis of variance was performed with LSD/Duncan pairwise comparison testing.Different lowercase letters indicate significant differences (P < 0.01).Source data for a and e are provided as a Source Data file.f Overlapping genes between the DEGs of WT vs. tcpDUO under 22 °C or 28 °C and the TCP4-binding gene loci.g ChIP-seq data showed that TCP4 protein was enriched in the promoter regions of CRC, SPT, and HEC1 in vivo.h EMSA analysis indicated that TCP4 protein was directly bound to the promoter of CRC and SPT in vitro.The EMSA experiments were repeated at least three times.The results in a-d were from RNA-seq analysis and the results of (e) were from RT-qPCR analysis.Source data for e and h are provided as a Source Data file.

Fig. 5 |
Fig.5| TCPs directly interacted with both AG and SEP3 to repress the transactivation activity of the AG-SEP3 complex.a The CIN-like TCP4 and the CYC-like TCP18 interacted with AG using yeast two-hybrid assays.The transformed yeast were spotted on the selection SD-Leu-Trp-His medium with 10-, 100-, and 1,000fold dilutions.2.5 mM 3-amino-1,2,4-trizole was supplemented to inhibit the selfactivation activity of AG.Full-length AG was used as bait.AD, activation domain; DBD, DNA-binding domain.b TCP4 interacted with AG in vivo as shown by LCI in tobacco leaves.The miR319-resistant TCP4 protein was fused with nLUC at the Cterminus, and AG was fused with cLUC at the N-terminus.nLUC and cLUC fused with GAL4BD were used as negative controls.c Co-IP confirmed that TCP4 interacted with AG.35Spro-GFP-mTCP4 or the control 35Spro-GFP was co-transformed into Arabidopsis protoplasts with 35Spro-AG-FLAG.The extracted proteins were immunoprecipitated with GFP-trap Agarose beads and detected by western blots with antibodies against GFP or FLAG.d TCP4 and TCP18 interacted with SEP3 using yeast two hybrid assays.The full-length SEP3 was used as the bait.The selection conditions were as described above.e LCI analysis showed that TCP4 interacted with SEP3.SEP3 was fused with cLUC at the N-terminus.f Co-IP confirmed that TCP4 interacted with SEP3.35Spro-GFP-mTCP4 or the control 35Spro-GFP was cotransformed into Arabidopsis protoplasts with 35Spro-SEP3-FLAG.g The schematic diagram of the constructs used for the transient expression assays.h, i The transactivation activity of the AG-SEP3 complex was significantly suppressed by coexpressing TCP4 using the reporter 35Spro-REN-CRCpro-LUC (h) or 35Spro-REN-SPTpro-LUC (i) using a transient expression system in Arabidopsis protoplasts.Source data for c, f, h, and i are provided as a Source Data file.The experiments of c and f were repeated at least three times, with at least three independent biological replicates.Significant differences are indicated by different letters (P < 0.05).Three biologically independent experiments were performed.Data are presented as mean values ± SEM.One-way two-sided ANOVA with post hoc Tukey HSD tests was used.

Fig. 7 |
Fig. 7 | The genetic analysis between tcpDUO and bel1 or stk. a Schematic diagram showing the deletions and one-bp insertion created by CRISPR/Cas9 in bel1 mutants in the wild-type and tcpDUO backgrounds.b Schematic diagram showing the deletions created by CRISPR/Cas9 in stk mutants in the wild-type and tcpDUO backgrounds.The green (a) and blue (b) squares indicate exons, and the black lines represent introns.c-n SEM analysis of ovules or carpelloid structures from wildtype and mutants under normal temperature (22 °C).c-f Mature ovules of wild-type (c), tcpDUO (d) and bel1 (e, f).g All ovules were converted into carpelloid structures in the tredecuple mutant tcpDUO bel1.h-j Close-up views of the carpelloid structures in tcpDUO bel1.The papilla cells and secondary ovules were clearly observed in the carpelloid structures.k Mature ovules of stk.l, m Carpelloid structures in the tredecuple mutant tcpDUO stk.Even longer funiculi and clear papilla cells were observed.Images of c-n are representatives of at least ten independent individuals.Scale bars, 100 μm in c-e, g-i, and k-m, 10 μm in f and k.

Fig. 8 |
Fig. 8 | Seesaw working model of TCP transcription factors in the control of ovule identity.a Ovule development in wild-type under 22°C.The activity of AG-SEPs-STK/SHPs was predominant due to the repressive effects of TCPs and BEL1 on the activity of AG-SEPs.b Ovule development in wild-type under 28 °C.Although the degradation of BEL1 was induced by HT, TCPs and the remaining BEL1 were sufficient to repress the activity of AG-SEP quartets.The activity of AG-SEPs-STK/ SHPs quartets was still predominant.c Ovule development in tcpDUO under 22 °C.Without the repressive effects of TCPs on AG-SEP quartets, the relatively stable BEL1 protein only slightly inhibited the activity of AG-SEP quartets, causing a low proportion of carpeloid structures to be derived from ovule conversion in tcpDUO.d Ovule development in tcpDUO under 28 °C.Without the repressive effects of Article https://doi.org/10.1038/s41467-023-41416-1Nature Communications | (2023) 14:5673

Table 1 |
The reciprocal crosses between tcpDUO and the wild type *Numbers are mean values with standard deviations in parentheses.For reciprocal crossing, n = 10.The different superscript letters (a, b) of the fertility percentages indicate significant differences determined using two-sided one-way ANOVA analysis, P < 0.0001.Source data are provided as a Source Data file.