Angiosperm leaves exhibit great morphological diversity and are conventionally classified as simple or compound based on the number of blades on their petioles. A simple leaf consists of a single, undivided, continuous leaf blade, whereas a compound leaf consists of several discrete blade units termed leaflets. Leaf development follows a common developmental program, which can be flexibly adjusted species-specifically in a spatiotemporal manner1. Morphogenesis and differentiation are two key stages that determine the final leaf form in leaf development. Compared to simple leaf, compound leaf formation requires a prolonged morphogenetic phase that allows for the initiation of leaflets and specification of the midrib, petiole, and rachis2. This extended morphogenetic phase is enabled by the sufficient maintenance and modulation of the transient morphogenetic activity at specific leaf margin meristematic regions termed marginal blastozones2. Thus, maintaining a transient morphogenetic window during leaf morphogenesis is crucial for the formation of compound leaves.

The KNOTTED-like homeobox (KNOX) and BEL1-like homeobox (BLH) genes constitute the plant-specific three-amino-acid loop extension (TALE) superfamily, which plays a crucial role in plant development3. The KNOX and BELL proteins form heterodimers that are implicated in nuclear localization and regulate DNA-binding affinity4,5,6,7. KNOX genes can be further categorized into three subclasses, Class I KNOX, Class II KNOX, and Class M KNOX8. Class I KNOX (KNOXI) transcription factors are essential for maintaining the shoot apical meristem (SAM) activity and regulating leaf complexity in angiosperms9,10,11,12. In most compound-leafed species, including tomato (Solanum lycopersicum) and hairy bittercress (Cardamine hirsuta), KNOXI proteins are crucial for maintaining transient morphogenetic activity during leaf development, as their reactivation in the leaf primordium induces leaflet initiation13,14. Furthermore, KNOXI has been thought to promote morphogenesis in most compound-leafed species, and its overexpression dramatically increases the number of leaflets. In tomato, the BLH protein BIPINNATE (BIP) interacts with the KNOXI protein to form a heterodimer that is re-located to the nucleus, resulting in determinate growth for leaf morphogenesis15. The loss-of-function of BIP leads to increased primary and secondary leaflet production and higher expression of KNOXI. In legume plants belonging to the inverted repeat-lacking clade (IRLC), such as pea and Medicago truncatula, KNOXI seems to be not involved in compound leaf development. The expression of KNOXI is not reactivated in developing leaf primordia, as distinguished from most compound-leafed species16,17. The role of KNOXI is replaced by the transcription factor LEAFY (LFY) orthologues, UNIFOLIATA (UNI) in pea and SINGLE LEAFLET1 (SGL1) in M. truncatula, which are essential for regulating the transient morphogenetic window of leaf growth18,19,20. The loss-of-function sgl1 mutant fails to generate trifoliate adult leaves but produces a single blade. However, the constitutive expression of SGL1 did not increase leaf complexity, which is distinguished from the overexpression of STM/BP-like KNOXI expression in M. truncatula, indicating that SGL1 and MtKNOXI act in parallel pathways to regulate different targets17.

Medicago truncatula is a model legume with leaves having a typical trifoliate pattern. Recent findings have revealed the regulatory mechanisms of compound leaf patterning in M. truncatula, for instance, the BLH protein PINNATE-LIKE PENTAFOLIATA1 (PINNA1) represses the transcription of SGL1 in a spatiotemporal manner to define the trifoliolate form in wild-type plants21. However, the leaf pattern in SGL1-overexpressing plants is similar to that in wild-type plants17, implying that the increased activity of SGL1 is not responsible for the generation of ectopic leaflets in the pinna1 mutant. Here, we characterized a novel regulatory pathway modulated by PINNA1 for the formation of additional leaflets in pinnate-like compound leaves in M. truncatula. We found that mutations in PINNA1 led to the formation of ectopic leaflets in pinnately compound leaves, and the defects in compound leaf patterning in pinna1 could be partially rescued by the simultaneous disruption of the STM/BP-like and KNAT2/6-like MtKNOXI genes. Genetic and biochemical analysis revealed that PINNA1 forms a heteromeric complex with MtKNOXI and sequesters it to the cytoplasm, preventing the direct activation of the target gene, UNUSUAL FLORAL ORGANS (UFO) ortholog MtUFO, whose ectopic expression leads to a pinnate-like form with five leaflets. In this work, we proposed a model to understand how trifoliate leaves transform into pinnate-like pentafoliate leaves and shed light on the regulatory mechanisms of compound leaf formation in M. truncatula.


Comprehensive characterization of the roles of Class I MtKNOX in compound leaf development

Phylogenetic analysis revealed that the STM-like genes (MtKNOX1 and MtKNOX6), a BP-like gene (MtKNOX2), and a KNAT2/6-like gene (MtKNOX7) are grouped into Class I KNOX genes (MtKNOXI) in M. truncatula (Fig. 1a). Additionally, we found that MtKNOX8 and its homologs in legume species formed a distinct clade which is not grouped with other MtKNOXI genes (Fig. 1a and Supplementary Fig. 1a). Multiple sequence alignment analysis showed that the N-terminus region of the MtKNOX8 protein had a shorter amino acid sequence compared to other KNOXI proteins (Supplementary Fig. 2). This result implies that the function of MtKNOX8 is different from other members of the MtKNOXI proteins. In compound-leafed species, such as S. lycopersicum and C. hirsuta, the ectopic expression of KNOXI genes results in a dramatic increase in leaf complexity3,13,22,23. To comprehensively assess if MtKNOX1/2/6/7 genes have conserved roles, they were introduced into wild-type plants under the control of the CaMV35S promoter. The resulting transgenic plants displayed a similar phenotype that the reiteration of higher-order leaflets was produced along petiolules (Supplementary Fig. 1b–j), but those plants had a seedling-lethal phenotype. The transgenic plants with relatively low expression of MtKNOXI could survive, and showed a weak phenotype with two ectopic leaflets formed in a pinnately compound leaf form (Fig. 1b–f and Supplementary Fig. 1g–j). We also assessed the function of MtKNOX8; however, the 35S:MtKNOX8 transgenic plants showed no obvious changes in leaf morphology (Fig. 1g and Supplementary Fig. 1k), suggesting distinct roles of MtKNOX8 and MtKNOX1,2,6,7 proteins. These observations indicate that ectopic MtKNOXI activity is sufficient for increasing leaf complexity.

Fig. 1: Comprehensive characterization of the roles of MtKNOXI in compound leaf development.
figure 1

a Phylogenetic tree of KNOX proteins from several angiosperms (A. thaliana, M. truncatula, C. cajan, G.max, L. japonicas, C. arietinum, P. vulgaris, A. duranensis, V. angularis, V. radiata, and P. sativum). b Leaf morphology of wild-type (WT) plants. cf Representative leaves derived from transgenic plants overexpressing 35S:MtKNOX1 (c), 35S:MtKNOX2 (d), 35S:MtKNOX6 (e), and 35S:MtKNOX7 (f) with a weak phenotype, respectively. g Mature leaf in transgenic plants overexpressing 35S:MtKNOX8. (w/o) in (cg) indicates that w in o total transgenic plants shows the displayed features. h Gene model of MtKNOX7 and Tnt1 insertion positions in mtknox7 alleles. Boxes represent exons and lines represent introns. i Reverse transcription-PCR (RT-PCR) analysis of MtKNOX7 transcripts in WT and mtknox7 mutants. MtActin was used as the loading control. Similar results were obtained from three independent experiments. jn Representative leaves of WT (j), mtknox7-1 (k), mtknox7-2 (l), mtknox1-1 mtknox2-1 mtknox6-1 triple mutants (m), and mtknox1-1 mtknox2-1 mtknox6-1 mtknox7-1 quadruple mutants (n). Scale bars, 5 mm in (bg and jn). Source data are provided as a Source Data file.

To better understand the functions of MtKNOXI genes, their loss-of-function mutants were investigated. A previous study showed that the simultaneous disruption of STM/BP-like MtKNOXI genes did not lead to obvious defects in leaf morphology17. To determine whether the KNAT2/6-like gene is involved in compound leaf development, two mutant alleles of MtKNOX7 were obtained by the reverse genetic screening of the Tnt1-tagged mutant population (Fig. 1h). A reverse transcription-PCR showed no detectable MtKNOX7 transcripts in mutants (Fig. 1i), and the mtknox7 mutant showed no obvious defects in leaf morphology (Fig. 1j–l). To further assess functional redundancy between STM/BP-like and KNAT2/6-like MtKNOXI genes, a quadruple mutant was generated. Similar to the mtknox1/2/6 triple mutant (Fig. 1m), knockout of MtKNOX1/2/6/7 in M. truncatula did not affect leaf development (Fig. 1n), indicating that MtKNOXI genes are not involved in compound leaf patterning.

MtKNOX1/2/6/7 and PINNA1 interact physically

Ectopic expression of MtKNOXI led to enhanced morphogenetic activity in developing compound leaves, implying that the function of MtKNOXI was repressed during leaf primordia development in M. truncatula. To identify a possible repressor, we used the full-length MtKNOX7 cDNA as a bait to identify potential interacting partners by a yeast two-hybrid screening. We found that MtKNOX7 interacted strongly with a BLH protein that was reported previously, PINNA121. Further analysis of the yeast two-hybrid X-a-Gal filter assay showed that MtKNOX1/2/6/7 were also able to interact with PINNA1 (Supplementary Fig. 3). To verify these interactions in vivo, a bimolecular fluorescence complementation (BiFC) experiment and a coimmunoprecipitation (co-IP) assay were carried out in Nicotiana benthamiana. For the BiFC assay, the MtKNOX1/2/6/7 proteins were fused with the N-terminus of YFP (YN), and PINNA1 was fused with the C-terminus of YFP (YC). A co-expression of PINNA1-YC and each MtKNOX1/2/6/7-YN resulted in significant YFP signals in the cytoplasmic space of leaf epidermal cells, suggesting their strong extranuclear interaction (Fig. 2a and Supplementary Fig. 4). Furthermore, these YFP signals were found to co-localize with the puncta of the autophagosome marker mCherry-ATG8e and the autophagy receptor NBR1-mCherry24,25 (Supplementary Fig. 4a, b). This observation indicates that the complexes were localized within autophagosomes or autophagic bodies and could potentially undergo degradation through the autophagy process. Additionally, the co-IP assays showed that PINNA1-7Myc could be coimmunoprecipitated with MtKNOX1/2/6/7-GFP (Fig. 2b), suggesting that PINNA1 can heterodimerize with MtKNOX1/2/6/7. Previous studies indicate that BLH proteins heterodimerize with KNOX proteins through their KNOX MEINOX domain26,27, which is composed of two subdomains, MEINOX1 (also referred to as KNOX1) and MEINOX2 (also referred to as KNOX2), essential for BLH-KNOX dimerization. To identify which MEINOX domain interacted with PINNA1, a yeast two-hybrid assay was performed. The data showed that MEINOX1 was sufficient for the interaction between MtKNOX7 and PINNA1, and MEINOX2 was sufficient for the interaction between MtKNOX1/2/6 and PINNA1 (Supplementary Fig. 5). Such different interaction modes between MtKNOX1/2/6-PINNA1 and MtKNOX7-PINNA1 likely suggested a functional difference between the STM/BP-like and KNAT2/6-like MtKNOXI genes, as the domains of a protein regulate its functions and interplays28.

Fig. 2: Physical and genetic interaction of MtKNOXI and PINNA1.
figure 2

a BiFC assay was performed in tobacco epidermal cells, and it was observed that PINNA1 interacts with MtKNOX1, MtKNOX2, MtKNOX6, and MtKNOX7, as indicated by the presence of a yellow fluorescent signal. DAPI (4’,6-diamidino-2-phenylindole) was used to stain the nuclei. Similar results were obtained from three independent experiments. b Coimmunoprecipitation assay confirmed that PINNA1 interacts with MtKNOX1/2/6/7 in tobacco epidermal cells. Total proteins were immunoprecipitated using GFP-Trap beads, and the coimmunoprecipitated proteins were detected by the anti-Myc antibody. Three biological repeats were performed for each interaction. c Schematic diagram of PINNA1 gene structure and Tnt1 insertions in pinna1 alleles. Boxes represent exons and lines represent introns. d Representative leaves of WT. eg Representative leaves of three pinna1 mutants, pinna1-6 (e), pinna1-7 (f), and pinna1-8 (g). h Proportion of leaves with different leaflet numbers in WT, pinna1-6 single mutant, and pinna1-6 mtknoxi multiple mutants. Fifty-day-old plants were used for counting the leaflet numbers of adult leaves at least three biologically independent replicates. Data represent means ± SD (n = 3 biological replicates). Scale bars, 20 μm in (a) and 5 mm in (dg). Source data are provided as a Source Data file.

MtKNOXI activities partially contribute to defects in compound leaf patterning in pinna1 mutants

To characterize the relationship between MtKNOXI and PINNA1, we isolated three new pinna1 knockout mutant alleles by reverse genetic screening (Fig. 2c and Supplementary Fig. 6a). Approximately 89% of adult leaves in pinna1 produced five leaflets in a pinnately compound leaf form (Fig. 2d–g and Supplementary Fig. 6b–d). A scanning electron microscopy (SEM) analysis showed that a pair of incipient ectopic leaflets of pinna1-6 emerged as two bulging cell groups from the proximal zones of the terminal leaflet primordium (Supplementary Fig. 6e–h). It is noteworthy that the compound leaf pattern in pinna1 was similar to that of 35S:MtKNOX1/2/6/7 plants with a weak phenotype (Fig. 1c–f). These observations implied that the MtKNOX1/2/6/7 activity may contribute to the formation of ectopic leaflets in pinna1 mutants. To validate this hypothesis, we generated double, triple, quadruple, and quintuple mutants among the mtknoxi and pinna1 mutants. Among different double mutant combinations, approximately 49% of leaves in pinna1-6 mtknox7-1 were trifoliate, but only 10.9 to 21.8% of leaves were rescued in the other three double mutant combinations (Fig. 2h). Moreover, knockout of all MtKNOX1, 2, 6 in pinna1-6 could rescue 53.2% of leaves (Fig. 2h), which was similar to that in pinna1-6 mtknox7-1, indicating a comparable contribution of MtKNOX1, 2, 6 and MtKNOX7 to the defects in pinna1-6. Furthermore, knockout of all MtKNOXI genes in pinna1-6 rescued 79.8% of leaves of the mutant (Fig. 2h), indicating the redundant roles of STM/BP-like and KNAT2/6-like MtKNOXI in the regulation of compound leaf patterning in pinna1-6 mutants. Taken together, this genetic evidence suggests that the leaf phenotype seen in pinna1 is caused partially by the activities of MtKNOXI.

PINNA1 mainly represses the activities of MtKNOXI by regulating its subcellular localization

To our knowledge, BLH-KNOX heterodimerization is functional to the translocation of the BLH-KNOX complex from the cytoplasm to the nucleus4,5,6,7. However, all PINNA1-MtKNOXI complexes were not trafficked into the nucleus in this study (Fig. 2a), indicating the existence of a distinct BLH-KNOX regulatory circuitry in M. truncatula. To investigate the underlying possible reasons for this, the subcellular localizations of these proteins were first determined. In plants, the relative accessibility of nuclear export signals (NES) and/or nuclear localization signals (NLS) generally determines the nucleo-cytoplasmic partitioning of proteins29. We found that PINNA1 and all MtKNOX1/2/6/7 proteins have conserved amino acids that match the NES amino acid sequence (Supplementary Fig. 7), and speculated that they are not localized in the nucleus. In fact, both PINNA1-GFP and PINNA1-mCherry fusions were excluded from the nucleus in M. truncatula protoplasts (Fig. 3a, b and Supplementary Fig. 8a). This finding is distinctly different from a previous report that PINNA1 was localized to the nucleus in tobacco leaf epidermal cell21. We also transiently expressed PINNA1-GFP or GFP-PINNA1 in M. truncatula epidermal cells, and both proteins were found localized to the cytoplasmic space (Supplementary Fig. 8b). To further confirm these results, we performed a nuclear-cytoplasmic fractionation assay to test the localization of GFP- or Myc-tagged PINNA1 proteins by immunoblotting the nuclear and cytoplasmic fractions of transfected M. truncatula protoplasts. The immunoblot analysis showed that both GFP- and Myc-tagged PINNA1 proteins were mostly present in the cytoplasmic fraction (Supplementary Fig. 8c). Therefore, it is unlikely that PINNA1 functions as a transcriptional factor to regulate the downstream genes. Next, we determined the subcellular localization of MtKNOX1/2/6/7-GFP. We found that less than 10% of GFP signals of each MtKNOX1/2/6-GFP were nucleo-cytoplasmic; however, MtKNOX7-GFP showed 100% nucleo-cytoplasmic distribution (Fig. 3a, b and Supplementary Fig. 8d), indicating that different regulation mechanisms are mediated by MtKNOX7.

Fig. 3: PINNA1 regulates the subcellular localization of MtKNOXI and represses their activities.
figure 3

a Co-localization of PINNA1-mCherry and each MtKNOX1/2/6/7-GFP in the Medicago protoplasts. Nuclei were stained by Hoechst (Hoechst 33342). b Quantification of subcellular localization patterns of PINNA1-mCherry, PINNA1-GFP, and each MtKNOX1/2/6/7-GFP in protoplasts respectively. c Quantification of subcellular localization patterns of GFP/mCherry fluorescence signal when expressed PINNA1-mCherry together with each MtKNOX1/2/6/7-GFP in protoplasts. Ratios of subcellular localization patterns were calculated from 50 protoplasts with at least three biologically independent replicates in (b) and (c). Data represent means ± SD (n = 3 biological replicates). d Transcriptional activation assay in Arabidopsis protoplast indicated that PINNA1 represses the activity of MtKNOX7. Data represent means ± SD (n = 3 biological replicates) and P values were calculated by unpaired two-tailed t-test. e RNA in situ hybridization of MtKNOX7 in the leaf primordia of WT. f RNA in situ hybridization of PINNA1 in the leaf primordia of WT. g Schematic representations of expression patterns of MtKNOX7 and PINNA1 in the leaf primordia of WT. h RNA in situ hybridization of MtKNOX7 in the leaf primordia of pinna1-6. i Schematic representations of the inhibitory effect of PINNA1 on the functions of MtKNOX1/2/6/7. PINNA1 has stronger inhibitory effects on MtKNOX7 than that on MtKNOX1/2/6. The orange full curve represents the stronger inhibitory effects and the orange dotted curve represents the weaker ones. The subcellular localization experiments in (a) and the RNA in situ hybridization in (e, f, h) were independently repeated three times with similar results. Scale bars, 5 μm in (a), and 50 μm in (e, f, h). Source data are provided as a Source Data file.

To further understand the biological effects of PINNA1-MtKNOXI heterodimerization, we transiently co-expressed each MtKNOX1/2/6/7-GFP with PINNA1-mCherry in protoplasts. The data showed that all GFP/mCherry signals were excluded from the nucleus (Fig. 3a, c). As the control, 35S-mCherry did not affect the subcellular localization of each MtKNOX1/2/6/7-GFP protein (Fig. 3a, c). These results indicated that PINNA1 prevents the nuclear localization of MtKNOX1/2/6/7 from promoting leaflet development. To further confirm this, a dual-luciferase reporter (DLR) system assay was performed by transiently expressing the fusion protein GAL4BD-MtKNOX1/2/6/7 with 6XUASpro:LUC reporter in protoplasts. MtKNOX1/2/6 displayed relatively low transactivation activity, except MtKNOX7 (Fig. 3d and Supplementary Fig. 8e), consistent with their different subcellular localization patterns. Subsequently, each of the GAL4BD-MtKNOXI members and 35S:PINNA1 were co-expressed in protoplasts along with the 6XUASpro:LUC reporter, and it was found that the activity of MtKNOX7 was repressed by PINNA1 (Fig. 3d and Supplementary Fig. 8e). The activities of MtKNOX1/2/6 were not altered significantly by PINNA1, probably because of their mostly cytoplasmic localization. Previous reports showed that the KNOXI expression is affected in the mutants of the BLH gene, which is involved in the generation of leaf-form variations15,30. We also found that the transcriptional levels of MtKNOXI were slightly increased in the pinna1 mutant (Supplementary Fig. 9). To further investigate the relationship between MtKNOX7 and PINNA1, the expression patterns of MtKNOX7 and PINNA1 were examined by RNA in situ hybridization analysis. The results showed that MtKNOX7 transcripts were only present in shoot apical meristem (SAM), but not in the incipient leaf primordia (S0) and developing leaf primordia in the wild type (Fig. 3e). However, the spatial localizations of PINNA1 were detected in SAM, S0, S1, and developing leaf primordia in the wild type (Fig. 3f). Thus, MtKNOX7 and PINNA1 have the overlapped expression domain in the SAM and the boundary between SAM and the emerging leaf primordia (Fig. 3g), suggesting that there is a potential for the interaction between them. Additionally, the ectopic expression signal of MtKNOX7 was detectable in the S0, S1, and S2 primordia in the pinna1-6 mutants (Fig. 3h). These data suggest that the leaf phenotype seen in pinna1 is probably caused, at least partially, by the ectopic expression of MtKNOX7. Taken together, PINNA1 has diverse inhibitory effects on MtKNOXI (Fig. 3i), which are mainly mediated by the regulation of nucleo-cytoplasmic partitioning and repression of the expression domain of MtKNOX7.

Pentafoliate leaves in pinna1 result from the increased activity of MtUFO

In M. truncatula, all adult leaves in sgl1 mutant were simple18, indicating that SGL1 is necessary for regulating indeterminacy during leaf development. The role of PINNA1 in producing lateral leaflets was further examined by generating double mutants with the sgl1 mutant. Most leaves in the sgl1 pinna1 double mutant were simple (Supplementary Fig. 10a–f), suggesting that sgl1 is genetically epistatic to pinna1 in leaf patterning. In our previous report, however, leaf complexity was not changed in SGL1-overexpressing plants (Supplementary Fig. 10g, h)17, indicating that the ectopic SGL1 activity is not sufficient for increasing leaf complexity. Therefore, the generation of the two extra leaflets in pinna1 was not due to the ectopic expression of SGL1. To determine the potential regulator of the defects in pinna1, we identified the genes involved in the regulatory network of LFY/SGL1. It has been shown that LFY functions with its coregulatory gene UNUSUAL FLORAL ORGANS (UFO) in flower development31. In addition, the ectopic expression of UFO in Arabidopsis and C. hirsuta led to deeply serrated leaf margins and increased leaf complexity, respectively31,32. These observations indicate that UFO promotes indeterminacy in leaf development, implying that UFO orthologs may play an important role in compound leaf patterning in M. truncatula. To test this hypothesis, we overexpressed MtUFO under the CaMV35S promoter. The transgenic plants exhibited five leaflets in a pinnate-like form, similar to those in pinna1 (Fig. 4a–c) and 35S:MtKNOXI plants with weak phenotype (Fig. 1c–f). This result indicated that the ectopic MtUFO activity promotes morphogenetic activity during leaf morphogenesis in M. truncatula. To investigate the role of MtUFO in the development of pentafoliate leaves in pinna1, we compared the expression levels of MtUFO in WT and pinna1. Our qRT-PCR analysis revealed a significant upregulation of MtUFO expression in pinna1 (Fig. 4d). To further explore the function of MtUFO, a reverse genetic screening of the Tnt1-tagged mutant population was performed (Supplementary Fig. 11a–c). In total, four loss-of-function mutants of MtUFO were isolated, all of which produced normal trifoliate leaves (Fig. 4e). Subsequently, mtufo-1 was crossed with pinna1-6 to generate the double mutant, and the leaf defects in pinna1-6 were fully rescued in pinna1-6 mtufo-1 (Fig. 4f). Furthermore, RNA in situ hybridization experiments showed that the expression domain of MtUFO was expanded in the leaf primordium (S1) of the pinna1-6 mutant, in comparison to the WT (Supplementary Fig. 11d, e). These observations indicate that the increased activity of MtUFO is responsible for the formation of extra leaflets in pinna1.

Fig. 4: Pentafoliate leaves in pinna1 result from the increased activity of MtUFO.
figure 4

ac Representative leaves of WT (a), pinna1-6 (b), and 35S: MtUFO (c). d The expression levels of MtUFO in the shoot apices of WT and pinna1 were determined by qRT-PCR. MtUBIQUITIN was used as the internal control. Data represent mean ± SD (n = 3 biological replicates) and P values were calculated by unpaired two-tailed t-test. e, f Representative leaves of mtufo-1 (e) and pinna1-6 mtufo-1 double mutants (f). Scale bars, 5 mm in (ac, e, f). Source data are provided as a Source Data file.

MtUFO functions downstream of the PINNA1-MtKNOX7 complex

To assess the effects of PINNA1 or MtKNOXI on the expression of MtUFO, a DLR assay was performed using the GFP-tagged PINNA1 or MtKNOXI effectors. PINNA1 and MtKNOX1/2/6 had no influence on the expression of LUC driven by a 2.9-kb promoter of MtUFO; however, it was effectively activated by MtKNOX7 semi-in vivo (Supplementary Fig. 12). In addition, the MtUFO promoter-driven LUC expression was significantly activated by GR-tagged MtKNOX7 upon treatment with dexamethasone (DEX) (Fig. 5a, b). The KNOXI protein is reported to bind to DNA sequences containing tandem TGAC core motifs33,34,35,36,37. Electrophoretic mobility shift assay (EMSA) revealed that the MtUFO promoter contains the cis-element TGACTTGAC, which acts as a binding site for MtKNOX7, at 1974 bp upstream of the translation initiation codon (Fig. 5c). The binding of MtKNOX7 to the TGACTTGAC cis-element was also confirmed by chromatin immunoprecipitation (ChIP) assay. A ChIP-qPCR analysis showed that the MtUFO promoter region containing the TGACTTGAC cis-element was significantly enriched in the precipitated DNA from 35S:MtKNOX7-GFP plants. (Fig. 5d). Moreover, to test the effect of the PINNA1-MtKNOX7 complex on the activities of the MtUFO promoter (proMtUFO), we performed a DLR assay. The activity of proMtUFO co-expressed with PINNA1 and MtKNOX7 was effectively repressed compared to when it was activated by MtKNOX7 alone, indicating that PINNA1 represses the expression of MtUFO by interacting with MtKNOX7 (Fig. 5e). To further confirm this, the expression levels of MtUFO were measured and found significantly upregulated in 35S:MtKNOX7 plants, compared to wild-type plants (Fig. 5f). In addition, the ectopic leaflets in 35S:MtKNOX7 plants were completely recovered when MtUFO was knocked out (Fig. 5g, h). These data suggested that MtUFO functions downstream of the PINNA1-MtKNOX7 complex for promoting indeterminacy in leaf development.

Fig. 5: MtUFO functions downstream of the PINNA1-MtKNOX7 complex.
figure 5

a Schematic representation of a reporter and an effector for DLR assay. The LUC (Firefly luciferase) was under the control of the MtUFO promoters (proMtUFO) as the reporter, while the REN (Renilla luciferase) was under the control of the CaMV35S promoter (35Spro) as the internal control. b A DLR assay indicated that the promoter of MtUFO was significantly activated by a GR-tagged MtKNOX7 when it was treated with dexamethasone (DEX). Data represent mean ± SD (n = 4 biological replicates). c EMSA assay showed that MtKNOX7-HIS fusion protein directly binds to the probe of the MtUFO promoter containing the TGACTTGAC cis-element. The arrow indicates the shifted bands. Similar results were obtained from three independent experiments. d ChIP-qPCR assays of the MtKNOX7-GFP protein binding to the promoter of MtUFO. Data represent enrichment values normalized to MtUBIQUITIN. Data represent mean ± SD (n = 3 biological replicates). e A DLR assay in tobacco cells showed that the PINNA1-MtKNOX7 complex represses the MtKNOX7-mediated activation of MtUFO transcription. Data represent mean ± SD (n = 3 biological replicates). f The expression levels of MtUFO in the shoot apices of WT and 35S:MtKNOX7 were determined by qRT-PCR. MtUBIQUITIN was used as the internal control. Data represent mean ± SD (n = 3 biological replicates). g, h Representative leaves of 35S:MtKNOX7-GFP (g) and 35S:MtKNOX7-GFP mtufo-1 plants (h). P values in (b, d, f) were calculated by unpaired two-tailed t-test, while those in (e) were evaluated by one-way ANOVA with Tukey’s multiple comparisons test. Scale bars, 5 mm in (g, h). Source data are provided as a Source Data file.


Several studies have shown that interactions between TALE (three-amino acid loop extension) homeodomain proteins play important roles during plant and animal development26,38. In animals, the interactions between the TALE proteins MEIS and PBC mask the NES in the PBC protein from the nuclear export receptor CRM1/exportin-1, thereby facilitating the nuclear localization of the MEIS-PBC heterodimers39. In plant development, the combinatorial interactions among TALE proteins make up functional BLH-KNOX heterodimers40. Selective BLH-KNOX interactions guide the correct subcellular localization of BLH-KNOX complexes and mediate DNA-binding activity, which plays distinct roles in plant growth and development. Here, our data suggest conserved and diverged molecular mechanisms by which the functional BLH-KNOX complexes control compound leaf development in M. truncatula. Our data showed that the BLH protein PINNA1 heterodimerizes with STM/BP-like and KNAT2/6-like MtKNOX in the cytoplasmic space, in contrast to the previously reported nuclear localization of the BLH-KNOX complexes4,5,6,7. In particular, the cytoplasm-localized PINNA1-MtKNOX7 heterodimers prevent MtKNOX7 protein trafficking to the nucleus, which inhibits transcriptional regulation of the target genes controlling compound leaf development (Fig. 6).

Fig. 6: A proposed working model for PINNA1-MtKNOX7 complex and MtUFO/MtKNOX7 module action in the compound leaf development in M. truncatula.
figure 6

In the wild type, PINNA1 represses the expression of MtKNOX7. Moreover, PINNA1 forms heterodimers with MtKNOX7 protein in the cytoplasmic space and attenuates the shuttling of MtKNOX7 from the cytoplasm into the nucleus. In pinna1 mutant, the expression domain of MtKNOX7 is expanded to leaf primordia, and more nuclear-localized MtKNOX7 proteins activate the transcription of MtUFO, thereby promoting indeterminacy in developing leaves.

Nucleo-cytoplasmic transport is an important means of regulating the subcellular localization of various TALE proteins, which is important for their proper functioning. In TALE transcription factors, NES and NLS are generally short peptides recognized by a variety of nuclear importin and exportin proteins required for nucleus-cytoplasm trafficking41. In plants, the nuclear import mechanism of the BLH-KNOX complexes is similar to that of the PBC-MEIS complex in animals. The heterodimerization of BLH-KNOX likely masks the NES to preclude the recognition of CRM1, thus causing BLH-KNOX heterodimers to accumulate in the nucleus6. Therefore, the nuclear import of BLH-KNOX complexes is correlated to the effect of masking the NES by heterodimerization and balance between the activities of the NES and NLS. In this study, we used NLStradamus software ( to predict the NLS of PINNA1 and MtKNOXI. All MtKNOXI proteins had a predictable NLS (Supplementary Fig. 13), but PINNA1 did not, suggesting that the cytoplasmic localization of PINNA1 was probably due to the lack of efficient NLS. Moreover, all PINNA1-MtKNOXI complexes were localized in the extranuclear space, probably because the heterodimerization of PINNA1-MtKNOXI did not mask the NES that was recognized by CRM1 or because the heterodimers lacked an efficient NLS. Therefore, treatment with leptomycin B (LMB) specifically inhibits the CRM1 activity42 or the fusion of an efficient NLS to PINNA1, thus providing an insight into the mechanism of the nuclear export of PINNA1-MtKNOXI complexes in the future.

The subcellular localization diversity existing among KNOXI orthologs in M. truncatula has also been observed in other species. In rice, three KNOXI (Oskn1/OSH1, Oskn2/OSH71, and Oskn3/OSH15) proteins showed different nuclear and cytoplasmic localization patterns. Different from the only nucleo-cytoplasmic localization of Oskn2 or Oskn3, the localization of Oskn1 varied between different cells, showing either cytoplasmic or nucleo-cytoplasmic distribution43. The subcellular localization of KNOXI was influenced by plant hormones. For example, the application of gibberellic acid (GA3) and 1-naphthylacetic acid (1-NAA) completely changed the localization of all Oskn1-3 proteins from the nucleus to the cytoplasm, whereas the application of naphthylphthalamic acid (NPA) led to their nuclear concentration43. Moreover, the subcellular localization of KNOXI might be influenced by other proteins, such as the members of the multiple C2 domain and transmembrane region protein (MCTP) family and a microtubule-associated protein. The MCTP protein FTIP3/4 prevents intracellular trafficking of STM from endosomes to the plasma membrane and facilitates STM recycling to the nucleus but does not affect KNAT2 in Arabidopsis44, and OsFTIP7 also mediates the nucleo-cytoplasmic translocation of OSH1 in rice45. MPB2C, a microtubule-associated protein, can alter the subcellular distribution of KNOXI protein KN146. Therefore, different subcellular localizations between MtKNOX1/2/6 and MtKNOX7 in single cells could be due to multiple regulatory mechanisms.

MtKNOX8 and its homologs in legume species formed a distinct clade in our phylogenetic analyses. While MtKNOX8 exhibits close association with MtKNOXI proteins and demonstrates a physical interaction with PINNA1 in N. benthamiana leaves (Supplementary Fig. 14), we found distinct functional roles of MtKNOX8 and MtKNOXI proteins in regulating the leaf complexity. Furthermore, MtKNOXI proteins act redundantly to modulate the leaf form in pinna1, but interact with PINNA1 through different functional subdomains of MtKNOXI MEINOX. These results suggest that the MtKNOX gene family can be subdivided and neofunctionalized in M. truncatula. Previous studies showed that alterations in the homeostasis of BLH/KNOX heterodimers can affect compound leaf development. In tomato, the mutations in BIP resulted in KNOX1 overexpression and a highly complex leaf phenotype, indicating the BIP-mediated repression of KNOXI activities15. The dominant mutant Petroselinum (Pts), characterized by a more compound leaf and higher KNOX1 expression than the wild type, was proposed to disrupt the BIP-KNOXI interaction by competing with the KNOXI protein to bind BIP15. In this study, the constitutive expression of MtKNOXI generated more leaflets, indicating that the downstream effectors were still sensitive to the MtKNOXI activity. Thus, we proposed that MtKNOXI was involved in the defects of the pinna1 mutant in two ways. On the one hand, PINNA1 knockout leads to the reactivation of MtKNOXI expression in leaf primordia, whereas on the other hand, the disruption of PINNA1-MtKNOXI complexes further promotes the retention of MtKNOXI in the nucleus, allowing it to function as a transcriptional factor in the regulation of compound leaf development (Fig. 6).

The regulatory mechanism of KNOXI in leaf development has been studied in the past years. The cytokinin biosynthesis gene, IPT7, the target of KNOXI transcription factor STM in Arabidopsis23, acts downstream of KNOXI in the regulation of compound leaf development in tomato47. In lettuce, LsKN1 promotes leaf complexity by directly binding to the promoters of PINIOD, LsCUC3, LsAS1, and LsGA3ox1 and regulating their expression48. These findings suggest that KNOXI regulates leaf development in multiple ways, such as hormonal pathways and transcriptional regulatory mechanisms. UFO is also the possible target gene of STM during the regulation of embryogenesis and flower meristem identity in Arabidopsis49,50. Additionally, the ectopic expression of UFO and its homologs is able to promote indeterminacy in leaf development, which is evident in both Arabidopsis and C. hirsuta31,32. In this study, we identified the MtUFO/MtKNOXI module that functions in the transition from trifoliate to pinnate-like pentafoliate leaf patterns. In M. truncatula, the genetic and biochemical analysis revealed that the increased activity of MtUFO is necessary for the formation of ectopic leaflets in both pinna1 mutants and MtKNOX7-overexpressing lines, and MtUFO was identified as one of the direct targets of MtKNOX7. Moreover, the homologs of MtUFO, STP in pea (Pisum sativum), and PFO/LjUFO in Lotus japonicus, were found to participate in the leaflet initiation during compound leaf development51,52. Thus, it might be conserved that UFO homologs play a positive role in increasing leaf complexity among legume species.

In M. truncatula, although the subcellular localization of PINNA1 is mainly cytoplasmic, it is also likely recruited in the nucleus through interaction with other co-factors, which might be Class II KNOX homeobox (KNOX II) or other proteins. In Arabidopsis, KNOXII proteins confer opposing activities with the KNOXI protein to suppress leaflet initiation during leaf development53,54. In S. lycopersicum, the members of KNOXII have also been reported to play a role in repressing leaflet formation55. However, whether KNOXII is involved in compound leaf patterning by interacting with BLH in M. truncatula remains an open question. Characterization of genetic relationship of the mutants with pentafoliate leaves and a comparison of regulatory targets between MtKNOXI and MtKNOXII may shed light on their roles in compound leaf patterning.


Plant material and growth conditions

Medicago truncatula ecotype R108 was used as the wild type to compare with Tnt1 insertion mutants. The mutant lines of PINNA1 (Medtr3g112290), MtKNOX1 (Medtr2g024390), MtKNOX2 (Medtr1g017080), MtKNOX6 (Medtr5g085860), MtKNOX7 (Medtr5g033720), SGL1 (Medtr3g098560), and MtUFO (Medtr4g094748) were isolated from a Tnt1 retrotransposon-tagged mutant collection of M. truncatula (Supplementary Table 1). As the sgl1-2 and mtufo-1 mutants were sterile, seeds were produced by the heterozygous parents for propagation. Homozygous lines were isolated from the self-pollination of heterozygous plants and genotyped in each generation. The mtknox1-1, mtknox2-1, mtknox6-1, mtknox7-1 homozygous plants, sgl1-2 heterozygous, and mtufo-1 heterozygous plants were crossed with pinna1-1 homozygous lines to generate double, triple, quadruple, or quintuple mutants in the F1 progeny and were identified on the basis of PCR genotyping in the F2 progeny. The plants were grown in the greenhouse at 22 °C with 150 μmol m−2 s−1 light and 70 to 80% relative humidity under a 16-h (light)/8-h (dark) photoperiod cycle.

Plasmid construction for transgenic plants

The coding sequence (CDS) of MtKNOX1, MtKNOX2, MtKNOX6, MtKNOX7, MtKNOX8, and MtUFO was PCR amplified and cloned into binary vectors pEarleyGate100 or pEarleyGate103-GFP to construct overexpression vectors respectively. These plasmids were introduced into the disarmed Agrobacterium tumefaciens strain EHA105 strain and used for plant transformation via the Agrobacterium-mediated transformation method to generate transgenic plants56. The primers used are listed in Supplementary Data 1.

RNA extraction, RT-PCR, and qRT-PCR experiments

Total RNA was extracted from shoot apices of 6-week-old plants using RNAzol RT RNA Isolation Reagent (MRC, Cat: RN190). Plant materials were thoroughly ground using the Tissuelyser-48 (Shanghai Jingxin). Two-microgram RNA samples were subjected to reverse transcription using the All-In-One 5X RT MasterMix (abm, Cat: G592). Quantitative real-time PCR analysis was conducted on a Bio-Rad CFX Connect real-time PCR detection system (CFX96, Bio-Rad) using FastStart Universal SYBR Green Master (Rox) reagent (Roche, Cat: 4913850001). The relative transcript level was calculated using the 2−ΔΔCt method, with MtUBIQUITIN (Medtr3g110110) serving as the internal control. Each reaction was conducted in triplicate. The primers used are listed in Supplementary Data 1.

Yeast two-hybrid assay

The Matchmaker Gold Yeast Two-Hybrid System (Clontech, USA) was performed in yeast two-hybrid assays. The full-length CDS of MtKNOX1, MtKNOX2, MtKNOX6, MtKNOX7, and MtKNOX8 were PCR amplified and cloned into the pGBKT7 vector to construct the bait plasmids (MtKNOX1-BD, MtKNOX2-BD, MtKNOX6-BD, MtKNOX7-BD, and MtKNOX8-BD), respectively. For yeast two-hybrid screening, MtKNOX7-BD was used as bait to screen a Medicago cDNA library. For the yeast two-hybrid assay, the full-length CDS of PINNA1 was PCR amplified and cloned into the pGADT7 vector to construct the prey plasmid (PINNA1-AD). The bait and prey constructs were transformed into the yeast strain AH109 and grown on SD/-Trp/-Leu plates at 30 °C for 3 d. The interacting transformants were then examined on the selective medium SD/-Leu/-Trp/-His/X-a-gal/15Mm 3-AT. Protein–protein interactions were further visualized by blue color when X-α-gal was used to detect the expression of the LacZ reporter. The MEINOX domains of STM/BP-like KNOXI and MtKNOX7 proteins were respectively divided into three domains, MEINOX1, MEINOX2, MEINOX (MEINOX1 and MEINOX2). Truncated versions of the coding regions of MtKNOX1, MtKNOX2, MtKNOX6, and MtKNOX7 were PCR amplified and cloned into the pGBKT7 vector as the bait constructs, respectively. These bait plasmids were then co-transformed with PINNA1-AD into the yeast strain. The subsequent steps followed the procedure described above. The primers used are listed in Supplementary Data 1.

Bimolecular fluorescence complementation (BiFC) assay

The full-length CDS of MtKNOX1, MtKNOX2, MtKNOX6, and MtKNOX7 were individually cloned into the pEarly201-ccdB-nYFP (YN) vector, while the full-length CDS of PINNA1 was cloned into the pEarly202-ccdB-cYFP (YC) vector. Additionally, the full-length CDS of ATG8e and NBR1 were cloned into the pCAMBIA3300-mCherry vector to serve as markers for autophagosomes or autophagic bodies. Subsequently, the corresponding binary vectors were introduced into Agrobacterium tumefaciens strain GV3101 and co-infiltrated into the leaves of Nicotiana benthamiana to test the interaction. The fluorescence signal was monitored using a Zeiss LSM880 confocal laser scanning microscope (Zeiss, Germany). All primers used are listed in Supplementary Data 1.

Coimmunoprecipitation (co-IP) assay

The full-length CDS of PINNA1, MtKNOX1, MtKNOX2, MtKNOX6, and MtKNOX7 were PCR amplified and cloned into pENTR/D-TOPO cloning vector (Invitrogen, Cat: K240020). Following this, they were transformed into either the pCAMBIA1390-ccdB-7Myc-6His or pEarleyGate103-GFP vector using the Gateway LR Clonase II (Invitrogen, Cat: 11791020). These constructs were expressed individually or co-expressed in the leaves of 4-week-old N. benthamiana plants after Agrobacterium infiltration. Equal amounts of samples (0.3 g) were collected after a 48-h incubation and powdered in liquid nitrogen. Total protein was extracted with 1 mL extraction buffer (20 mM HEPES-KOH (pH 7.5), 150 mM NaCl, 1 mM EDTA, 0.5% Triton X-100, 0.2% β-mercaptoethanol, 10% glycerol, 1 mM phenylmethylsulfonyl fluoride (PMSF), and 1x protease inhibitor cocktail). Samples were incubated at 4 °C for 20 min with gentle shaking and centrifuged twice at 15,000×g for 20 min at 4 °C. The supernatant was incubated with 10 μL of GFP-Trap magnetic agarose beads (Chromotek, Cat: gtma-20) at 4 °C for 4 h by vertical rotation. Subsequently, the beads were washed seven times with extraction buffer and eluted with 2× SDS protein sample loading buffer for 5 min at 95 °C. The supernatants were electrophoretically separated by 10% SDS-PAGE and transferred to a nitrocellulose membrane (Bio-Rad, Cat: 1620115). Immunoblots analysis was performed using a mouse anti-Myc antibody (ABclonal, Cat: AE010, 1:4000 dilution) and a mouse anti-GFP antibody (TransGen Biotech, Cat: HT801-01, 1:4000 dilution). The primers used are listed in Supplementary Data 1.

Subcellular localization assay

For subcellular localization in M. truncatula mesophyll protoplast, full-length CDS of PINNA1, MtKNOX1, MtKNOX2, MtKNOX6, and MtKNOX7 were PCR amplified and cloned into PBI221-GFP or PAN583-mCherry vectors respectively. These constructs were transiently expressed alone or co-expressed in Medicago mesophyll protoplasts using the polyethyleneglycol (PEG)–mediated transformation method57. The infiltrated protoplasts were incubated at 22 °C for 12–16 h in the dark. For subcellular localization in Medicago leaf epidermal cells, PINNA1-GFP, GFP-PINNA1, and 35S-GFP constructs were transiently expressed in leaves by Agrobacterium infiltration. The fluorescence signal was monitored by a Zeiss LSM880 confocal laser scanning microscope (Zeiss, Germany). The primer sequences are listed in Supplementary Data 1.

Scanning electron microscopy (SEM) assay

For SEM, shoot apices samples from 4-to 6-week-old plants were fixed in a fixative solution (3% glutaraldehyde in 1×PBS buffer) under vacuum for 10 min at 4 °C, followed by overnight incubation at 4 °C. The samples were then washed five times in 1×PBS buffer, with each wash lasting 10 min. Subsequently, the samples were dehydrated in graded ethanol (30, 50, 60, 70, 85, 95, and two rounds of 100% ethanol, each for 20 min). After dehydration, the samples underwent critical point drying by carbon dioxide and were then sprayed with gold powder. Finally, the prepared samples were observed by Quanta 250 FEG scanning electron microscope (FEI, USA) at an accelerating voltage of 5 kV.

RNA in situ hybridization

The RNA in situ hybridization treatments were conducted following the previously established methods58. Briefly, 777-bp, 518-bp and 640-bp fragments were isolated and PCR amplified from the CDS of MtKNOX7, PINNA1, and MtUFO, respectively. The resulting PCR products were ligated into the pGEM-T vector (Promega, Cat: A3600) using the primer sequences provided in Supplementary Data 1. Both the sense and antisense probes were labeled with digoxigenin (Roche, Cat: 11175025910) according to the manufacturer’s instructions. RNA in situ hybridization was implemented on shoot apices of 4-week-old wild-type and pinna1-6 plants with the digoxigenin-labeled sense or antisense probes. The fresh shoot apices were fixed in FAA solution at 4 °C and then embedded in paraffin (Leica, Cat: 39601095). After the sectioning and hybridization processes, the 8 μm-thick slices of shoot-apex samples were examined using ECLIPSE Ni Series light microscope (Nikon, Japan).

Nuclear-cytoplasmic fractionation assay

For nuclear-cytoplasmic fractionation, the full-length CDS of PINNA1 was PCR amplified and cloned into PAN583-GFP or PAN583-4Myc vectors respectively. The primer sequences are listed in Supplementary Data 1. The nuclear and cytoplasmic fractions were isolated from protoplasts expressing GFP- or Myc-tagged proteins, following the previously established protocol59. Medicago mesophyll protoplast transformed with plasmids were cultured for 16 h at 22 °C under dark conditions, then resuspended in 1 mL extraction buffer (20 mM Tris-HCl pH 7.5, 250 mM sucrose, 25% glycerol, 20 mM KCl, 2 mM EDTA, 2.5 mM MgCl2, 5 mM DTT, 1× protease inhibitor cocktail, and 0.3% Triton X-100). After incubating on ice for 10–15 min, 100 μL of the mixture was taken as the total protein fraction. The crude cytoplasmic and nuclear fractions were separated by centrifugation at 3000×g for 10 min at 4 °C. The supernatant was centrifuged at 12,000×g for 10 min at 4 °C and collected as the cytoplasmic fraction. The pellet was washed three times at 3000 × g for 5 min with 1 mL of nuclear wash buffer (20 mM Tris-HCl, pH 7.4, 25% glycerol, 2.5 mM MgCl2, 0.1% Triton X-100, and 1× protease inhibitor cocktail). Subsequently, it was resuspended in with 500 μL of NRB2 buffer (20 mM Tris-HCl, pH 7.5, 0.25 M Sucrose, 10 mM MgCl2, 0.5% Triton X-100, 5 mM β-mercaptoethanol, and 1× protease inhibitor cocktail) and carefully overlaid on top of 500 μL NRB3 buffer (20 mM Tris-HCl, pH 7.5, 1.7 M Sucrose, 10 mM MgCl2, 0.5% Triton X-100, 5 mM β-mercaptoethanol and 1× protease inhibitor cocktail). These were centrifuged at 16,000 × g for 30 min at 4 °C. The final nuclear pellet was resuspended in 100 μL extraction buffer. To ensure the quality of the fractions, PEPC protein and Histone H3 were used as cytoplasmic or nuclear markers. Immunoblots analysis was performed using a mouse anti-Myc antibody (ABclonal, Cat: AE010, 1:4000 dilution), a mouse anti-GFP antibody (TransGen Biotech, Cat: HT801-01, 1:4000 dilution), a rabbit anti-Histone-H3 antibody (Proteintech, Cat: 17168-1-AP, 1:2000 dilution), and a rabbit anti-PEPC antibody (PhytoAB, Cat: PHY2038S, 1:2000 dilution).

Transcriptional activation assay

The transcriptional activation assay was conducted in Arabidopsis mesophyll protoplasts using a dual-luciferase reporter (DLR) assay system (Promega, Cat: E1960). The reporter 6×UASpro pGREENII vector contains six GAL4 binding elements (upstream activating sequence, UAS), a Renilla luciferase (REN) gene (as an internal control), and a Firefly luciferase (LUC) gene (as a reporter). For effectors, coding regions of GAL4BD-MtKNOXIs protein fusion were amplified using specific primers from the GAL4BD fusion plasmids of yeast two-hybrid assays and cloned into PBI221 to construct effector plasmids. The coding region of PINNA1 was also amplified and cloned into PBI221 as another effector. The primer sequences are listed in Supplementary Data 1. These constructs were transiently expressed in Arabidopsis protoplasts. The infiltrated protoplasts were kept in the dark for 12–16 h, and the relative LUC and REN activity was measured by the DLR assay system on the Centro XS LB960 (Berthold, Germany). The plasmid ratio for the experimental setup was 6×UASpro pGREENII: BD-MtKNOXI-PBI221: PINNA1-PBI221 = 1: 4: 4 (μg: μg: μg), compared with the control (6XUASpro pGREENII: BD-MtKNOXI-PBI221: PBI221 = 1: 4: 4). Finally, the activity of transactivation was indicated by the ratio of LUC to REN.

Luciferase imaging assay

The full-length CDS of PINNA1, MtKNOX1, MtKNOX2, MtKNOX6, and MtKNOX7 were PCR amplified and fused to either the pCAMBIA3300-GFP or pCAMBIA3300-GR vector to generate effectors, respectively. The 2.9-kb promoter fragment upstream of MtUFO was cloned into the pGreenII-0800-Luc vector to generate a reporter construct. The constructs were introduced into A. tumefaciens GV3101 together with the pSoup19 helper plasmid. At least four tobacco leaves from independent plants were infiltrated and harvested after 48 h. For GR induction, another set of at least four tobacco leaves from independent plants were infiltrated with infiltration buffer containing 10 μM DEX (Coolaber, Cat: SL4070) or an equal volume of ethanol, and harvested after 48 h. The luciferase fluorescence signals were observed by a plant living imaging system (Tanon 5200, China) after these leaves were sprayed with 1 mM d-Luciferin solution (GoldBio, Cat: LUCK-1G) and incubated for 5 min in the dark. Then the rest of the leaves were punched and powdered in liquid nitrogen to measure the LUC and REN activity. Relative LUC and REN activity were measured by the DLR assay system on the Centro XS LB960 (Berthold, Germany). Finally, the relative firefly luciferase activity was calculated as the ratio of LUC to REN for each sample. The primer sequences are listed in Supplementary Data 1.

Electrophoretic mobility shift (EMSA) assay

To express MtKNOX7 in E. coli, the full-length CDS of MtKNOX7 was cloned into the pET28a vector. The MtKNOX7-HIS fusion protein was expressed in E. coli Rosetta (DE3) cells (TransGen Biotech, Cat: CD801-02) by induction with 0.5 mM isopropyl β-d-thiogalactoside (IPTG) at 16 °C for overnight, and then purified with Ni Sepharose 6 Fast Flow (GE Healthcare, Cat: 17531803). For the EMSA assay, two complementary oligonucleotides with 5′ biotin labeling were synthesized and then hybridized to prepare double-strand DNA probes. The EMSA assay was performed using the LightShift Chemiluminescent EMSA Kit (Thermo Fisher Scientific, Cat: 20148) following the manufacturer’s instructions. The primer sequences are listed in Supplementary Data 1.

Chromatin immunoprecipitation (ChIP) assay

For ChIP assays60, plant tissues (1 g) were cross-linked in cross-linking buffer (10 mM Tris-HCl pH = 8, 1 mM PMSF,1 mM EDTA 0.4 M sucrose, and 1% formaldehyde) for 10 min and then quenched the crosslinking reaction with Glycine (0.25 M) for 5 min at room temperature. After being powdered in liquid nitrogen, the pellet was washed twice with Nuclei Extraction Buffer (15 mM PIPES pH 6.8, 5 mM MgCl2, 50 mM KCl, 15 mM NaCl, 1 mM CaCl2, 0.25 M sucrose, 0.4 % Triton X-100, 5 mM PMSF, and 1x protease inhibitor cocktail), and then centrifuged at 6000 × g at 4 °C for 5 min to isolate nuclei. The pelleted nuclei were resuspended in Nuclear Lysis Buffer (50 mM Tris-HCl (pH 8.0), 10 mM EDTA (pH 8.0), 1% SDS, 1 mM PMSF, and 1x protease inhibitor cocktail) and sonicated with Bioruptor to shear the DNA into fragments of 200 to 500 bp. Immuno-precipitate chromatin complexes were incubated with anti-GFP antibody (Abcam, Cat: ab290) at 4 °C overnight, and subsequently mixed with protein A/G magnetic beads (MCE, Cat: HY-K0202) at 4 °C for 2 h. DNA was eluted and purified from the complexes, and the enrichment of DNA fragments was analyzed by quantitative PCR using the primer listed in Supplementary Data 1.

Phylogenetic analysis

To conduct the phylogenetic analysis, KNOX homologs or UFO orthologs were retrieved from PLAZA for most species, respectively. Multiple sequences were aligned using ClustalW online ( with default parameters. The neighbor-joining phylogenetic tree was built using the MEGA11 program (, and the phylogenetic trees with bootstrap values from 1,000 trials were shown. All valid KNOX homologs were listed in Supplementary Data 2,  3, while Supplementary Data 4 included all valid UFO orthologs.

Statistical analysis

Statistical analysis was conducted using the program GraphPad Prism version 9.0 software. An unpaired two-tailed t-test was performed to calculate the P values between two groups, while for comparisons involving three or more groups, a one-way ANOVA with Tukey’s multiple comparisons test was conducted. The results of the statistical analyses are shown in the Source Data.

Accession numbers

Sequence data from this article can be found in GenBank under the following accession numbers: PINNA1 (Medtr3g112290), MtKNOX1 (Medtr2g024390), MtKNOX2 (Medtr1g017080), MtKNOX6 (Medtr5g085860), MtKNOX7 (Medtr5g033720), MtKNOX8 (Medtr1g084060), SGL1 (Medtr3g098560), MtUFO (Medtr4g094748), MtActin (Medtr7g026230), MtUBIQUITIN (Medtr3g110110), NBR1(AT4G24690), and ATG8e (AT2G45170).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.