2.1 Plant materials and growth conditions

The sltcp29/24 double mutant with the background of M82 (Solanum lycopersicum) was screened from the previously constructed CRISPR/Cas9 mutant population (Bi et al. 2023). The T₂ generation homozygous sltcp29/24 mutants were used in this study. Wild-type (WT) tomato plants M82 and sltcp29/24 mutants were grown in temperature-controlled incubators at 25°C ± 1°C during the day and 18°C ± 1°C at night, with a photoperiod containing 16 h of light and 8 h of darkness.

2.2 Leaf phenotyping

The leaf phenotypes of WT and sltcp29/24 mutants were observed 45 days after germination (DAG). The numbers of primary, secondary and intercalary leaflets in successive leaves were counted, then the leaf morphology was recorded using a scanner. At least 7 plants were used for each genotype. The terminal and lateral leaflets of the fifth leaf from WT and mutant plants were scanned, and leaf area and leaf dissection index were calculated using ImageJ software. The method refers to previous research (He et al. 2020). At least 10 leaves from different plants were used to observe and record each genotype. The leaf margin epidermal cell morphology of WT and mutant terminal leaflets was observed under a laser microscope; then, the cell counts and areas were analyzed using Photoshop and ImageJ. At least 5 leaves per genotype were observed and three different fields of vision per leaf were photographed.

2.3 In situ hybridization

The unique sequence of SlTCP29 and SlTCP24 were amplified from the WT cDNA and ligated into pSPT18 and pSPT19 vectors. The constructs were linearized, and T7 or SP6 RNA polymerase was used for the RNA labeling reaction according to the protocol of manufacturer (Roche). Leaf primordia from the m-P6 stage were selected for section preparation, and RNA in situ hybridization was performed as previously described (Nakayama et al. 2021).

2.4 Leaf primordium imaging

The fifth leaf primordium of WT and mutant plants was used for observation. Excess leaf primordia of the plants were trimmed off using a surgical blade to preserve the fifth leaf primordium. Changes in the morphology of the fifth leaf primordium of WT and mutant plants were regularly observed and photographed under a Zeiss digital microscope. At least 6 leaf primordia per genotype were observed at a time.

2.5 Collection of leaf primordia and RNA-seq analysis

The leaf primordia of the WT and mutant seedlings grown in an incubator for approximately 20 days were sampled using a ZEISS digital microscope, and the leaf primordium of m-P6 stage containing meristem was collected as the RNA sequencing sample. At least 0.3 g of leaf primordia were collected for each sample, and three independent samples from each group were collected for sequencing. The RNA-seq was conducted in SeqHealth (Wuhan, China) using the Illumina HiSeq X Ten sequencing platform. The raw data were qualified using Base Calling and Trimmatic software. The clean data were compared to the tomato genome in the Solanaceae Genomics Network (http://solgenomics.net/) to obtain comprehensive transcriptional information. Finally, the absolute value of logFC >1 and P-value <0.05 were used as criteria to screen differentially expressed genes (DEGs).

2.6 RT-qPCR analysis

Total RNA was extracted from different tomato tissues using an RNA pure plant kit (CWBIO). RNA samples of 1 μg were reverse-transcribed into cDNA using the Prime Script™ RT Master Mix (Takara). The tomato ACTIN gene (NCBI: NM_001330119.1) was used as an internal control. Primers are listed in Table S1. RT-qPCR analysis was performed using SYBR Green PCR Master Mix (Takara) on qTOWER3 G (Analytik Jena AG) following the manufacturer's protocol. Relative expressions were analyzed by the 2^-ΔΔCt method (Livak and Schmittgen 2002). Three biological and technical replicates were performed for each assay.

2.7 Yeast one-hybrid assays

The promoter fragments of SlTKNII5, SlMIR164a, and SlACS1A, containing TCP transcription factor-binding motifs were cloned into the pAbAi vector as bait. After digestion with BstBI, the linearized construct was transformed into the Y1H Gold yeast strain using the Yeastmaker Yeast Transformation System 2 (Clontech). The full-length coding sequences (CDS) of SlTCP29 and SlTCP24 were cloned into the pGADT7 vector. Primers are listed in Table S1. The recombinant pGADT7 vector was transformed into Y1H Gold(bait/AbAi) strains and spread on SD/−Leu medium. Subsequently, positive DNA-protein interaction factors were screened based on the protocol of the manufacturer (Clontech).

2.8 Dual-luciferase assays

The full-length CDS of SlTCP29 and SlTCP24 were constructed into the pCAMBIA1300 vector driven by the CaMV 35S promoter as the effector. The SlTKNII5, SlMIR164a and SlACS1A promoter fragments were respectively inserted into the pGreen II 0800-LUC vector. Primers are listed in Table S1. The pCAMBIA1300 vector was used as a negative control. The recombinant plasmids were transformed into Agrobacterium tumefaciens GV3101 competent cells. The effector and reporter were mixed at a ratio of 1:1; then the mixture was infiltrated into tobacco leaves using a syringe. After 72 h incubation, the activities of firefly luciferase and renilla luciferase were measured using the dual-luciferase reporter assay system (Promega) on an Infinite 200 pro luminometer (TECAN). For firefly luciferase imaging, 1 mM D-Luciferin was sprayed on the abaxial sides of tobacco leaves. The leaves were kept in the dark for 10 min, and then luciferase images were captured using the CCD imaging system.

2.9 Electrophoretic mobility shift assay (EMSA)

The full-length CDS of SlTCP29 and SlTCP24 were cloned into the pGEX 6P-1 vector containing a GST tag and the pET-32a vector containing a His tag, respectively. Primers are listed in Table S1. Escherichia coli strain Rosetta2 cells were used to express the fusion proteins. The recombinant protein was incubated at 16°C for 16 h with 0.25 mM IPTG, and then purified using a GST-labeled purification kit (Beyotime). The probes of the SlTKNII5, SlMIR164a and SlACS1A promoters were synthesized and the binding site was labeled with biotin. The unlabeled probes were used as the cold competitors. The mutated biotin probes were used to detect binding specificity. GST protein and His protein were used as the negative controls. The EMSA reactions were conducted according to the instructions of the Chemiluminescent EMSA Kit (Beyotime).

2.10 Yeast two-hybrid assays

To test the interaction of KNOXII members at the protein level, the full-length CDS of SlTKNII5 was constructed into the pGBKT7 vector as bait, and the full-length CDS of SlTKNII3/4/7 were each constructed into the pGADT7 vector as prey. Primers are listed in Table S1. The recombinant plasmids were co-transformed into Y2H Gold yeast strain according to Yeastmaker Yeast Transformation System 2 (Clontech), and the transformants were cultured on SD/−Leu/−Trp (DDO) medium for 3–5 days. Positive transformants were then screened on SD/−Leu/−Trp/−Ade/-His (QDO) medium containing X-α-gal and AbA.

2.11 Firefly luciferase complementation imaging assays

The full-length CDS of SlTKNII5 with the stop codon removed was cloned into the pCAMBIA1300-nLUC vector, and the full-length CDS of SlTKNII3/4/7 were cloned into the pCAMBIA1300-cLUC vector. Primers are listed in Table S1. Then, the recombinant plasmids were transformed into Agrobacterium tumefaciens GV3101. Different combinations were mixed in a ratio of 1:1; then, the mixture was infiltrated into the tobacco leaves using a syringe. The plants grew in the dark for one day and in normal light conditions for two days. Finally, the fluorescence signal was captured using the CCD imaging system. At least three replicates of each combination were conducted for fluorescence detection.

2.12 Accession numbers

Tomato genes were collected from the Solanaceae Genomics Database (http://solgenomics.net/) under the following accession numbers: SlTCP29, Solyc08g048370. 2; SlTCP24, Solyc08g048390.2; GOB, Solyc07g062840.2; SlTKN1, Solyc04g077210.2; SlTKN2, Solyc02g081120.2; SlTKN3, Solyc05g005090.2; SlTKN4, Solyc01g100510.2; SlTKNII3, Solyc07g007120.2; SlTKNII4, Solyc12g010410.1; SlTKNII5, Solyc08g041820.2; SlTKNII7, Solyc08g080120.2; SlACS1A, Solyc08g081550.2; and SlMIR164a, MI0027570.

3.1 Simultaneous mutations in SlTCP29 and SlTCP24 affect the morphology of tomato compound leaves

We obtained a double-gene mutant with a frameshift mutation in the SlTCP29 gene and a codon mutation in the SlTCP24 with one amino acid deletion in the TCP domain from our previously constructed tomato transcription factor CRISPR/Cas9 mutant population (Figure S1a). SlTCP29 and SlTCP24 belong to the CIN-like TCP transcription factor subfamily and have high genetic homology (Figure S1b). The compound leaf morphology was the most obvious change in plant architecture caused by the changes in SlTCP29 and SlTCP24 protein sequences. It is worth mentioning that the leaf phenotype of our double mutant is consistent with that of the sltcp29/24 double mutant in the background of different tomato varieties reported in previous studies (Hu et al. 2023). In order to further investigate the specific details of the effect of sltcp29/24 mutants on leaf morphology, we collected scanning observations of leaf morphology and statistics of leaflets per leaf unit for the first six units in WT and sltcp29/24.

Initially, we observed the additional leaf margin serrations in the fully expanded cotyledons of sltcp29/24. Statistics on all types of leaflets in each leaf unit displayed that the number of primary leaflets of WT and mutant remained essentially the same. However, with leaf development, the mutant developed more secondary leaflets on the primary leaflet petiolule and intercalary leaflets on the rachis, resulting in a significantly higher number of leaflets in each leaf order in the mutant than the WT, and a more complex morphology of the compound leaves (Figures 1a,b and S2). The result that simultaneous mutation of SlTCP29 and SlTCP24 increased the number of leaflets in tomato is consistent with previous reports (Hu et al. 2023).

With the gradual differentiation of leaves, the leaflet margin was serrated more obviously in the mutant than in the WT, changing from lobed leaves into deeply serrated leaves (Figure 1c). Analyses of WT and mutant single-leaf complexity revealed that the leaf dissection index of both the terminal and lateral leaflets of the mutant was significantly higher than that of WT (Figure 1d). However, the leaf area of single leaflets of mutant primary leaflets was slightly reduced; but not significantly compared to the WT (Figure 1e). Observations of the epidermal cells of functional leaves that had completed differentiation revealed a significant reduction in the area of mutant leaf epidermal cells per unit field of view and a significant increase in their number compared to the WT (Figure 1f-h). The above results suggest that SlTCP29 and SlTCP24 synergistically maintain the deterministic growth of tomato leaflets and leaf margin normalization.

3.2 Temporal and spatial expression of SlTCP29 and SlTCP24 in the development of tomato compound leaves

We divided leaf development into m-P3 to P10 stages and then conducted RT-qPCR analysis to explore the expression patterns of SlTCP29 and SlTCP24 during compound leaf development (Figure S3). The results showed that SlTCP29 expression gradually increased with leaf primordium development, with the highest expression at the P7 stage, followed by a slight decrease in expression level, but continued to be expressed at the stage of P8-P10 (Figure 2a). In contrast, SlTCP24 was expressed at all stages of leaf development, and its expression level was higher in the m-P6 to P8 stages (Figure 2b).

To further investigate the spatial expression of SlTCP29 and SlTCP24, we performed in situ hybridization analysis on the leaf primordium at the WT m-P6 stage. The results revealed that SlTCP29 and SlTCP24 were expressed in similar locations, with expression detected on the adaxial side of the leaf primordium and initiating leaflet primordium (Figure 2c, d, e), as well as in the leaf margin meristem in differentiation (Figure 2f, g). In addition, the expression of SlTCP29 and SlTCP24 was also detected in the SAM as well as in the floral meristem (Figure 2h-k), implying that SlTCP29 and SlTCP24 may play a role in the maintenance of meristematic tissues and floral bud differentiation. These above results indicate that SlTCP29 and SlTCP24 begin to play a role in the early stage of leaf primordium development and continue to accompany the various stages of leaf differentiation.

3.3 Simultaneous mutation of SlTCP29 and SlTCP24 delays the differentiation of tomato leaf primordium

The elaboration of tomato compound leaf requires maintaining a transient indeterminate growth in the MBs of leaf primordium. To further investigate the impact of sltcp29/24 on leaf primordium development in tomato, we conducted morphological observations on the differentiation process of the fifth leaf (L5) primordium in both WT and sltcp29/24.

After 6 days of germination, the L5 primordium of WT and sltcp29/24 began to initiate (Figure 3a, f). After 8 days of germination, the growth of L5 primordium in sltcp29/24 was slower than in WT (Figure 3b, g). Twelve days after germination, the first leaflet primordium on L5 primordium in WT was differentiated, the second leaflet primordium began to initiate, and the leaf margins of the terminal leaflets also started to differentiate. In contrast, the first leaflet primordium on L5 primordium in sltcp29/24 began to initiate (Figure 3c, h). 16 days after germination, the differentiation of the L5 leaflet primordium was further evident in WT. Moreover, the leaflet primordium on L6 primordium was also initiated. Whereas the second leaflet primordium on L5 primordium in sltcp29/24 began to initiate, the leaflet primordium on the L6 primordium has not yet differentiated (Figure 3d, i). After 20 days of germination, the terminal leaflet of L5 in WT was distinctly differentiated and began to gradually enter the secondary morphogenesis stage. In contrast, the L5 primordium in sltcp29/24 remained in the morphogenetic stage (Figure 3e, j). These results suggest that simultaneous editing of SlTCP29 and SlTCP24 prolongs the duration of leaf primordium differentiation and decelerates the transition of leaf development from the primary to the secondary morphogenetic stage. This prolongation of transition time may be responsible for the increased indeterminate growth of compound leaves in the sltcp29/24 mutant.

3.4 Differential gene expression analysis in the leaf primordia between sltcp29/24 and WT

We analyzed RNA-seq in the leaf primordium containing the apical meristem from sltcp29/24 mutant and WT to investigate the transcriptional regulatory pathways of SlTCP29 and SlTCP24 in regulating leaf development. A total of 2,185 genes were differentially expressed; 763 up-regulated and 1,422 down-regulated (Figure 4a, b). Genes associated with the cell cycle and proliferation were up-regulated in mutants compared to WT. In contrast, genes associated with cell cycle inhibition and promotion of cell expansion were down-regulated in mutants (Figure 4c). The differential expression of some transcription factors associated with leaf margin establishment, such as GOB, and leaf elaboration, such as KNOXI genes, suggests that SlTCP29 and SlTCP24 may recruit multiple factors regulating tomato compound leaf development.

We observed that members of the Class I KNOX associated with leaf morphogenesis were generally up-regulated in the mutant, whereas the expression of the Class II KNOX gene LeT12/SlTKNII3, functionally antagonistic to Class I KNOX, was down-regulated (Figure 4c). To verify the accuracy of the RNA sequencing, we performed RT-qPCR analysis of several leaf development-related genes, specifically members of the KNOXI genes, and the results were consistent with the trend of the transcriptomic data (Figure 4d). These results imply that the increased leaf complexity caused by sltcp29/24 mutations may be due to the up-regulation of morphogenesis-related KNOXI members and the leaflet boundary gene GOB during early leaf primordium development, as well as the down-regulation of KNOXII genes which may promote organ differentiation.

3.5 SlTCP29 and SlTCP24 can promote SlTKNII5 transcription by binding to its promoter

The KNOXII genes are expressed in differentiated organs and exert an inhibitory effect on leaf complexity (Furumizu et al. 2015; Wang et al. 2023). Our RT-qPCR analysis found that four KNOXII members were significantly down-regulated in the sltcp29/24 mutant compared to WT (Figure 5a). Differential expression of other members, except for the SlTKNII3 gene, was undetected in the transcriptome, possibly due to their low abundance of transcripts.

We analyzed the promoter sequences of the KNOXII members to clarify whether SlTCP29 and SlTCP24 promote leaf differentiation by regulating the KNOXII genes' transcription. We found that only the SlTKNII5 promoter contained a classical CIN-TCP transcription factor binding element (Figure 5b). To verify whether SlTCP29 and SlTCP24 can bind to the SlTKNII5 promoter, we performed dual luciferase assays and firefly luciferase imaging. The results demonstrated that SlTCP29 and SlTCP24 could transcriptionally activate SlTKNII5 expression in vivo (Figure 5c, d). Yeast one-hybrid and EMSA assays further demonstrated that SlTCP29 and SlTCP24 bind directly to the ‘GGACCA’ motif in the SlTKNII5 promoter (Figure 5e, f). RT-qPCR analysis indicated that SlTKNII5 expression gradually increased with the development of leaf primordia and peaked at the P8 leaflet stage, with a similar expression trend as that of SlTCP29, suggesting that SlTCP29 is co-expressed with SlTKNII5(Figure 5g).

To further validate the function of SlTKNII5 in leaf development, we obtained sltknII5 mutants using CRISPR/Cas9. However, we did not observe significant leaf morphology changes in the single-gene mutant (Figure S4). Previous studies have shown that heterodimers can be formed among KNOX family members to function (Zeng et al. 2021; Chen et al. 2023). The results of our yeast two-hybrid and firefly luciferase complementation imaging assays also indicate that SlTKNII5 could physically interact with SlTKNII3/4/7 (Figure 5h, i). Thus, these results suggest that SlTCP29 and SlTCP24 can directly activate the transcription of SlTKNII5, which may be genetically redundant with other KNOXII members, and that mutations in SlTKNII5 may affect only a portion of these protein complexes, and that there is a potential for the formation of homo- or heterodimeric combinations between other KNOXII members as well.

3.6 SlTCP29 and SlTCP24 promote the expression of tomato SlMIR164a by directly binding to its promoter

The tomato GOB gene is expressed at the initiating leaflet boundary and plays a role in the specification of developing leaflet separation and boundary (Berger et al. 2009). Our RNA-seq data revealed up-regulation of GOB gene expression in the sltcp29/24 mutant (Figure 4d), whereas the promoter sequence of GOB gene does not contain the classical motif bound by CIN-TCPs. Previous studies have shown that TCP3 in Arabidopsis can activate miR164a expression, and that miR164a is a negative regulator of the boundary gene CUC2(Koyama et al. 2010). Moreover, miR164 has been reported to be a negative regulator of GOB in tomato (Arazi et al. 2005; Berger et al. 2009). We speculated that SlTCP29 and SlTCP24 may indirectly regulate the expression of GOB by regulating the miR164 level, and consequently affect leaf margin serration morphology.

To confirm this hypothesis, we analyzed the upstream DNA regions of two precursor genes of tomato miR164 and identified a CIN-TCPs binding site in the promoter of SlMIR164a gene (Figure 6a). Subsequently, dual luciferase reporter assays and firefly luciferase imaging confirmed that SlTCP29 and SlTCP24 promoted SlMIR164a transcription (Figure 6b, c). Yeast one-hybrid and EMSA assays further confirmed that SlTCP29 and SlTCP24 could bind to the ‘GGACCA’ motif in the SlMIR164a promoter (Figure 6d, e). These results suggest that SlTCP29 and SlTCP24 may indirectly repress GOB gene expression at the leaf margin through direct transcriptional activation of SlMIR164a expression to adjust miR164 level, thus affecting the depth of the leaf margin tooth.

3.7 SlTCP29 and SlTCP24 may indirectly inhibit the expression of KNOXI genes by affecting ethylene biosynthesis

The balance between morphogenesis and differentiation of compound leaves has been shown to involve the engagement of several phytohormones (Ali et al. 2020; Israeli et al. 2021). In our RNA-seq data, several ethylene biosynthesis-related genes 1-AMINOCYCLO-PROPANE-1-CARBOXYLIC ACID (ACC) SYNTHASES (ACSs) and ACC-oxidase 1(ACO1) were significantly down-regulated in sltcp29/24 mutant (Figure 7a). Hormone content measurements at the shoot apical leaflets, showing that the content of ACC, the precursor of ethylene biosynthesis, was significantly down-regulated in the sltcp29/24 mutant (Figure 7b). The hormone level was consistent with gene expression.

Previous studies have shown an antagonistic effect between ethylene signaling and KNAT2 protein in SAM, and ethylene may regulate meristem activity (Olivier et al. 2002). We further found that exogenous ACC treatment could decrease the expression of SlTKN1 and SlTKN2 in tomato stem tips (Figure 7c). We hypothesized that SlTCP29 and SlTCP24 may indirectly inhibit the expression of KNOXI genes during leaf primordium development by affecting ethylene biosynthesis. To confirm this hypothesis, we analyzed the promoter sequences 2000 bp upstream of the start codon of the differentially expressed SlACSs and SlACO1. Two potential CIN-TCP binding cis-elements were identified in the SlACS1A promoter (Figure 7d). Dual luciferase reporter analysis and firefly luciferase imaging demonstrated that SlTCP29 and SlTCP24 can transcriptionally activate SlACS1A expression in vivo (Figure 7e, f, g). Yeast one-hybrid and EMSA assays indicated that SlTCP29 and SlTCP24 can bind directly to ‘GGACCA’ and ‘TGGTCC’ motifs on the SlACS1A promoter (Figure 7h, i). These results suggest that SlTCP29 and SlTCP24 directly transcriptionally activate the expression of SlACS1A, further increasing the accumulation of ACC in shoot apical leaflets. SlTCP29 and SlTCP24 may indirectly affect the expression of KNOXI genes in leaf primordium through the ethylene biosynthesis pathway.