The WOX family transcriptional regulator SlLAM1 controls compound leaf and floral organ development in Solanum lycopersicum

The LAM1 transcription factor regulates expansion of primary leaflets in the compound leaves of tomato, and also affects floral organ development, fruit size, and initiation of secondary leaflets.


Introduction
Leaves and floral organs are typical lateral organs and originate from the flanks of the shoot apical meristem (SAM) and the floral meristem, respectively. These determinate lateral organs possess inherent polarity, with the adaxial side of the lateral organ primordium being the side closest to the meristem and the abaxial side being away from it (Eshed et al., 2001;Bowman et al., 2002). The HD-ZIPIII, KANADI, YABBY, MYB, and LOB domain proteins together with small RNAs, KNOX, and WUSCHEL-related homeobox (WOX) proteins play important roles in this process in plants (Laux et al., 1996;Waites et al., 1998;Kerstetter et al., 2001;Husbands et al., 2009;Ikeda et al., 2009;Moon and Hake, 2011;Sakakibara et al., 2014;Youngstrom et al., 2019). The WOX family, a large group of transcription factors, belongs to the homeodomain (HD) super family that has a typical DNA-binding domain of ~60 amino acids (Gehring et al., 1994;van der Graaff et al., 2009;Burglin and Affolter, 2016). The WOX homeodomain (WOX HD) is slightly larger (~65-70 aa) due to extension at the C-terminus of the HD (Haecker et al., 2004;van der Graaff et al., 2009;Tadege et al., 2011;Catarino et al., 2016;Romani et al., 2018). Previous reports have shown that WOX genes can be divided into three clades based on their phylogenetic relationships and conserved domains, namely the ancient clade, the intermediate clade, and the WUS/modern clade (Haecker et al., 2004;van der Graaff et al., 2009). The WOX transcription factors in Arabidopsis have four main specific motifs in addition to the HD that is conserved in all WOX family members, and these are the WUS box, the EAR-like motif, the STF box, and the MAEWEST/WOX4 box (Haecker et al., 2004;van der Graaff et al., 2009;Ikeda et al., 2009;Tadege et al., 2011). Members of the WUS/modern clade have a particular WUS box near the C-terminus that is required for repressive activity, and this distinguishes the modern clade members from the rest of the WOX family (Haecker et al., 2004;Ikeda et al., 2009;Lin et al., 2013). The other three motifs are found in some specific subclades of the WUS clade. An EAR-like motif is found at the C-terminus of the WUS, WOX5, and WOX7 homologues comprising the WUS and WOX5 subclades (Vandenbussche et al., 2009), and this might contribute to repressive activity, but its contribution is dispensable for at least the WUS function (Ikeda et al., 2009). The MAEWEST/WOX4 box is located at the N-terminus of the HD and exists only in the WOX1 and WOX4 homologues, but its function is unknown (Costanzo et al., 2014). The STF box is found at the C-terminus of only WOX1 and WOX6 homologues and has been shown to be required for the STF (STENOFOLIA) repressive function in leaf and flower development (Zhang et al., 2014).
In Arabidopsis, the wox1 single-mutant has no discernible phenotype (Haecker et al., 2004;Vandenbussche et al., 2009) and WOX3 (also called PRS) is required for development of lateral stipules in leaves and lateral sepals and stamens in flowers (Matsumoto and Okada, 2001;Shimizu et al., 2009), but wox1 prs double-mutants display narrow leaves and floral organs that are affected in lateral expansion of the leaf blade and in fusion of petals (Vandenbussche et al., 2009;Nakata et al., 2012). This indicates that WOX1 and PRS redundantly regulate the expansion of lateral organs including leaves and flowers. However, this redundancy appears to be specific to Arabidopsis because WOX1 homologues in other eudicot species display strong leaf and flower phenotypes as single-mutants. The WOX1 homologue genetic mutants lam1 in Nicotiana sylvestris, maw (maewest) in petunia, stf in Medicago, and lath (lathyroides) in pea show narrow leaf blades and petals (McHale, 1992;Vandenbussche et al., 2009;Tadege et al., 2011;Zhuang et al., 2012), indicating non-redundant WOX1 function in the expansion of lateral organs in these species. In addition, the Medicago WOX3 homologue mutant loose flower (lfl) is affected in flower development but not in outgrowth of the leaf blade, and the stf lfl doublemutant is identical to stf in the leaf-blade phenotype (Niuet al., 2015), indicating that LFL/MtWOX3 has no redundant function with STF in blade development. Interestingly, in all of the WOX1 homologue mutants, the growth defects appear to be specific to the medial-lateral axis and are variable in strength without a significant effect on leaf length and complexity. For example, the lam1 mutant is extremely narrow in width with almost no blade tissue, but the blade length appears normal (McHale, 1992). The lam1 mutant is also non-bolting and nonflowering under standard growth conditions (McHale, 1992;Tadege et al., 2011). Similarly, the wox1 prs, maw, stf, and lath mutants do not display obvious proximal-distal defects, but the blade phenotypes are relatively weak compared to lam1, and they flower normally. Arabidopsis, N. sylvestris, and petunia have simple leaves, which makes it difficult to evaluate the effect of WOX1 and its homologues on leaf complexity; in contrast, Medicago and pea have compound leaves, but the trifoliate identity in stf mutants (Tadege et al., 2011) and multifoliate identity in lath mutants (Zhuang et al., 2012) are indistinguishable from their respective wild-types under standard growth conditions. In addition, both stf and lam1 (induced to flower by high temperature) mutants are female-sterile due to defective ovule development, but the wox1 prs, maw, and lath mutants are fertile, albeit with reduced fertility (Vandenbussche et al., 2009;Nakata et al., 2012;Zhuang et al., 2012). These observations indicate that the function of WOX1 and its homologues is generally conserved in mediating lateral outgrowth, but also show that there is considerable specificity in the different eudicot species examined, suggesting a dynamic role.
WOX1 homologues are found only in eudicots and the ancestral species Amborella trichopoda, and not in monocots and other taxa (Vandenbussche et al., 2009;Tadege et al., 2011;Zhang et al., 2014). In monocots, leaf-blade expansion is controlled by WOX3 homologues. In maize, the WOX3 homologues NARROW SHEATH 1 and 2 (NS1 and NS2) redundantly regulate blade outgrowth, and the double-mutant shows a strong blade ablation phenotype (Nardmann et al., 2004). Similarly in rice, the NARROW LEAF2 (NAL2) and NAL3 genes encode the WUSCHEL-related homeobox3 (OsWOX3A) protein (Cho et al., 2013;Ishiwata et al., 2013), and the nal2 nal3 double-mutant shows strong pleotropic phenotypes in leaf, spikelet, tiller, and lateral root development (Cho et al., 2013). Likewise, NARROW LEAFED DWARF1 (NLD1), encoding a WUSCHEL-related homeobox3 protein, controls the development of lateral organs in barley, as demonstrated by the narrow leaf-blade phenotype of the nld1 mutant (Yoshikawa et al., 2016). But these monocots have simple leaves, and it is unclear if WOX3 homologues control leaf complexity. In addition, WOX1/6 orthologs are expressed in different stages of young leaf primordia but are absent from the SAM, and are also enriched at the adaxial-abaxial boundary layer (Vandenbussche et al., 2009;Tadege et al., 2011;Nakata et al., 2012;Zhuang et al., 2012;Niu et al., 2018). In monocots, expression of WOX3 orthologs is enriched in shoot meristems and in the marginal edges of leaf primordia (Nardmann et al., 2004;Cho et al., 2013;Yoshikawa et al., 2016), whilst in both dicots and monocots species, abundant transcripts of WOX1/3/6 are detected in different reproductive tissues. The not-strictlyconserved expression pattern of WOX1/3/6 implies potential functional specificities. Despite the WOX1 and WOX3/PRS redundancy in Arabidopsis, the WOX1 function in regulating medial-lateral expansion of lateral organs appears to be specific to eudicots, and this function is taken over by WOX3 homologues in monocots, further fueling the hypothesis that WOX1 function is evolutionarily dynamic. However, whether WOX1 homologues in compound leaf species outside of legumes also play a role in regulating leaf complexity is unknown.
Here, we examined the WOX1/STF/LAM1 orthologous gene, SlLAM1, from the compound-leaf model species tomato, Solanum lycopersicum, and determined that it functions in controlling both leaf outgrowth and complexity. The loss-of-function mutant of SlLAM1 displayed fewer and much smaller secondary leaflets, and also had defects in the outgrowth of the mediolateral axis of leaves and flowers, indicating that SlLAM1 is involved in secondary leaflet initiation in the Solanaceae in addition to the conserved function in promoting lateral organ expansion. Our data shed new light on the dynamic role of WOX1 functioning and its contribution to the evolution of eudicot leaf architecture.

Plant materials and growth conditions
Plants of tomato (Solanum lycopersicum, cv. Ailsa Craig) and woodland tobacco (Nicotiana sylvestris) were grown from seed in a greenhouse under 16/8 h light/dark conditions (150 μE m -2 s -1 ) at 24/20 °C with relative humidity of 50-60%. The plants were well watered and supplied with adequate nutrients.

Vector construction, plant transformation, and genotype analyses
The full-length coding sequence of SlLAM1 was amplified and then cloned into the pCAMBIA3301 vector to generate the 35S::SlLAM1 construct, which was transformed into Agrobacterium tumefaciens EHA105 strain for transformation into the N. sylvestris lam1 mutant (Van Eck et al., 2006). Two targets for CRISPR/Cas9-mediated editing of SlLAM1 were designed by using the CRISPR-GE tool (http://skl. scau.edu.cn/) (Ma et al., 2015). The gRNAs were amplified and cloned into the pYLCRISPR/Cas9P 35S -N binary vector using the Golden Gate method (Ma et al., 2015). The resulting pYLCRISPR/Cas9P35S-N-SlLAM1 construct was used for Agrobacterium (EHA105 strain)mediated transformation of tomato (Van Eck et al., 2006). Genomic DNA was isolated from young leaves of T 0 and T 1 transgenic plants for PCR amplification, and the products were sequenced to verify the mutation status. The T 1 plants of CR-Sllam1-1 with genotypes that were homozygous for Target 2 (9 bp or 11 bp deletion) were chosen to perform phenotype analyses. For RNA-seq experiments, shoot apices fromT 1 plants with mutant phenotypes (including bi-allelic ones) were collected.

Real-time quantitative PCR
Total RNA from different tissues of the tomato and tobacco wildtypes was extracted using TransZol regent (TransGen), and samples of 2 μg were reverse-transcribed using a TransScript II One-Step gDNA Removal and cDNA Synthesis SuperMix kit (TransGen) according to the manufacturer's instructions. Real-time quantitative PCR (RT-qPCR) was performed using a TransStart Tip Green qPCR SuperMix kit (TransGen) and a LightCycler480II device (Roche). The primer sequences used are given in Supplementary Table S1. The ACTIN genes of tomato and tobacco were employed as the internal controls; tomato Tubulin and tobacco Ubiquitin were also used as alternative references in some experiments. Three biological replicates were performed for each sample, together with three technical replicates for each biological one. All the primers used in this study are listed in Supplementary Table S1.

SEM analysis
Shoot apices and leaves from 1-month-old seedlings were subjected to vacuum infiltration in a fixative solution of 5% formaldehyde, 5% acetic acid, and 50% ethanol for 30 min and then kept at room temperature overnight. Before observations, the tissues were dehydrated using an ethanol series (45%, 55%, 65%, 75%, 85%, 90%, and 95%) with each step lasting for at least 1 h, and with a final step of 100% ethanol overnight. The ethanol was removed by drying in liquid CO 2 using a Samdri critical-point dryer, and the tissues were then dissected under a stereomicroscope (SZX16, Olympus). The samples were sprayed with gold and scanned at 5 kV using an EVOLS10 device (Zeiss).

RNA in situ hybridization
RNA in situ hybridization was performed using 8-μm sections from shoot apices of 3-week-old plants according to the method previously described by Coen et al. (1990). The full-length CDS of SlLAM1 was amplified for generating DIG-labelled probes. The signals were visualized with an Olympus BX63 microscope under the DIC channel.

Histological sectioning
Mature leaves were collected and fixed overnight in FAA solution (formalin/acetic acid/alcohol). The samples were then twice dehydrated sequentially in 50% alcohol, 100% alcohol, isopropanol, and n-butanol solutions for 4-6 h, and then submerged in glycol methacrylate resin following the protocol of Feder and O'Brien (1968). A Leica microtome was used to cut 2-μm sections, which were stained with Schiff reagent and Toluidine Blue for visualization under an Olympus BX63 microscope.

Detection of pollen viability
Matured anthers were submerged in Alexander staining solution (Peterson et al., 2010) for 30 min in darkness, and then observed under an Olympus BX63 stereomicroscope.

RNA-seq analysis
A total of 20 shoot apex samples from 3-week-old seedlings of both the wild-type and CR-Sllam1-1 (T 1 ) plants were harvested with three biological replicates. Total RNA was extracted using TRIzol™ Reagent (Invitrogen) according to the manufacturer's protocol. RNA quality was assessed on an Agilent 2100 Bioanalyzer. The cDNA libraries were constructed using a NEBNext® Ultra™ RNA Library Prep Kit for Illumina® (New England Biolabs) and then sequenced using an Illumina HiSeq2500 by the Gene Denovo Biotechnology Co. (Guangzhou, China). Clean reads were obtained after filtering and were mapped to the reference genome of tomato from the Tomato Sol Genomics Network (SGN) database (https://solgenomics.net/) using HISAT2, and the StringTie software was used to assemble the mapped reads and to quantify the expression of each gene as values of fragments per kilobase of transcript per million mapped reads (FPKM) (Kim et al., 2015;Pertea et al., 2015). Genes with P<0.05 and an absolute fold-change value ≥2 were considered as differentially expressed genes (DEGs),as determined using the DESeq2 software (Love et al., 2014). The free online OmicShare tools (www.omicshare.com) were used for Gene Ontology (GO) analysis and examination of enrichment of Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways of the DEGs, and for violin plots and sample-to-sample correlation analysis, with their default settings. The expression levels of 26 DEGs that were selected for their possible roles in leaf development were verified using RT-qPCR.

Results
SlLAM1 is the ortholog of STF and LAM1 There are 15 WOX genes in Arabidopsis and several of them have been reported to play important roles in leaf and floral organ development, embryogenesis, and stem cell maintenance in shoot and root apical meristems (Laux et al., 1996;Haecker et al., 2004). Out of these 15 genes, WOX1 and WOX3/PRS redundantly control leaf-blade outgrowth (Vandenbussche et al., 2009;Nakata et al., 2012), and PRS is reported to function in the recruitment of leaf primordial founder initials from the SAM (Nardmann et al., 2004). If this assumption is true, WOX1 and PRS may also be expected to be involved in the control of leaflet numbers in compound leaf species. However, in M. truncatula, a legume species with compound leaves, leaf complexity appears to be unaffected in the stf loss-of-function mutant, but STF/MtWOX1 and LFL/MtWOX3 function in separate pathways (Niuet al., 2015). To gain better insights about the independent functions of WOX1 homologues in leaf-blade outgrowth and possible contributions to leaf complexity, we studied the tomato WOX1 homologue SlLAM1. With its sympodial stem growth habit and cymose inflorescences, tomato is a compound leaf species in the Solanaceae and is an excellent model to study leaf, inflorescence, and fruit development, in which the functions of WOX genes are likely to be apparent. To verify the WOX proteins in tomato, we used the sequences of Arabidopsis WOXs to conduct a BLAST search of the tomato genome using an E-value of <10 -5 as the cut-off criterion, and this identified 10 typical WOX proteins (Supplementary  Table S2). We also probed the M. truncatula, pea, C. sativus, N. Sylvestris, rice, and maize genomes, in which mutants of WOX1 or WOX3 have been reported, and identified a total of 106 WOX genes, representing species from the Fabaceae, Cucurbitaceae, Poaceae, and Solanaceae. We then conducted a phylogenetic analysis of the corresponding WOX proteins, which included 15 from Arabidopsis, 11 from C. sativus, 15 from M. truncatula, 15 from pea, 11 from rice, 18 from maize, 10 from tomato, 10 from N. sylvestris, and one from petunia (MAW), using the IQ-TREE software (Nguyen et al., 2015). This analysis indicated that whilst the ancient clade was represented by just one subclade (WOX10/13/14), the intermediate clade was divided into two subclades (WOX8/9 and WOX11/12), and the WUS/Modern clade into six subclades (WUS, WOX1/6, WOX2, WOX3, WOX4, and WOX5/7) (Fig. 1A, Supplementary Fig. S1), consistent with the phylogeny of the WOX transcription factor family in plants (Wu et al., 2019). The 10 SlWOX members could also be grouped into three clades: the WUS/modern clade (7), the intermediate clade (2), and the ancient clade (1), consistent with the phylogeny of Arabidopsis WOX proteins (van der Graaff et al., 2009). The WOX3 subclade was represented by two members and each of the other eight subclades was represented by a single SlWOX (Fig. 1B, Supplementary Fig. S1). Based on this comprehensive phylogenetic analysis, we identified SlLAM1 (Solyc03g118770) in tomato as being a singlecopy gene closely related to LAM1, MAW, and STF (Fig. 1B).

SlLAM1 is strongly expressed in leaves, fruit, and flowers
To determine the function of SlLAM1, we first examined its expression pattern using real-time quantitative PCR (RT-qPCR). SlLAM1 was expressed widely in the vegetative and reproductive organs, with relatively high levels of transcript accumulation detected in leaf primordia at the P4-P6 stages, in flowers, and in young fruit ( Fig. 2A). To examine expression patterns in the early stages of leaf and flower development in more detail, RNA in situ hybridization assays were used. Transcripts of SlLAM1 were found at relatively low levels in leaflet primordia at the P1-P3 stages, but not in the SAM (Fig. 2B-G).Subsequently, enrichment was observed in the adaxial-abaxial boundary layer of the leaf (P4), and primary leaflet primordia (Fig. 2D, E) and floral organ primordia (Fig. 2F), similar to that of STF in M. truncatula (Tadege et al., 2011;Zhang et al., 2014). These results suggested that SlLAM1 may function in leaf and flower development, consistent with WOX1 functioning in other eudicots.

Overexpression of SlLAM1 can rescue the leaf phenotype of the tobacco lam1 mutant
To further investigate the function of SlLAM1 in Solanaceae, we conducted a complementation test by generating a 35S::SlLAM1 construct and transforming it into the lam1 mutant of N. sylvestris (woodland tobacco). We obtained six transgenic lines that showed various degrees of complementation in leaf-blade expansion (Fig. 3A, Supplementary Fig. S2). The lam1 mutant has a very short stem and is non-flowering, whereas two of the 35S::SlLAM1/lam1 transgenic plants had a wild-type-like stem and were able to flower under standard growth conditions (Fig. 3B).The other four transgenic plants did not show rescue of the non-flowering phenotype, and this might have been due to the higher accumulation of SlLAM1 transcripts in these lines (Fig. 3C, Supplementary Fig. S2). However, the 35S::SlLAM1/lam1 plants were still sterile, probably due to only partial complementation ( Supplementary  Fig. S3). These results suggested that the protein features of SlLAM1 responsible for leaf expansion are conserved in tomato and tobacco; however, the ectopic expression assays did not enable us to determine whether differential effects existed in reproductive organ growth and flowering. Transgenic plants with SlLAM1 driven by the native or LAM1 promoter would be useful for obtaining information in this regard.

SlLAM1 regulates compound leaf development in tomato
To better understand the function of WOX1 in tomato, we generated loss-of-function mutants of SlLAM1 using CRISPR/ Cas9-mediated genome editing (Ma et al., 2015). Two targets were selected and integrated to one construct, with the aim of enhancing the editing efficiency (Fig. 4A). The two targets were located in the first and the third exon of the CDS sequence of SlLAM1. We acquired two representative mutant lines and confirmed their mutation sites by PCR amplification and sequencing (Fig. 4B). The CR-Sllam1-1 (T 0 ) line was a bi-allelic mutant with a 9-bp deletion (in-frame) in target 1, and 11-bp and 7-bp deletions in target 2 (Fig. 4B). The CR-Sllam1-3 (T 0 ) line was homozygous with a 86-bp deletion near target 1 and a 7-bp deletion in target 2 (Fig. 4B). Compared with the amino acid sequences of STF, LAM1, and Expression is relative to that of ACTIN, which was used as the internal control. Data are means (±SD), n=3. Significant differences compared with the WT were determined using unpaired two-sample t-tests: *P<0.05, **P<0.01, ***P<0.001; nd, not detected. wild-type SlLAM1, the C-terminal domain and part of the middle domain of SlLAM1 were deleted in the CR-Sllam1-1 mutant, while a large fragment frame shift occurred in the CR-Sllam1-3 mutant (Fig. 4C, Supplementary Fig. S4).
The two independent CRISPR-generated mutant alleles displayed similar phenotypes, suggesting that this was indeed caused by the loss-of-function of SlLAM1 (Fig. 4D).
In the CR-Sllam1-1 (T 0 ) line, the phenotype was mild and the mediolateral axis growth of the leaf was not distinctly altered, but the leaf margin was slightly serrated and leaflets were shorter than that of the wild-type (WT). In contrast, the CR-Sllam1-1 (T 1 ) and CR-Sllam1-3 (T 0 ) lines displayed , and CR-sllam1-3 (T 0 ). Scale bars are 2 cm. The CR-Sllam1-1 (T 1 ) mutants all displayed the same defective phenotypes and no WT-like plants were found in these lines. Compared with the WT, CR-Sllam1-1 and CR-Sllam1-3 have narrower lamina and under-developed secondary leaflets. (E) Boxplot of primary and secondary leaflet numbers for the WT and CR-Sllam1-1 (T 1 ) mutants. Measurements were taken on plants at 2 months old, and three leaves (5th-7th) from 10 individual plants were measured. Significant differences were determined using unpaired two sample t-tests: ***P<0.001; ns, not significant. (F) SEM images of shoot apical morphology in the WT and CR-Sllam1-1 (T 1 ) mutant. Scale bars are 100 μm. No lateral leaflet primordia were initiated at P3 or later stages in CR-Sllam1-1. strong defects in leaf growth in the mediolateral axis (Fig. 4D). These CR mutants had narrower terminal and lateral leaflets, and in some cases the blades developed into vestigial strip-like structures, similar to that of the tobacco lam1 mutant (McHale, 1992). This narrow-leaflet phenotype of CR-Sllam1 indicated a conserved function of SlLAM1 in governing blade expansion, similar to previous reports for its orthologous genes (McHale, 1992;Vandenbussche et al., 2009;Tadege et al., 2011;Zhuang et al., 2012). In addition, CR-Sllam1 exhibited alterations in leaflet number and length (Fig.4D). The number was variable in a single mutant; compared with the WT, many of the early leaves displayed fewer or hardly any primary leaflets, while some later leaves had more leaflets but secondary leaflets had almost vanished. We quantified the number of leaflets in compound leaves of the CR-Sllam1-1 (T 1 ) mutant and found that there were no obvious differences for the primary leaves, but the number of secondary leaflets was significantly reduced (Fig. 4E). We also used SEM to examine shoot apical tissue from 1-month-old plants and found that the primary lateral leaflet primordia were initiated at the P3 stage in the WT, but no primordia were present in the CR-Sllam1-1 (T 1 ) mutant at either the P3 or P4 stages (Fig. 4F). These results suggested that the loss-of-function of SlLAM1 might affect the initiation of secondary leaflets in tomato, in addition to its conserved function in blade expansion.
Since the Sllam1 mutants sometimes showed a strip-like leaf, we used semi-thin sections to examine the morphological alterations in the severely defective leaves of CR-Sllam1-3 (T 0 ) to see how the growth of the adaxial/abaxial surfaces were affected. In the WT, the lamina extended from the midrib to the blade margin with a distinct arrangement of tissue layers (Fig. 5A). The arrangement of the layers was similar in the CR-Sllam1-3 mutant but the leaf blade was not extended (Fig. 5B). Examination using SEM showed that although there were more trichomes on the abaxial leaf surface of the CR-Sllam1-3 mutant, the epidermal cells of the WT and the mutant were similar on both the adaxial and abaxial surfaces ( Fig. 5C-I, Supplementary Fig. S5), suggesting that the adaxial/ abaxial identity was not significantly affected. These results therefore indicated that SlLAM1 regulates tomato leaf development mainly by governing the mediolateral axis growth of the blade.

SlLAM1 regulates reproductive organ development in tomato
The CR-Sllam1 mutants also showed significant variation in their floral organs. The WT had fused carpels and six petals fused at the base of the corolla (Fig. 6A), whilst in the CR-Sllam1-1 mutant the petals were narrower and separated from each other, and the style was slightly twisted (Fig. 6B). In the CR-Sllam1-3 (T 0 ) mutant, the petals appeared choripetalous and acicular, the sepals were narrower than those of the WT and CR-Sllam1-1 mutant, and the carpels were dehiscent with multiple slender styles pointing down (Fig. 6B). Consistent with the strength of the mutant phenotypes, we observed that CR-Sllam1-1 (T 0 ) was fertile, while the single CR-Sllam1-3 (T 0 ) plant and the CR-Sllam1-1 (T 1 ) plants were sterile and no seeds could be obtained from them. We tested the viability of mature pollen using Alexander staining and found that it was not significantly affected (Fig. 6B), indicating that the defects in the gynoecium were the cause of the sterility phenotype. Interestingly, we found that CR-Sllam1-1 (T 0 ) plants produced smaller fruit compared to WT plants (Fig. 6C) but that there was no difference in seed size (Fig. 6C), indicating a critical role for SlLAM1 in fruit development in which the bi-allelic mutant leads to reduced fruit size and the homozygous mutant leads to complete sterility.

Differential gene expression between CR-Sllam1-1 and the WT
To gain insights into the potential regulatory pathway of SlLAM1, we conducted RNA-seq using tissue from the shoot apex of 3-week-old seedlings of the CR-Sllam1-1 mutant (T 1 ) and the WT (Supplementary Table S3). The RNA-seq showed high reproducibility (Supplementary Fig. S6) and we identified 1358 differently expressed genes (DEGs; adjusted P-value <0.05) in the mutant, of which 1179 were up-regulated and 179 were down-regulated (Fig. 7A, Supplementary Tables S4,  S5). Genes with negative regulatory functions were significantly enriched ( Supplementary Fig. S7), which is consistent with the primarily repressive function of WOX1 homologues (Lin et al., 2013;Zhang et al., 2014). Analysis of KEGG pathways showed that genes belonging to 'plant hormone signal transduction' (ko04075) were enriched ( Supplementary Fig. S8), and such genes have been widely reported to be involved in leaf development (Bar and Ori, 2014). Interestingly, the auxin biosynthetic and metabolic genes YUCCA3 and IAMT1, the auxin transport component gene PIN, and the auxin-responsive gene SAUR67 were all highly down-regulated (Fig. 7B, C, Supplementary Fig. S9, Supplementary Table S6), suggesting that SlLAM1 may regulate leaflet development via this hormone. Expression of several transcription factors related to leaf polarity and leaf growth were also down-regulated (Fig. 7B, C), indicating that SlLAM1 might recruit multiple genes in the process of regulating leaf development in tomato.

Discussion
WOX1 genes are reported to be involved in lateral organ growth and development in several species. Loss-of-function mutants usually display abnormal leaf and flower phenotypes, for example lam1 in N. sylvestris, maw in petunia, stf in M. truncatula, lath in pea, and mf in cucumber (Vandenbussche et al., 2009;Tadege et al., 2011;Nakata et al., 2012;Zhuang et al., 2012;Costanzo et al., 2014;Niuet al., 2018;Wang et al., 2020). Here, we characterized the WOX1 homologous gene SlLAM1 in tomato, and identified both novel and conserved aspects of WOX1 regulatory roles in plant growth and development. Our results demonstrated that SlLAM1 affected the leaflet number of compound leaves in addition to its commonly described functions in controlling lateral expansion of the leaf blade and floral organs (Figs 3, 4). It appeared that the number of primary leaflets was reduced in the early leaves of the CRISPR/Cas-9-edited line CR-Sllam1-1 but not in leaves that were produced later, suggesting that leaflet initiation may have been delayed. This has recently been confirmed by an independent study of SlLAM1 by Du et al. (2020), who carried out detailed quantification of leaflet numbers and examined the expression patterns of the DR5::VENUS auxin-response reporter and found that leaflet initiation is indeed delayed in Sllam1-1. Since fewer secondary leaflets were found in both CR-Sllam1 (Fig. 4) and the Sllam1-1 mutants (Du et al., 2020), it is reasonable to assume that SlLAM1 affects their initiation. We also found that knocking down SlLAM1 led to narrow leaflets in the CR-Sllam1-1 line in which the C-terminal and part of the middle domain were deleted, whilst the CR-Sllam1-3 line in which almost all the domains were deleted showed the most severe leaf defects. This difference might further indicate that functioning of the homeodomain and C-terminal domain are essential for leaflet outgrowth, which is consistent with previous results (Zhang et al., 2014).
Interestingly, our RNA-seq analysis highlighted the expression of auxin biosynthesis, auxin transport, and auxin-response  Expression is relative to that of ACTIN, which was used as the internal control. Individual data points are shown together with the means (±SD), n=3. Significant differences compared with the WT were determined using unpaired two-sample t-tests: **P<0.01, ***P<0.001. The arrowheads indicate the multiple styles in CR-sllam1-3 (T 0 ), and red shading indicates the exposed ovules that result from the unfused carpels. The viability of pollen grains is not significantly changed in the mutants, as indicated by Alexander staining (lower panels). (C) The fruits but not the seeds of the CR-sllam1-1 (T 0 ) mutant are smaller than those of the WT. Ripe fruits and seeds are shown. Scale bars are 2 cm in (A, C), 100 μm in (B, upper), and 10 μm in (B, lower).
genes as all being highly enriched in the CR-Sllam1-1 mutant compared with the WT (Fig. 7), and this phytohormone is known to be involved in a wide range of plant developmental processes. Similar enrichment in auxin-related gene expression has also consistently been identified in the mf mutant in cucumber (Niu et al., 2018), the stf mutant in M. truncatula (Tadege et al., 2011), and in transgenic plants with induced WOX1 expression in Arabidopsis (Nakata et al., 2018), suggesting that SlLAM1 may indeed interact with several auxin response and signaling components during leaflet initiation, leaf-blade expansion, and floral organ development. In tomato, LYRATE (LYR) is indispensable for regulating the auxin response during the initiation of leaflet primordia (David-Schwartz et al., 2009), and the auxin-response inhibitor ENTIRE (E, SIIAA9) and the CUC transcription factor GOBLET (GOB) can integrate with auxin to regulate leaflet initiation and serration (Ben-Gera et al., 2012). Specific interactions between SlLAM1 and LYR, CUC, or GOB are yet to be established. Notably, a recent report has shown that SlLAM1 functions downstream of E in mediating leaflet initiation and blade expansion (Du et al., 2020). The e mutation, which increases the auxin response, slightly rescues the blade width of Sllam1-1, indicating that there might be additional genes acting downstream of auxin signaling to promote leaf-blade expansion (Du et al., 2020). This inference is partially supported by our results that showed that several auxin-responsive SAUR genes and auxin-related transciption factors were down-regulated in CR-Sllam1-1, including members of the TEOSINTEBRANCHED1/ CYCLOIDEA/PROLIFERATION CELL FACTOR (TCP) and GROWTH-REGULATING FACTOR (GRF) famlies (Fig. 7). Further research is required to determine whether these auxin-related genes are indispensable for SlLAM1mediated leaf developemnt.
Medicago STF has been shown to promote cytokinin activity in transgenic switchgrass by repressing cytokinin-degrading ezymes , and a direct connection between WUS and cytokinin has been demonstrated in Arabidopsis (Leibfried et al., 2005;Meng et al., 2017;Wang et al., 2017;Zhang et al., 2017;Zubo et al., 2017;Ito et al., 2018). In addition, expression of WOX1 and WOX3 in apple is strongly up-regulated in response to cytokinin treatment (Li et al., 2019b), and auxin can also induce the expression of CsWOX1b and CsWOX3 in cumcuber (Gu et al., 2020). In Arabidopsis, expression of WOX1 and PRS can be directly up-regulated by the auxin signaling component Auxin Response Factor 5 (ARF5), but it is suppressed by ARF2, ARF3, and ARF4 (Guan et al., 2017). These studies together with the genetic interactions between auxin-related genes and SlLAM1 determined by Du et al. (2020) imply that both auxin and cytokinin are interconnected with WOX1, and that they might be in some form of regulatory feedback loop. Hence, future studies need to focus on auxin-and cytokinin-mediated pathways and their cross-talk in order to shed light on the multiple functions of SlLAM1 and to comprehensively understand the genetic networks involved in leaf and flower developmental.
In the petunia maw mutant, the petals and carpels are unfused and female fertility is reduced (Vandenbussche et al., 2009;Segatto et al., 2013). Similarly, in the stf mutant of M. truncatula, the petals are unfused and the carpel is open with protruding ovules, which results in complete female sterility (Tadege et al., 2011). The strong alleles in pea lath mutants also result in female sterility (Zhuang et al., 2012). In the Arabidopsis wox1 prs double-mutant, the carpels are fused, but fertility appears to be unaffected despite narrow and unfused petals (Vandenbussche et al., 2009). In the cucumber mf mutant, on the other hand, both male and female sterility are observed, with rare fertility under field conditions producing shorter seeds and fruits (Niu et al., 2018). The N. sylvestris lam1 mutant is completely blocked in stem elongation, but when bolting is induced with high temperature, characteristic unfused petals and carpels are observed together with sterility phenotypes (McHale, 1992;Tadege et al., 2011). These results indicate that the narrow leaf and petal phenotypes of wox1 loss-of-function mutants are common and they show a conserved role for WOX1 in lateral organ expansion; however, the defects in carpel development are variable and lead to phenotypes that range from reduced fertility to sterility depending on species. Compared with these wox1 mutants in other species, the tomato CR-Sllam1 mutants showed similar defects in the form of narrow leaves, unfused petals, and unfused and sterile carpels (Figs 4, 6), but they also showed additional effects in the development of the compound leaves and in the regulation of fruit size. The significant reduction in secondary leaflets in CR-Sllam1 (Fig. 4D-F) reveals that SlLAM1 plays a novel role in regulating their initiation in tomato that is not apparent in legumes. In addition, the significantly reduced fruit size in the heterozygotes (Fig. 6C) suggests that SlLAM1 is involved not only in the regulation of fertility but also in the regulation of embryo development that determines final fruit and seed size. This is consistent with the observation in cucumber that the rarely fertile mf mutants produce shorter fruits and abnormal seeds (Niu et al., 2018); the female-sterility phenotype might have precluded identification of this defect in other species. In agreement with this, the LATH in pea has been found to physically interact with the regulatory proteins controlling organ size BIGGER ORGANS (BIO) and ELEPHANT EAR-LIKE LEAF1 (ELE1), which targetPsGRF5 (Li et al., 2019a). A re-evaluation of the wox1 mutation in these species with conditional WOX1 activity might help to further uncover its role in the regulation of seed size and embryogenesis.
The potential involvement of WOX1 in the development of male reproductive organs has been proposed in cucumber, in which CsWOX1physically interacts with NOZZLE/ SPOROCYTELESS (NZZ/SPL), a protein required for cell division and differentiation of the anther cell wall, to regulate sporogenesis (Niu et al., 2018). RNA-seq analysis also revealed that homologues of genes known to be important for tapetal and microspore development are down-regulated in the cucumber mf mutant, such as CsSPL, CsDYT1, CsMS1, CsAMS, CsGAMYB, and CsMYB103 (Niuet al., 2018), suggesting that CsWOX1 plays a role in male reproductive organ development by directly or indirectly promoting the expression of genes involved in sporogenesis. Previous studies have demonstrated that Medicago STF physically interacts with MtTPL and acts as a transcriptional repressor in promoting the expansion of the leaf blade and floral organs (Lin et al., 2013;Zhang et al., 2014). A similar repressive activity of WUS has also been shown to mediate its functioning in stem cell maintenance in vegetative and reproductive meristems in Arabidopsis (Dolzblasz et al., 2016), even though WUS had previously been reported to be a bifunctional transcription factor that acts as a repressor in vegetative SAMs and as an activator in reproductive organ development (Ikeda et al., 2009). LOOSE FLOWER (LFL), the M. truncatula WOX3 gene, has also been identified to function as a repressor in regulating floral organ development (Niu et al., 2015). Therefore, it is likely that SlLAM1 also acts as a transcriptional repressor in regulating leaf-blade outgrowth, leaflet number, petal expansion, and carpel development. Identifying the key targets required for the accomplishment of each of these processes will further our understanding of the molecular mechanisms by which SlLAM1 orchestrates these multiple functions.
In summary, our study has uncovered the involvement of SlLAM1 in the regulation of secondary leaflet initiation and possibly fruit size in tomato, in addition to regulating leafblade expansion and floral organ development. We have thus expanded the range of developmental processes regulated by WOX1 genes in plants, paving the way for further improving our understanding of the evolution of complex lateral organs.

Supplementary data
The following supplementary data are available at JXB online. Table S1. List of primers used in this study. Table S2. Details of the 10 WOX proteins identified in tomato. Table S3. RNA-seq statistics for the wild-type and CR-Sllam1 samples. Table S4. Significantly up-regulated genes in CR-Sllam1 relative to the wild-type. Table S5. Significantly down-regulated genes in CR-Sllam1 relative to the wild-type. Table S6. FPKM values of selected genes that were significantly down-regulated in CR-Sllam1-1. Fig. S1. Phylogenetic relationships of WOX proteins among different species. Fig. S2. Complementation of the tobacco lam1phenotype by SlLAM1. Fig. S3. Sterility phenotype of transgenic 35S::SlLAM1/ lam1 plants. Fig. S4. Multiple sequence alignments of STF, LAM1, SlLAM1, and the two CR-Sllam1 mutant proteins. Fig. S5. SEM images of the adaxial and abaxial leaf surfaces of the wild-type and the CR-Sllam1-1 (T 1 ) mutant. Fig. S6. Evaluation of the reproducibility of the RNA-seq experiment. Fig. S7. GO enrichment plot of differentially expressed genes between the CR-Sllam1-1 mutant and the wild-type. Fig. S8. KEGG enrichment plot of differentially expressed genes between the CR-Sllam1-1 mutant and the wild-type. Fig. S9. RT-qPCR results obtained using alternative reference genes.