Shaping leaves through TALE homeodomain transcription factors

Abstract The first TALE homeodomain transcription factor gene to be described in plants was maize knotted1 (kn1). Dominant mutations in kn1 disrupt leaf development, with abnormal knots of tissue forming in the leaf blade. kn1 was found to be expressed in the shoot meristem but not in a peripheral region that gives rise to leaves. Furthermore, KN1 and closely related proteins were excluded from initiating and developing leaves. These findings were a prelude to a large body of work wherein TALE homeodomain proteins have been identified as vital regulators of meristem homeostasis and organ development in plants. KN1 homologues are widely represented across land plant taxa. Thus, studying the regulation and mechanistic action of this gene class has allowed investigations into the evolution of diverse plant morphologies. This review will focus on the function of TALE homeodomain transcription factors in leaf development in eudicots. Here, we discuss how TALE homeodomain proteins contribute to a spectrum of leaf forms, from the simple leaves of Arabidopsis thaliana to the compound leaves of Cardamine hirsuta and species beyond the Brassicaceae.


Introduction Plant TALE homeodomain transcription factors
Homeobox genes encode a 60 amino acid DNA-binding homeodomain with a characteristic protein fold composed of three α-helices connected by two short amino acid loops.Homeodomain proteins activate or repress transcription.The TALE (Three Amino Acid Loop Extension) homeodomain superclass is distinguished by three extra amino acids between helix 1 and helix 2 of the homeodomain (Bertolino et al., 1995).KNOTTED1 (KN1) in maize is the founding member of TALE homeodomain transcription factors in plants (Vollbrecht et al., 1991).TALE proteins in plants are encoded by KNOX (knotted1-like homeobox) and BLH (BEL1-like homeobox) genes, and are distinguished by the homeodomain sequence and the presence of conserved domains N-terminal to the homeodomain (Fig. 1).KNOX genes are further divided into two classes, class I KNOX (KNOXI) and class 2 KNOX (KNOXII) (Kerstetter et al., 1994).KNOXI proteins have a diagnostic MEINOX domain, which is composed of two conserved domains, KNOX1 and KNOX2, separated by a variable region, as well as shorter GSE and ELK motifs adjacent to the homeodomain (Fig. 1A) (Vollbrecht et al., 1991;Bharathan et al., 1997;Bürglin, 1997).In many plants, KNOXI and KNOXII comprise a relatively small number of genes; for instance, Arabidopsis thaliana has four KNOXI and four KNOXII genes, and Zea mays has five KNOXI and eight KNOXII genes (Gao et al., 2015).Genetic analysis indicates a degree of redundancy between members within a class, and phenotypes resulting from ectopic expression indicate that proteins within KNOXI and within KNOXII have somewhat different activities (Chuck et al., 1996;Parnis et al., 1997;Nishimura et al., 2000;Byrne et al., 2002;Furumizu et al., 2015;Rast-Somssich et al., 2015).The BLH class of proteins are uniquely characterized by two conserved domains upstream of the homeodomain, SKY and BELL, together called the MEINOX-interacting domain (MID).In addition, some BLH proteins have short conserved amino acid sequences, called ZIBEL, at both the N-and C-terminal ends of the protein (Fig. 1A) (Bellaoui et al., 2001;Müller et al., 2001;Becker et al., 2002;Mukherjee et al., 2009).The ZIBEL domain shares similarities with the ethylene-responsive element binding factor-associated amphiphilic repression (EAR) transcriptional repressor domain.A third TALE-related class of genes, class M KNOX, includes KNATM in A. thaliana.These genes encode proteins that have a MEINOX domain but lack the homeodomain (Fig. 1A) (Magnani and Hake, 2008).
TALE homeobox genes are found in plants, animals, and fungi.This deep evolutionary history suggests that these genes were present in the last common ancestor of eukaryotes (Bharathan et al., 1997;Bürglin, 1997Bürglin, , 1998)).Heterodimer pairing between TALE homeodomain proteins in the unicellular green algae Chlamydomonas and the unrelated brown algae Ectocarpus mediates a haploid to diploid life cycle transition (Lee et al., 2008;Arun et al., 2019).In fact, KNOX and BLH proteins, which are deeply rooted in the Viridiplantae, are both required for development of the diploid sporophyte in the bryophytes Marchantia polymorpha and Physcomitrella patens.As such, it has been proposed that an ancestral function of TALE homeodomain transcription factors was to promote a haploid to diploid transition in plants (Sakakibara et al., 2008(Sakakibara et al., , 2013;;Mukherjee et al., 2009;Hisanaga et al., 2021).Furthermore, expansion of the TALE class of genes throughout the evolution of plants potentially increased the number of possible protein interactions available to regulate development and thereby promote more complex sporophyte body plans (Bowman et al., 2016).

Serrations in the simple leaves
Angiosperm leaves are found in an enormous variety of shapes, but can fundamentally be considered as either simple, with a single continuous lamina, or compound, where the lamina is partitioned into separate leaflets (Fig. 2).The margin of leaf lamina may be further sculpted by differential growth, leading, for instance, to serrations or lobes.Leaves are typically flat structures with features for light interception and gas exchange to maximize photosynthesis.They emerge from the flanks of the shoot apical meristem and early in initiation define adaxial-abaxial (top-bottom), proximal-distal (base-tip), and medial-lateral (midvein to margin) axes that lead to growth of a flat organ.Cell growth and expansion build on this basic plan and modify shape through growth of secondary structures from meristematic regions, termed marginal blastozones, at the leaf margin (Hagemann and Gleissberg, 1996).
What we know about KNOX genes and how they affect leaf shape can be approached through an understanding of genetic regulatory mechanisms shaping simple leaves of A. thaliana.There are four well-studied KNOXI genes in A. thaliana, SHOOT MERISTEMLESS (STM), BREVIPEDICELLUS/ KNAT1 (BP), KNAT2, and KNAT6.All four genes are expressed in regions of the shoot meristem and, to varying degrees, contribute to maintaining the shoot meristem (reviewed in Maksimova et al., 2021).STM is the most crucial of these genes, and loss-of-function mutations result in loss of the shoot meristem.BP and KNAT2 are not essential for meristem function but act redundantly with STM (Byrne et al., 2002;Belles-Boix et al., 2006).All four KNOXI genes are down-regulated in cells of the initiating leaf primordium and, importantly, are not expressed in the developing leaf (Fig. 1B).This same KNOXI expression pattern, being present in the shoot meristem and down-regulated in initiating and growing leaves, is a common property of many angiosperm species that have simple leaves (Smith et al., 1992;Lincoln et al., 1994;Long et al., 1996;Bharathan et al., 2002;Groot et al., 2005).KNOXI down-regulation in the initiating leaf primordium is at least in part mediated by the hormone auxin (Heisler et al., 2005;Hay et al., 2006;Heisler and Byrne, 2020).KNOXI are also repressed in developing A. thaliana leaves by multiple transcription and chromatin-remodelling factors (Satterlee and Scanlon, 2019;Jia et al., 2023).Key transcription factors that repress KNOXI in the leaf are the MYB domain protein ASYMMETRIC LEAVES1 (AS1) and the LOB domain protein ASYMMETRIC LEAVES2 (AS2).Together, these proteins form a heterodimer that represses KNOXI genes BP, KNAT2, and KNAT6 (Byrne et al., 2000;Ori et al., 2000;Semiarti et al., 2001;Iwakawa et al., 2002;Xu et al., 2003;Guo et al., 2008).AS1 and AS2 also specify adaxial fate, with a subtle influence on A. thaliana leaves but a more extreme contribution to adaxial fate in other species.
An interesting series of observations made many years ago in a number of plant species was that ectopic expression of KNOXI, via transgenes expressing KNOXI from ubiquitous, inducible, or leaf-specific promoters, resulted in changes of simple leaf shape to leaves with lobing of the margin (Fig. 3) (Sinha et al., 1993;Lincoln et al., 1994;Chuck et al., 1996;Serikawa and Zambryski, 1997;Tamaoki et al., 1997;Sakamoto et al., 1999;Frugis et al., 2001;Hay and Tsiantis, 2006).In A. thaliana, and in the simple leaves of tobacco and lettuce, the degree of lobing that results from the induction of ectopic KNOXI depends on dosage.Likewise, different KNOXI members show different degrees of leaf lobing (Sinha et al., 1993;Chuck et al., 1996;Tamaoki et al., 1997;Hay et al., 2003;Shani et al., 2009).Elaboration of the leaf shape is proposed to be due to KNOXI activity delaying differentiation, thereby prolonging the window of growth permissive for patterning at the leaf margin (Shani et al., 2009;Kierzkowski et al., 2019).
Arabidopsis thaliana leaves are spatulate with serrated margins.The serrations are small regularly distributed protuberances interspaced by sinuses.Despite being a rather obscure developmental feature, the study of genetic regulation of serrations has yielded interesting findings.Serrations become evident on the margin of young leaf primordia after the basic adaxial-abaxial, proximal-distal, and medial-lateral axes of growth and differentiation have been established (Nikovics et al., 2006;Kawamura et al., 2010;Bilsborough et al., 2011;Jeon and Byrne, 2021).Essential genetic components of serration development are NAC-domain transcription factors, and in A. thaliana this includes CUP-SHAPED COYTYLEDON2 (CUC2) and CUC3.CUC2 and CUC3 are expressed in leaves and, along with the closely related gene CUC1, are also expressed in the shoot meristem.In the shoot meristem, these genes are restricted to a boundary region between organ primordia, where they repress growth and allow organ separation.They are required for meristem function and act in a positive feedback loop whereby CUC activates KNOXI gene expression and in turn STM activates CUC expression (Aida et al., 1997(Aida et al., , 1999;;Vroemen et al., 2003;Spinelli et al., 2011;Scofield et al., 2018).Within the leaf, CUC2 is expressed in the sinus region between serrations and is required to form serrations (Fig. 4A) (Nikovics et al., 2006;Bilsborough et al., 2011).CUC3 is partially redundant with CUC2 and is involved in outgrowth of serrations, but is spatially more restricted and acts later in leaf development compared with CUC2 (Hasson et al., 2011;Serra and Perrot-Rechenmann, 2020).The level of CUC2 is an important determinant of the degree of marginal outgrowths.One mechanism of regulation is via the miRNA, miR164, which represses the level and domain of CUC2 expression in leaf margins, and CUC1 and CUC2 in the shoot meristem (Laufs et al., 2004;Mallory et al., 2004;Nikovics et al., 2006;Sieber et al., 2007;Scofield et al., 2018).
CUC2 works alongside the auxin efflux carrier PIN FORMED1 (PIN1) and the hormone auxin (Koenig et al., 2009;Kawamura et al., 2010;Bilsborough et al., 2011).At early stages of leaf development, polar localization of PIN1 in leaf margin cells directs auxin to the distal tip of the leaf.As the leaf grows, the intracellular polar localization of PIN1 directs auxin towards recurrent convergence points along the leaf margin (Fig. 4A, C).Accumulation of auxin at these convergence points leads to tissue outgrowth (Hay et al., 2006;Scarpella et al., 2006;Kawamura et al., 2010;Bilsborough et al., 2011).CUC2 promotes PIN1 intracellular re-localization and is required to generate the auxin convergence points.Furthermore, auxin represses CUC2, thereby restricting expression to regions between outgrowths (Bilsborough et al., 2011).Altering the CUC2-PIN1-auxin genetic module results in notable leaf phenotypes.Loss of PIN1 or CUC2 leads to loss of serrations and smooth leaf margins, whereas increasing CUC2 or disrupting auxin signalling leads to highly lobed leaves that show ectopic expression of the KNOXI gene BP (Nikovics et al., 2006;Berger et al., 2009;Bilsborough et al., 2011).Thus KNOXI are not expressed in the A. thaliana leaf but this species has the genetic circuity for shaping a lobed leaf margin if KNOXI is provided (Hay et al., 2006;Nikovics et al., 2006;Hasson et al., 2011).
KNOXII genes, on the other hand, are expressed in leaves (Fig. 1C).Interestingly these genes act in the opposite manner to KNOXI, promoting leaf maturation and a simple leaf shape.There is considerable redundancy between KNOXII genes.Mutations in individual genes have no or minimal phenotypic effects on leaf development.However, progressive loss of function of the KNOXII genes KNAT3, KNAT4, and KNAT5  The function of TALE homeodomain proteins in leaf shape | 3225 results in highly lobed leaves in a dosage-dependent manner (Furumizu et al., 2015;Challa et al., 2021).Furthermore, the BLH paralogues SAW1 and SAW2 are expressed in the leaf and redundantly suppress serrations (Fig. 1D).Mutations in either SAW1 or SAW2 have little or no phenotypic effect, but double mutants have more prominent serrations compared with the wild type.Mutations in either SAW1 or KNAT3 have only subtle enhancement of leaf serrations, but more prominent serrations occur in plants lacking both SAW1 and KNAT3, consistent with functional heterodimer formation between these proteins (Kumar et al., 2007;Furumizu et al., 2015;H. Yu et al., 2020;Jeon and Byrne, 2021).Similarly KNOXI-BLH interactions appear to determine leaf shape differences in cultivated varieties of lettuce.In lettuce, the KNOXI gene LsKN1 and the BLH gene LsSAW1 are associated with leaf shape differences, including a serrated or wavy leaf phenotype in heading varieties compared with non-heading varieties.A CACTA transposon insertion mediates higher levels of LsKN1 transcription in heading varieties.This mutant allele probably arose after the first domestication of lettuce (C.Yu et al., 2020;Jia et al., 2022).Heading varieties also have a loss-of-function mutation in LsSAW1.LsKN1 and LsSAW1 can form heterodimers which may inhibit activity as individually they act antagonistically in regulation of downstream targets (An et al., 2022).
KNOXII appear to limit margin growth and promote maturation together with members of the CINCINNATA-TEOSINTE BRANCHED1/CYCLOIDEA/PCF (CIN-TCP) class of genes.Down-regulation of the three KNOXII genes KNAT3, KNAT4, and KNAT5 or the five CIN-TCP genes, TCP2, TCP3, TCP4, TCP10, and TCP24, that are targeted by miR319 results in deeper marginal serrations (Palatnik et al., 2003;Alvarez et al., 2016;Challa et al., 2021).However, reduced levels of both these KNOXII and CIN-TCP genes result in a highly ramified leaf lamina with distinct leaflet-like outgrowths, and outgrowths upon outgrowths.This phenotype requires the CUC2-auxin module and is associated with ectopic expression of KNOXI genes KNAT2 and KNAT6 (Fig. 4D) (H.Yu et al., 2020;Challa et al., 2021).TCP4 directly activates the KNOXII genes KNAT3 and KNAT4, while TCP5, which is not a target of miR319, also activates expression of KNAT3 as well as SAW1 (H.Yu et al., 2020;Challa et al., 2021).An emerging theme is one where the maturation status within the developing leaf is controlled by members of several different transcription factor classes, and considerable genetic redundancy serves to control marginal blastozone activity.

Brassicaceae species with lobed or compound leaves
In plants with simple leaves, KNOXI are typically not expressed in the leaf.In contrast KNOXI genes are expressed in the leaves of many angiosperms with compound leaves, although there are exceptions within the Fabaceae, as discussed below.As we shall see, the expression of KNOXI in compound leaves is a key driver of their complexity.
Arabidopsis species other than A. thaliana tend to have lobed leaves, which is predicted to be the ancestral shape within this plant genus, and multiple independent events have led to a derived simple leaf trait in several lineages (Piazza et al., 2010).The KNOXI gene STM is expressed in the leaves of A. lyrata, A. suecica, A. helleri, and Olimarabidopsis pumila.In A. suecica, RNAi-mediated silencing of STM reduces leaf lobing.In addition, interspecies crosses show that lobing is a dominant trait associated with higher transcription of the STM allele from A. lyrata and A. helleri relative to A. thaliana.STM has probably been a target for evolution of leaf shape.Potentially the simple leaf of A. thaliana evolved from an ancestor with lobed leaves through positive selection for changes in the promoter of STM that led to loss of expression in the leaf (Piazza et al., 2010).
Cardamine hirsuta is close relative of Arabidopsis within the Brassicaceae (Hay and Tsiantis, 2006;Hay et al., 2014;Nikolov et al., 2019;Hendriks et al., 2023).Cardamine hirsuta has pinnately compound leaves, with each leaf comprising a rachis with lateral leaflets and a terminal leaflet.The leaflet lamina is borne on a short stalk or petiolule attached to the rachis (Fig. 2).Cardamine hirsuta has provided key insights into the genetic and potential evolutionary changes required to transition between simple and compound leaf shapes.As in A. thaliana, KNOXI are down-regulated in the initiating leaf primordium of C. hirsuta, but in this species KNOXI expression is reactivated in developing leaves.This reactivation is required to produce leaflets.ChSTM, the orthologue of STM, has a central function in production of leaflets since reducing ChSTM results in loss of leaflets and simplifies the leaf (Fig. 3).On the other hand, loss of ChBP does not reduce leaflets, due to redundancy with ChSTM and possibly also redundancy with ChKNAT2 and ChKNAT6 (Hay and Tsiantis, 2006;Rast-Somssich et al., 2015).Interspecies transgene expression studies, where A. thaliana genes are expressed in C. hirsuta and C. hirsuta genes are expressed in A. thaliana, demonstrate that promoter regions drive gene expression in a pattern reflecting the species of origin.This tells us firstly, that cis-acting elements in promoter regions of STM and BP are responsible for the species differences in expression patterns and, secondly, that both species have all the necessary trans-acting elements for appropriate expression (Hay and Tsiantis, 2006).ChSTM and ChBP are expressed on the abaxial and adaxial sides of the leaf, respectively.This difference may, to some extent, account for their different contributions to leaf shape, although ChSTM and ChBP proteins also differ in their capacity to induce leaf development changes in A. thaliana (Hay and Tsiantis, 2006;Rast-Somssich et al., 2015).
The distinct contributions of ChSTM and ChBP to leaf development are further highlighted through their response to CUC genes.As in A. thaliana, a CUC-auxin genetic module is active in generating outgrowths along the C. hirsuta leaf margin (Hay and Tsiantis, 2006;Barkoulas et al., 2008;Blein et al., 2008).Leaflet initiation is associated with a peak of auxin along the margin of the rachis, and this requires ChPIN1.Disrupting auxin distribution through mutations in ChPIN1 or treatment with auxin produces few leaflets or ectopic leaf lamina in regions between leaflets, respectively.The regulated distribution of auxin along the leaf margin therefore promotes leaflet outgrowth and separation (Barkoulas et al., 2008).ChCUC2 and ChCUC3 expression in C. hirsuta, and CUC3 homologues in other compound leaf species, are restricted to the proximal and distal boundary of leaflets.ChCUC2 is regulated by a miR164 encoded by the PARSLEY (PAR) gene (Rast-Somssich et al., 2015).Overexpressing miR164 or down-regulation of ChCUC2 leads to fewer leaflets and to leaflet fusions.By contrast, par mutants have more leaflets and this is associated with increased ChBP but not ChSTM.Indeed, in the C. hirsuta leaf, ChSTM appears to regulate ChBP (Fig. 4E) (Blein et al., 2008;Berger et al., 2009;Rast-Somssich et al., 2015).At the leaf margin ChSTM is expressed in a pattern complementary to PIN1 and auxin peaks, and ChSTM can be repressed by synthetic auxin.In addition, increasing KNOXI promotes expression of C. hirsuta CUC genes (Barkoulas et al., 2008;Blein et al., 2008).A CUC-auxin genetic feedback module therefore regulates serrations in A. thaliana and leaflets in C. hirsuta.In C. hirsuta this regulatory module interacts with KNOXI to control leaflet development (Rast-Somssich et al., 2015).Interestingly, ChBP is localized to the base of leaflets in a region overlapping with ChCUC2.While KNOXI act as global repressors of differentiation in the leaf, more restricted spatial expression may direct the formation of boundaries between regions of active and repressed growth (Rast-Somssich et al., 2015;Challa et al., 2021).
The CUC-auxin module and KNOXI account for regions of growth and repression at the margin of the rachis in the C. hirsuta compound leaf, but they do not account for the shape of leaflets.This requires the growth repressor REDUCED COMPLEXITY (RCO).RCO is necessary for development of leaflets as loss-of-function mutations result in a lobed leaf (Vlad et al., 2014).KNOXI are expressed relatively broadly and act as global regulators of growth, delaying differentiation and thereby providing an extended morphogenic window for activity of local leaf margin patterning genes.On the other hand, RCO expression is restricted to the base of marginal outgrowths in a region nearly complementary to that of ChCUC2.RCO function is largely spatially and temporally separated from that of ChCUC2 and locally represses growth after ChCUC2-auxin-mediated patterning of the leaf margin (Fig. 4B, E) (Kierzkowski et al., 2019;Bhatia et al., 2023).This implies that there is no direct interaction between RCO and the ChCUC2-auxin module.Instead, RCO acts in the context of patterning set up by ChCUC2 and auxin.
Comparatively, serrations and leaflets are shaped through differential anisotropic growth in regions of high auxin peaks, with KNOXI delaying differentiation and permitting a long duration of outgrowth for leaflets relative to serrations (Kierzkowski et al., 2019).Two additional features distinguish serrations and leaflets.Firstly, serrations are formed at the margin of leaf lamina, whereas leaflets are formed at the margin of the rachis.Thus the outgrowths initiate in different morphological and cellular contexts.Secondly, confined expression of RCO along the base of outgrowths uniquely serves to restrict growth in the proximal zone and along the mediolateral axis of the leaflet to generate a narrow petiolule (Kierzkowski et al., 2019;Wang et al., 2022).Remarkably KNOXI or RCO expression in A. thaliana generates a lobed leaf, but a compound leaf with distinct leaflets can be generated through combined expression of KNOXI and RCO (Chang et al., 2019;Kierzkowski et al., 2019).Localized expression of an RCO transgene to the base of outgrowths in A. thaliana partly requires loss of KNOXI repressors, AS1 or AS2, suggesting that in this heterologous context there is subtle signalling between RCO and KNOXI growth regulators (Wang et al., 2022).However, the precise relationship between KNOXI and RCO in C. hirsuta is still to be determined.
RCO is a homeodomain protein related to LATE MERISTEM IDENTITY 1 (LMI1).The LMI1 genes interestingly have undergone duplications and gene loss within different lineages of the Brassicaceae.RCO is one of three tandem LMI1 genes in C. hirsuta, Capsella grandiflora, Capsella rubella, and A. lyrata.RCO is not present in the A. thaliana genome, which has only one LMI1 gene (Sicard et al., 2014;Vlad et al., 2014).Different LMI1 class proteins from various species induce lobed leaves to some extent in A. thaliana when driven by the ChRCO promoter.As such, a unique property of RCO relative to other LMI1 genes is the spatial expression pattern at the base of the leaflet (Vlad et al., 2014;Vuolo et al., 2016).RCO expression levels also account for leaf shape differences between the Capsella species C. grandiflora, which has a simple leaf shape, and C. rubella, which has lobed leaves.A single major quantitative trait locus (QTL) contributes to this phenotypic difference whereby the C. rubella RCO allele is more highly expressed than the RCO allele of C. grandiflora (Sicard et al., 2014).Likewise leaf shape in different cotton species (Gossypium sp.) varies from entire leaves lacking dissection, to deeply lobed leaves.In cultivars of G. hirsutum, different leaf lobing is largely due to the okra locus, which encodes an LMI1 homologue.Large lobes in okra leaves and severe leaf lobing in super-okra leaves is associated with higher levels of a full-length okra gene, whereas mild and intermediate-lobing cultivars have a mutation in okra leading to a truncated protein.LMI1-associated changes in leaf shape in cotton therefore appear to involve both transcription and protein activity (Zhu et al., 2016;Andres et al., 2017).

Beyond Brassicaceae: other eudicots with compound leaves
The ancestral leaf form in angiosperms is predicted to be a simple shape with compound leaves, and reversion from compound to simple leaf shape occurring multiple independent times.This raises the question as to whether the same or different genetic pathways control leaf shape complexity throughout angiosperms (Bharathan et al., 2002;Groot et al., 2005;Champagne et al., 2007;Sousa-Baena et al., 2014).The study of compound leaf development in several species has provided significant insights into which genetic pathways are conserved, and the pathways that differ between species.Beyond the Brassicaceae, well-studied species with compound leaves are Solanum lycopersicum (tomato), Pisum sativum (pea), Medicago truncatula (barrel clover), and Vigna radiata (mung bean).These species each offer unique opportunities to understand the genetic regulation and evolution of leaf shape diversity.
Wild-type or cultivated tomato has a bipinnately compound leaf composed of a terminal leaflet and three to four pairs of primary leaflets that themselves have secondary leaflets.Additionally, the leaflet margins are lobed, adding to shape complexity (Fig. 2).In these respects, tomato allows examination of whether the context and extent of margin patterning of the leaf lamina are regulated by the same genetic pathways or whether they are developmentally independent.Certainly, KNOXI proteins extend the timing of meristematic activity at the leaf margin, and their level of expression determines the degree of leaflet ramification.Reducing KNOXI levels results in a simplified leaf, whereas mutations or transgenes that increase KNOXI produce a leaf with a higher order of leaflets (Sinha et al., 1993;Hareven et al., 1996;Chen et al., 1997;Janssen et al., 1998;Shani et al., 2009).
Several interesting examples of changes to TALE homeobox gene expression and leaf shape diversity in wild and heirloom cultivars of tomato have been described.The Russian heirloom variety, Silvery Fir Tree (SiFT) generates more leaflets compared with standard cultivars.This phenotype is due to a nucleotide polymorphism that disrupts BIPINNATA (BIP), an orthologue of the A. thaliana BLH class genes SAW1 and SAW2.BIP and SAW repress KNOXI in the leaf (Kumar et al., 2007;Kimura et al., 2008;Nakayama et al., 2021).The second example relates to distinct wild populations of tomato on the Galapagos Islands that have different leaf complexity.Solanum cheesmaniae has a single order of leaflets whereas S. galapagense has three orders of leaflets.The mutation responsible for the increased leaf dissection in S. galapagense is a single nucleotide change in the promoter region of the gene PETROSELINUM (PTS) [also known as TOMATO KNOX-LIKE HOMEODOMAIN PROTEIN 1 (TDK1)].PTS is a class M KNOX gene, encoding a MEINOX domain but not a homeodomain.PTS is up-regulated in S. galapagense compared with S. cheesmaniae, and this higher level of expression influences KNOXI activity.The KNOXI gene TOMATO KNOTTED-1 (TKN1) is transcriptionally up-regulated in S. galapagense and in plants carrying the S. galapagense pts allele.In addition, PTS disrupts the in vitro protein-protein interaction between BIP and the KNOXI class protein LeT6.PTS therefore appears to regulate KNOXI activity in the leaf both transcriptionally and post-transcriptionally (Kimura et al., 2008;Nakayama et al., 2021).In addition to these two examples, gene regulatory network analysis indicates that levels of the transcription factor BLADE-ON-PETIOLE (BOP) correlate with variation between three species of tomato, the cultivated S. lycopersicum, which has intermediate leaf complexity, and two wild relatives, S. pennellii with low leaf complexity and S. habrochaites with high leaf complexity.Solanum pennellii has higher levels of BOP relative to S. lycopersicum, which has higher levels than S. habrochaites.BOP regulates transcription of PTS and BLH class genes (Ichihashi et al., 2014).Consistent with these examples where leaf shape is dependent on the regulation of KNOXI, an unbiased screen for leaflet formation in A. thaliana carrying the C. hirsuta RCO gene only identified mutations in AS1 and AS2, two repressors of KNOXI (Wang et al., 2022).The evolution of complex leaves in tomato and Arabidopsis species may therefore involve modifying KNOXI expression, through change in the promoter region of KNOXI genes that alter their expression, or change in components of the network regulating KNOXI expression or activity (Hay and Tsiantis, 2006;Ichihashi et al., 2014;Wang et al., 2022).
Leaflet production and separation as well as lobing of leaflets in tomato involves GOBLET (GOB), a transcription factor closely related to CUC2 (Berger et al., 2009).Loss-of-function mutations in GOB lead to simplified leaves with only primary, unlobed leaflets.On the other hand, GOB dominant mutants, which have increased expression due to a mutation in a miR164binding site, have deeply lobed leaflets and fewer leaflets (Berger et al., 2009;Israeli et al., 2021).GOB therefore appears to have context-dependent effects.At the margin of the leaflet lamina, GOB regulates the extent of lobing, whereas along the rachis GOB regulates production of leaflets.In addition to GOB, multiple auxin response regulators are required for leaflet production.These include the indole acetic acid (Aux/IAA) protein ENTIRE (E) and members of the AUXIN RESPONSE FACTOR (ARF) class such as SlMP/SlARF5.Aux/IAA proteins repress the activity of ARF proteins, but this repression is released in the presence of high auxin where Aux/IAA is degraded.E is expressed in the intercalary region between leaflets, and ARFs are expressed within the outgrowing leaflet.Both GOB and SlMP are required for the increased leaflet ramification observed in plants with elevated expression of KNOXI (Berger et al., 2009;Koenig et al., 2009;Ben-Gera et al., 2012;Israeli et al., 2019Israeli et al., , 2021;;Xiong and Jiao, 2019).The tomato compound leaf shape conferred by KNOXI is also dependent on CIN-TCP class transcription factors such as Lanceolate (LA).
Although LA dominant mutants have a simple leaf and suppress the highly ramified leaf phenotype induced by overexpression of KNOXI, reducing levels of LA promotes leaflet production in a manner that is additive with overexpression of KNOXI.LA may act independently of KNOXI but provide a permissive developmental environment for KNOXI activity (Ori et al., 2007).
The value of studying leaf development across a broad spectrum of the plant phylogenetic tree is aptly demonstrated by key findings on leaf development in compound leaves of species in the Fabaceae, commonly known as the legumes.Typically, these species have pinnately compound leaves with lateral leaflets and a terminal leaflet, each connected to a central rachis, although in some species leaves are trifoliolate or palmate.Further variations such as in P. sativum, the common garden pea, include prominent leaf-like stipules and reduction of leaflets into slender tendrils (Fig. 2).
Pea is a member of a large subclade of legumes, the inverted repeat-lacking clade (IRLC), characterized by a loss of one copy of a chloroplast genome inverted repeat sequence.Fabaceae species outside the IRLC clade, including Glycine max (soybean), Mimosa pudica, Acacia hindsii, Phaseolus vulgaris (bean), and Lotus japonicus, have compound leaves with KNOXI expression in the leaf, and in this respect are similar to C. hirsuta and tomato (Champagne et al., 2007).Conversely KNOXI genes are typically not detected in pea or M. truncatula leaves or in leaves of other IRLC species M. sativa (alfalfa), Vigna radiata (mung bean), Wisteria sinensis (wisteria), and Vicia faba (fava bean) (Gourlay et al., 2000;Hofer et al., 2001;Champagne et al., 2007;Ge et al., 2014;Zhou et al., 2014).Compound leaf development is instead dependent on transcription factor genes related to UNIFOLIATA (UNI) in pea and SINGLE LEAFLET1 (SGL1) in M. truncatula.Mutations in these genes have fewer lateral leaflets and may develop as a simple leaf (Fig. 3) (Hofer et al., 1997;Wang et al., 2008).UNI and SGL1 are orthologues of the transcription factors FLORICAULA (FLO) in Antirrhinum and LEAFY (LFY) in A. thaliana, which control floral meristem identity (Moyroud et al., 2010).Although UNI and SGL1 appear to be the principal drivers of leaf complexity in these species, in pea, M. truncatula, and M. sativa, mutations in orthologues of AS1 result in changes in leaf development and some ectopic KNOXI expression in leaves.The phenotypic consequences vary somewhat but include ectopic lamina, more prominent leaflet serrations, and more leaflets (Tattersall et al., 2005;Champagne et al., 2007;Ge et al., 2014;Zhou et al., 2014).These are relatively subtle phenotypes, possibly due to modest levels of KNOXI in these mutants since transgeneinduced high levels of KNOXI in leaves of M. truncatula induce variations in leaflet number and higher order leaflets (Zhou et al., 2014).Surprisingly, STM from A. thaliana can rescue the sgl1 mutant leaf phenotype, suggesting that both LFY and KNOXI class genes have common downstream targets (Pautot et al., 2022).In the non-IRLC species tomato, G. max, L. japonicus, and V. radiata, KNOXI are detected in the leaves but LFY orthologues also play a minor role in promoting production of leaflets (Molinero-Rosales et al., 1999;Dong et al., 2005;Champagne et al., 2007;Jiao et al., 2019).As such, LFY-related genes and KNOXI may have parallel functions or have shared downstream targets.From an evolutionary point of view, in the IRLC clade Fabaceae species both pathways may have been functional.The loss of a role for KNOX1 genes in leaf development in these species appears to have occurred via loss of expression in the leaf.This may have occurred through changes in cis-regulatory sequences of KNOX1 genes or could be the result of changes to upstream regulators of KNOX1 genes.
Interestingly the CUC-auxin genetic module appears to have a conserved function in leaflet development in IRLC species, but this module interfaces with UNI rather than KNOXI.In pea, as in C. hirsuta and tomato, CUC3 genes are expressed in the proximal and distal boundary of leaflets, and peaks of auxin occur where leaflets initiate.Disrupting CUC3 and auxin leads to a more simplified leaf (Blein et al., 2008;Demason et al., 2013).Likewise, in M. truncatula, mutations of SMOOTH LEAF MARGIN1 (SLM1), an orthologue of PIN1, result in a loss of leaf serrations and fewer lateral leaflets.However, in sml1 mutants there are additional terminal leaflets, suggesting that different mechanisms or different degrees of gene redundancy regulate M. truncatula lateral and terminal leaflets (Zhou et al., 2011).Mutations in a CUC3-related gene, MtNAM, result in fusion of leaflets but do not affect leaflet initiation or margin serrations possibly due to gene redundancy (Cheng et al., 2012).Although CUC-auxin represents a common genetic module for serrations, lobes, and leaflets in many plant species, the spatial and temporal activity of this pathway appears to depend on genetic redundancy of CUC class genes.

Conclusions
As it is theorized that compound leaves are the ancestral leaf form of many angiosperm lineages, with multiple incidences of reversion to simple leaves, we must question just how conserved leaf developmental pathways are.In this context, TALE homeodomain proteins present an enduring point of interest as their continuity between species and their function in leaf shape complexity allow a useful starting point to probe the evolution of leaf regulatory modules.
Several themes begin to emerge when discussing the importance of TALE homeodomain proteins in leaf development, which to date has largely focused on KNOXI proteins.Firstly, whilst KNOXI are conserved across land plant taxa, an expansion of this gene class has occurred throughout evolutionary history.Such changes result in gene redundancies that are predicted to have lent plasticity to the regulatory modules that underpin leaf development.In turn, this plasticity may have broadened the potential protein-protein interactions that can interface with established genetic regulatory networks.Likewise, changes to KNOXI expression domains through alterations in promoter regions are, in some cases, sufficient to generate morphological variation in leaves.These perturbations to regulatory mechanisms of development account for some of the diverse range of leaf forms observable in the natural world today.From here, we can more broadly assess the types of changes leading to natural variation in leaf shape and whether the same regulatory modules are the target for evolutionary change.Several plant species undergo radical changes in leaf shape in response to environmental influences of temperature, light quality, and in response to aquatic or terrestrial conditions (Chitwood and Sinha, 2016).Whether TALE homeodomain transcription factor-mediated changes in leaf shape have adaptive significance remains to be uncovered.

Fig. 1 .
Fig. 1.TALE homeodomain transcription factors in plants.(A) Schematic diagram of key domains in KNOX, KNATM, and BLH proteins.(B) The KNOXI gene STM is expressed in the shoot apical meristem (SAM) and is not expressed in initiating leaf primordia and leaves.(C) The KNOXII gene KNAT3 (C) and the BLH gene SAW1 (D) are expressed in the leaves and not in the shoot meristem.Scale bars indicate 50 µm.

Fig. 2 .
Fig. 2. Simple and compound leaves.Different shapes of leaves with the basic anatomy labelled.From left to right: Arabidopsis thaliana simple leaf with serrations, Cardamine hirsuta pinnately compound leaf, Solanum lycopersicum bipinnately compound leaf, Pisum sativum pinnately compound leaf with terminal tendrils, and Medicago truncatula trifoliate compound leaf.Scale bars indicate 1 cm.

Fig. 3 .
Fig. 3. KNOXI is associated with changes in leaf shape complexity.Left: Arabidopsis thaliana wild-type simple leaf with undissected lamina.Ectopic expression of STM in the leaf results in leaf lobing.Middle: Cardamine hirsuta leaf reduced ChSTM has fewer leaflets compared with the wild-type compound leaf.Cardamine hirsuta leaf with overexpression of ChSTM has more leaflets compared with the wild type.Right: Pisum sativum wild-type compound leaf and uni mutant with a simple leaf.Leaf sizes are not to scale.Schematic diagrams of loss-of-function mutants and ectopic expression phenotypes are based on data in Hay and Tsiantis (2006), Demason et al. (2013), and Alvarez et al. (2016).

Fig. 4 .
Fig. 4. Growth patterning at the leaf margin.Schematic representations of young developing leaves of A. thaliana (A) and C. hirsuta (B).Growth along the leaf margin at serrations is promoted in regions of high auxin (dark blue) and is repressed in sinus regions where CUC2 (orange) is expressed.In C. hirsuta, growth is also repressed at the base of leaflet outgrowths through the activity of RCO (yellow).In C. hirsuta (B), ChSTM (pale blue) has a broad expression domain and acts to delay differentiation.(C) At serrations in A. thaliana, a feedback loop promotes PIN1-mediated auxin peaks in serrations and CUC2 expression in sinus regions.(D) In wild-type A. thaliana, morphogenetic potential at leaf margins is reduced by CIN-TCP and KNOXII (black), which together repress expression of several KNOXI genes (light grey) and partially repress CUC2 (grey).In the absence of CIN-TCP and KNOXII (light grey), expression of several KNOXI genes is activated and CUC2 is up-regulated, leading to leaflet-like outgrowths.(E) In C. hirsuta, early leaflet outgrowth is promoted by a ChCUC2-auxin module that interacts with KNOXI genes.RCO is expressed at the base of developing leaflets.A regulatory interaction between RCO and KNOXI may or may not occur in C. hirsuta.Based on data in Bilsborough et al. (2011), Rast-Somssich et al. (2015), and Challa et al. (2021).