Targeted manipulation of leaf form via local growth repression


For correspondence (fax +0114 2220002; e-mail


A classical view is that leaf shape is the result of local promotion of growth linked to cell proliferation. However, an alternative hypothesis is that leaf form is the result of local repression of growth in an otherwise growing system. Here we show that leaf form can indeed be manipulated in a directed fashion by local repression of growth. We show that targeting expression of an inhibitor of a cyclin-dependent kinase (KRP1) to the sinus area of developing leaves of Arabidopsis leads to local growth repression and the formation of organs with extreme lobing, including generation of leaflet-like organs. Directing KRP1 expression to other regions of the leaf using an miRNA target sequence tagging approach also leads to predictable novel leaf forms, and repression of growth in the leaf margin blocks the outgrowth of lobes, leading to a smoother perimeter. In addition, we show that decreased growth around the perimeter and across the leaf abaxial surface leads to a change in 3D form, as predicted by mechanical models of leaf growth. Our analysis provides experimental evidence that local repression of growth influences leaf shape, suggesting that it could be part of the mechanism of morphogenesis in plants in the context of an otherwise growing system.


Leaf morphogenesis occurs by a process of differential growth (Fleming, 2006). Although it is commonly assumed that the various lobes and serrations that define much of leaf shape are generated by a process involving localized promotion of growth, it is conceptually equally robust to perceive shape formation as a process of localized growth repression against a background of otherwise maintained growth (Figure 1). This hypothesis is derived from classical studies of comparative leaf morphology, in which it was suggested that the observed spectrum of leaf shape represents different degrees and patterns of local growth promotion/suppression (Kaplan, 2001). Recent evidence from molecular genetic analysis of Arabidopsis suggests that tightly regulated growth repression may be involved in the control of leaf shape. In particular, the spatial/temporal control of CUC2-related transcription factor activity has emerged as a paradigm for the determination of aspects of leaf shape in a number of plants, including serration, lobing and leaflet formation in compound leaves (Nikovics et al., 2006; Peaucelle et al., 2007; Blein et al., 2008). During the early stages of Arabidopsis leaf development, expression of the CUC2 transcription factor becomes restricted, defining the boundaries of presumptive serrations rather than the serrations themselves (Figure S1; Nikovics et al., 2006), suggesting that, although the serrations appear to grow out from the leaf, they actually form by a process involving growth repression at the serration boundaries while maintaining growth within the serrations, leading to the difference in growth distribution required for morphogenesis (Nikovics et al., 2006).

Figure 1.

 Growth promotion and repression as alternative means of morphogenesis.
If an imaginary oval-shaped organ (a) undergoes uniform growth, a larger but similar shaped organ results (b). If growth occurs at a higher rate in localized regions within the organ indicated by circles in (a), morphogenesis occurs such that a series of protrusions are formed at these locations, leading to a serrated perimeter (c). If growth occurs at a lower rate in the same localized regions within the organ, morphogenesis occurs, also leading to a serrated perimeter (d). Depending on the temporal pattern of growth promotion/repression and the overall growth rate of the organ, distinguishing options (c) and (d) by simple visual observation may be difficult.

To test the hypothesis that localized growth repression could play a role in leaf morphogenesis, we engineered an artificial system in which growth could be repressed in an inducible and targeted fashion during Arabidopsis leaf development. If such manipulations led to predictable changes in leaf form, this would support the hypothesis. The first challenge was to identify a tool to suppress growth. Although a number of studies have reported alteration of leaf form via gene manipulation, the majority of these investigations have utilized upstream regulators of development (e.g. transcription factors and hormonal signalling systems), making it difficult to define the actual downstream growth processes involved in generating the observed morphogenic response (Autran et al., 2002; Nath et al., 2003; Palatnik et al., 2003; Dinneny et al., 2004). Testing the hypothesis that growth repression could play a role in leaf morphogenesis requires the use of factors that are much closer to the final downstream effectors of growth. Plant growth is the result of co-ordinated cell division and cell growth. At its core, plant growth involves a balance of turgor-generated pressure (providing the force for growth) and restrictive forces within the cell wall (Tomos and Pritchard, 1994; Fleming, 2006). Ideally, to locally repress growth, one would either locally reduce turgor pressure or locally decrease cell-wall extensibility. However, implementing such changes on the causative agents of growth presents a number of problems. For example, decreasing turgor is likely to cause a number of pleiotropic stress responses that complicate any observations made, and increasing cell-wall stiffness in young growing tissue is likely to require repression of a minimum of several genes encoding known positive effectors of cell-wall extensibility (Schipper et al., 2002; Somerville et al., 2004; Zhong and Ye, 2007; Peaucelle et al., 2008), which is technically challenging. To overcome this challenge, we focused on manipulation of a repressor of cell division, AtKRP1. This encodes an inhibitor of the cyclin-dependent kinases that are required for progression through the plant cell cycle, and it has been shown in a number of studies that over-expression of this gene leads to termination of division, and, as a consequence, growth repression at the tissue level (Wang et al., 2000; De Veylder et al., 2001; Inzé, 2005). Moreover, it has been shown that the protein moves only over short intercellular distances, making it a useful tool to locally suppress cell division and growth (Weinl et al., 2005). We therefore decided to use expression of AtKRP1 as a tool to repress growth in a targeted fashion during leaf development. To regulate the spatial/temporal expression pattern of this growth repressor, we used established inducible promoter and enhancer trap systems (Craft et al., 2005; Laplaze et al., 2005). These systems provide a means of generating novel patterns of AtKRP1 expression at specific time points in leaf development, thus allowing us to observe the effect of new patterns of growth repression on leaf form.

In the experiments described here, we show that local inhibition of growth is indeed sufficient to alter leaf form in a targeted way, leading to a range of shapes, including formation of leaflet-like organs and cup-shaped leaves. These data support the hypothesis that localized growth repression could be a mechanism for shape control in leaves.


Inhibition of growth in the CUC2 domain leads to leaf lobing and formation of leaflet-like organs

Wild-type (Col-0) Arabidopsis leaves develop an ovate form with a limited number of serrations towards the proximal base of the leaf (Figure 2a). At a very early stage of leaf development, the region of future perimeter deformation is defined by CUC2 expression (as visualized by CUC2>>GUS reporter gene activity (Figure S1a; Nikovics et al., 2006). CUC2 expression then gradually becomes restricted to regions on the leaf perimeter adjacent to the emerging lobe (Figure S1b,c; Nikovics et al., 2006). To investigate the effect of enhanced growth repression in the CUC2 domain, an AtKRP1 coding sequence was placed under Dex-inducible transcriptional control via a CUC2 promoter (CUC2>>KRP1 plants), and the effect of Dex induction on leaf form was observed. If growth repression normally occurs but is not maximal in the CUC2 domain, local over-expression of AtKRP1 in this area should lead to enhanced repression of growth, and thus deeper leaf dissection. Our results show that, when AtKRP1 expression was targeted to the CUC2 domain of Arabidopsis leaves, more extreme lobing did indeed occur (Figure 2a). This frequently led to formation of distinct leaflet-like organs along the lower flank of the leaves (Figure 2a,b,d). Although the depth of sinus indentation was clearly increased towards the base of the leaves, a smaller change was observed towards the leaf tip, suggesting that the basic pattern of lobbing/serration was not altered by these manipulations. The pOp synthetic promoter used in these experiments (Craft et al., 2005; Samalova et al., 2005) resulted in simultaneous transcription of AtKRP1 and a GUS reporter gene after Dex induction, allowing verification that target gene expression was localized to the expected region of the leaf flank (Figure S1d–g). Analysis of leaf histology showed that the cells in the region of CUC2-directed AtKRP1 expression were greatly enlarged relative to adjacent cells (Figure 2e–g) and relative to cells in equivalent positions in wild-type leaves (Figure 2c). Cell enlargement is a diagnostic for AtKRP1 over-expression (Verkest et al., 2005), as, when cell division is inhibited, cell size increases to a limit, at which time cell growth is terminated. This differential cell size was observed in both the epidermis and parenchyma of the targeted tissue (Figure 2f,g), indicating that CUC2-directed KRP1 activity was present in all layers of the leaf in the CUC2 domain.

Figure 2.

 Over-expression of KRP1 in the CUC2 domain leads to the formation of leaflet-like structures.
(a) Comparison of CUC2>>KRP1 plants grown for 21 days on 0.5× MS medium without inducer (−DEX) or supplemented with 10 μm dexamethasone (+DEX). Leaves L3–L10 from representative plants are shown.
(b) Forty-eight-day-old fully expanded leaves of Dex-induced CUC2>>KRP1 plants showing lobes and leaflet-like organs.
(c) Close-up of non-induced leaf 6 (inset) of a CUC2>>KRP1 plant showing the cell size on the lower flank of a serration.
(d) Close-up of a fully expanded 48-day-old leaf of a Dex-induced CUC2>>KRP1 plant. Note that an organized vein architecture is observed.
(e–g) Increased cell size in a Dex-induced CUC2>>KRP1 leaf indicates the area of KRP1 expression. (e) shows the region boxed in (d) at the boundary of a lobe. Larger cells are visible at the lobe boundary (arrows), whereas cells within the lobe (asterisk) are smaller. Enlarged cells are visible in both the epidermis (f) and parenchyma (g) layers at the leaf/lobe boundary.
Scale bars = 5 mm (a, b), 100 μm (c–g) and 500 μm [magnified image in (c)].

Although the CUC2 promoter leads to a specific expression pattern within the developing leaf, the promoter is also active within the shoot apical meristem, defining the boundary between the emerging leaf and the meristem (Laufs et al., 2004). To discriminate between the effect of ectopic KRP1 activity during leaf initiation and during subsequent leaf growth, we performed a series of experiments in which the Dex inducer was supplied to CUC2>>KRP1 plants during defined time windows. Under our growth conditions, leaves L8, L9 and L10 were already initiated in control plants by day 16 (Figure 3a). When Dex was supplied to plants from days 16–21 (i.e. post-initiation of leaves L8, L9 and L10), extreme lobing was observed (Figure 3b,c), indicating that the induced CUC2-directed KRP1 expression was active after leaf initiation and was responsible for the observed changes in shape in the proximal part of the leaf where cell division normally occurs at this stage of development (Donnelly et al., 1999).

Figure 3.

 Temporal control of Dex induction of CUC2>>KRP1 plants, indicating that deeper lobes and leaflet-like structures are generated via CUC2 activity after leaf initiation.
(a) Non-induced CUC2>>KRP1 plants at day 16. Leaf primordia L8, L9 and L10 are already initiated. In the L10 primordium, development of the first pair of lobes is just visible, whereas the L8 primordium shows initiation of a second pair of lobes.
(b) Leaves L8, L9 and L10 from CUC2>>KRP1 plants after Dex induction from days 16–21 (+DEX D16–D21). Extreme lobes and leaflet-like structures are present along the proximal flanks by day 21; these are not observed in control leaves (−DEX D21). This indicates that the observed effect of CUC2>>KRP1 induction is not due to altered growth activity at the point of leaf initiation on the apical meristem.
(c) Comparison of all leaves (L1–L10) in 21-day-old CUC2>>KRP1 plants that were not induced (−DEX D21) or induced with Dex from days 16–21 (DEX 16–21). The older leaves (L1–L6) are not altered in shape. These data indicate that the action of CUC2>>KRP1 is restricted to early stages of leaf development (L7–L10) when sinus formation normally occurs.
Scale bars = 200 μm (a) and 5 mm (b, c).

The morphogenic response observed in the CUC2>>KRP1 plants was distinct from that observed in plants in which KRP1 expression was induced throughout the leaf (pOpON::KRP1 plants shown in Figure S2). In these plants, a smaller leaf size was observed, with a similar pattern of dissection but deeper indentations than in non-induced plants, comparable to that previously reported for 35S::KRP2 plants (De Veylder et al., 2001). However, the depth of sinus indentation observed in induced pOpON::KRP1 leaves was distinctly less than that observed in the CUC2>>KRP1 leaves, indicating that a greater gradient of growth repression between the sinus and lobes was achieved in CUC2>>KRP1 plants.

The extreme lobing and leaflet-like organ formation observed in our experiments with CUC2>>KRP1 plants was reminiscent of the changes observed when transcription factors involved in leaf development are mis-expressed (Sinha et al., 1993; Barkoulas et al., 2007). To investigate whether the morphogenic response observed involved altered expression of these transcription factors, we performed a series of RT-PCR analyses using primers for a variety of such genes. The results of these experiments showed that increased KRP1 expression in the CUC2 domain did not influence expression of the major patterning genes implicated in Arabidopsis leaf morphogenesis (Figure S3), indicating that the obtained phenotypes were the result of local growth repression alone, rather than an indirect affect via alteration of the transcriptional network controlling leaf shape.

Growth repression in the distal region of the leaf using miR164ts-tagged KRP1 leads to goose foot-shaped leaves

The results from analysis of CUC2>>KRP1 plants indicated that enhanced growth repression in the CUC2 domain led to deeper lobe formation. To investigate whether induction of growth repression in other domains of the leaf could be used to alter leaf shape, we created transgenic lines in which a modified form of KRP1 [KRP1miR164ts, in which the target sequence for miR164 (Nikovics et al., 2006) was spliced to the 3′ end of the KRP1 coding sequence] could be inducibly expressed throughout the leaf (pOpON::KRP1miR164ts plants). The miRNA164 family has been shown to target cleavage of NAM/CUC transcripts in the shoot apex and developing leaves (Sieber et al., 2007; Raman et al., 2008). These studies also showed that the three Arabidopsis genes encoding miR164 (miR164a,b and c) are expressed in a complex fashion during early leaf development, including a phase during which miR164a is present in a domain that is essentially the same as that defined by the CUC2 promoter expression pattern (Figure S1). By inducibly over-expressing KRP1miR164ts throughout the leaf at this stage of leaf development, we aimed to target degradation of KRP1 to mainly the CUC2 domain while allowing over-expression of KRP1 throughout the rest of the leaf, i.e. to create essentially the opposite pattern of growth repression to that induced in the CUC2>>KRP1 plants described above.

At the time of Dex induction (days 8–10, primordium length 400–500 μm), the endogenous CUC2/miR164 expression pattern within the leaf encompasses the lower flanks of the primordium and a region just above the first emerging serration/lobe (Figure S1b). These regions are protected from the effect of elevated KRP1miR164ts expression and are expected to undergo a normal pattern of growth, whereas the upper parts of the leaf and the first pair of lobes (in which miR164 is absent) are expected to undergo growth repression. At later time points, a new serration/lobe is expected to appear towards the base of the leaf, but by this time (after day 10), the Dex inducer has been removed, so no induction of KRP1miR164ts should occur in this region (i.e. growth at the base of the leaf would continue as normal and would not be repressed). Our observations (described below) fit with these predictions.

When KRP1miR164ts expression was induced during an early phase of growth of leaf 4 (days 8–10), growth was repressed in the distal domain of the leaf (Figure 4a). Small serrations/lobes were produced in the upper part of the leaf, reflecting a pattern already present at the time of induction, but these remained as small peg-like structures (white arrow in Figure 4b). The cells at a distance from this structure were relatively large (both in the epidermis and parenchyma) (Figure 4c), indicating the non-CUC2 domain (at the time of induction) in which over-expressed KRP1miR164ts was not cleaved and thus tissue growth was repressed via KRP1 activity. In the lower part of the leaf, in the axis of the lower lobe and main body of the leaf (green arrow, Figure 4b), cell size was normal (both in epidermis and parenchyma) (Figure 4d), indicating the CUC2 domain (at the time of induction) in which over-expressed KRP1miR164ts was cleaved and a normal pattern of growth occurred. Molecular analyses of these leaves confirmed that AtKRP1 mRNA accumulated after Dex induction (Figure 4e) and that the expected degradation product of the KRP1miR164ts mRNA was present in the induced leaf (Figure 4f). Similar morphogenic responses were observed when similar inductions were performed to target KRP1miR164ts expression during specific growth phases in earlier and later leaves (Figure S4).

Figure 4.

 Growth repression in the distal region of the leaf using miR164ts-tagged KRP1 leads to goose foot-shaped leaves.
(a) Leaf number 4 from a series of CUC2>>KRPmiR164ts plants at days 8, 10 or 21 (non-induced, −DEX) or induced from days 8–10 (+Dex D8–10) before analysis at day 21 (D21). This transient induction leads to formation of goose foot-shaped leaves.
(b–d) Detail of a goose foot-shaped leaf resulting from Dex induction (days 8–10) of a CUC2>>KRPmiR164ts plant. The white arrow in (b) indicates a vestigial serration initiation point. Enlarged cells are present in both the epidermis and parenchyma of areas predicted to have increased KRP1 expression (c), whereas the cell size (epidermal and parenchyma) is smaller in the adjacent lobe region (d), in which endogenous miR164a is predicted to lead to degradation of the KRPmiR164ts transcript.
(e) RT-PCR analysis of CUC2>>KRPmiR164ts plants indicated an elevated level of KRP1 transcript in Dex-treated leaves (+DEX) relative to untreated controls (−DEX).
(f) 3′ RACE analysis of RNA shows that the predicted miR164-triggered degradation product of the KRPmiR164ts transcript is only detectable in Dex-treated CUC2>>KRPmiR164ts plants. The miR164 cutting site was verified by sequence analysis of the amplified fragment.
Scale bars = 5 mm (a, b), 200 μm [inset D8 in (a)], 500 μm [inset D10 in (a)] and 100 μm (c, d).

Over-expression of KRP1 in the margin blocks lobe outgrowth and leads to an altered 3D shape

The results presented above show that the spatial/temporal control of growth suppression is sufficient to define the degree of leaf serration/lobing and to change 2D leaf shape. They did not address the role of growth repression within the lobes themselves with respect to leaf morphogenesis, nor did they provide information on the potential role of growth suppression in the mechanism of establishment of the 3D leaf form. To investigate these issues, we targeted repression of growth to the leaf perimeter using an enhancer trap (E4907) to drive AtKRP1 expression (Laplaze et al., 2005). E4907 shows dynamic margin-localized reporter gene expression during early stages of leaf development, with GFP signal extending around the leaf perimeter from the distal tip, followed by extension of GFP expression across the abaxial epidermis at later stages of development (Figure 5). Thus, the E4907 line can be used to drive expression of a target gene (in our case, AtKRP1) to create gradients of growth repression, initially across the surface of the leaf (from the perimeter inwards), and, at later stages of leaf development, along the adaxial/abaxial axis.

Figure 5.

 The E4907 driver line can be used to target gene expression to the leaf margin and abaxial epidermis.
(a, b) Series of leaves (L9–L15) from E4907 plants showing that GFP expression (a) is restricted to the distal tip of the leaf at early stages. At later stages, GFP expression gradually becomes extended around the leaf perimeter. Bright-field microscope images of the same leaves are shown in (b).
(c, d) Cross-sections of early-stage (c) and later-stage (d) E4907 leaves, indicating lack of GFP expression at the very early stages of primordium development (arrow) and restriction of GFP expression to the leaf margins at later stages (asterisks).
(e) Cross-section of a young E4907 leaf (approximately 1.6 mm wide) showing margin-localized GFP expression and very weak GFP expression in the abaxial epidermis.
(f, g) Close-up of GFP expression in the marginal region of a leaf (f), and the same leaf under bright-field microscopy (g).
(h) At later stages of E4907 leaf development (leaf width 2.6 mm), GFP expression is evident both at the leaf margin and across the abaxial epidermis, with a gap in the region of the mid-vein.
(i, j) Close-up of a leaf margin region showing GFP expression (i) and combined GFP and tissue autofluorescence (j).
ab, abaxial; ad, adaxial; mv, mid-vein. Scale bars = 50 μm (a, b) and 100 μm (c–j).

Imposition of growth suppression around the leaf perimeter should tend to smooth any gradients of growth around the margin and thus decrease serration. When AtKRP1 expression was driven via the E4907 line (E4907>>KRP1), a smoother leaf perimeter resulted relative to the control line (Figure 6a,b). Closer examination revealed that serration initiation had occurred in these leaves (Figure 6c,d), but that the subsequent differential growth that normally leads to the curved form of the serration was lacking, with the serrations themselves remaining as small pegs consisting of relatively enlarged cells. As the E4907 driver line does not target gene expression to the entire margin at the very earliest stages of leaf development (Figure 5a,b), our data do not address the question of the role of cell division at the stage of serration initiation, but they do indicate that serration outgrowth requires the maintenance of cell proliferation.

Figure 6.

 Over-expression of AtKRP1 in the margin blocks lobe outgrowth and leads to formation of cup-shaped leaves.
(a, b) Leaf silhouettes (L7–L14) of 22-day-old E4907 (a) and E4907>>KRP1 (b) plants. To enable flattening of the leaves, the central portion was removed prior to scanning. A smoother perimeter is observed in the E4907>>KRP1 leaves.
(c, d) Details of perimeters from E4907(c) and E4907>>KRP1 (d) leaves [boxed in (a) and (b)]. Lobe initiation pegs are visible in the E4907>>KRP1 leaves but they do not grow to form lobes.
(e) RT-PCR analysis reveals an elevated level of KRP1 transcript (arrowhead) in E4907>>KRP1 plants compared to the E4907 control lines.
(f–i) Cell size increases in the epidermis (f, g) and parenchyma (h, i) of the margin region of E4907>>KRP1 plants (g, i) compared with the control E4907 driver line (f, h).
(j) Young E4907>>KRP1 leaf showing buckling to form a cup shape.
(k) Pairwise adaxial and abaxial images of leaves from 38-day-old E4907>>KRP1 plants showing abaxial curling of the lamina edge and spoon-shaped leaves.
Scale bars = 1 mm (a, b, j), 200 μm (c, d), 50 μm (f, i) and 5 mm (k).

Molecular analysis confirmed an elevated level of AtKRP1 transcript in the E4907>>KRP1 lines (Figure 6e), and analysis of leaf histology revealed an accumulation of larger cells in the region of the margin (both epidermis and parenchyma), extending inwards from the leaf perimeter (Figure 6f–i). Analysis of leaves at a very early stage of development also indicated accumulation of larger cells around the leaf perimeter and tip in the E4907>>KRP leaves, extending inwards from the margin where GFP expression was detectable (Figure 7), consistent with the expected pattern of KRP1 accumulation.

Figure 7.

 Early-stage E4907>>KRP1 leaves have an increased epidermal cell size in the tip and margin.
(a, b) Cellular outline of the abaxial epidermis of leaf L3 from 7-day-old control E4907 (a) and E4907>>KRP1 (b) plants. Larger cells are visible at the leaf tip and the leaf margin in the E4907>>KRP1 leaf, but the cell size in the mid-vein and flanking regions are similar in both lines.
(c, d) Bright-field image of the leaves in (a) and (b) from which cellular outlines were drawn.
(e, f) GFP expression in the leaves shown in (a) and (b), indicating the site of AtKRP1 expression driven by E4907.
Scale bars = 50 μm.

Previous work involving mechanical modelling of plant tissue has indicated that repression of growth around the periphery of a growing planar body (such as a leaf) should automatically lead to the tissue buckling out of plane to produce a 3D cup shape (Green, 1992; Selker et al., 1992; Nath et al., 2003; Coen et al., 2004). As predicted by these models, KRP1-mediated growth suppression in the leaf margin led to formation of leaves with a distinctive 3D cup shape (Figure 5j). However, as growth of the leaf progressed and KRP1 over-expression (and consequential growth repression) appeared in the abaxial epidermis, ‘flipping’ of this cup occurred so that the proximal edge of the leaves curled downwards while the leaf tip curled upwards (Figure 6k). These data show that imposition of growth suppression initially around the margin but later across the abaxial/adaxial axis was sufficient to affect 3D leaf form.


Growth repression and leaf morphogenesis

Our experiments with the CUC2>>KRP1 and CUC2>>KRP1miR164ts plants indicate that spatial control of growth suppression can be used to control aspects of 2D leaf shape. Enhanced repression of tissue growth via termination of cell division in the CUC2 domain accentuated the normal pattern of growth suppression to the extent that individual leaflet-like organs were generated from Arabidopsis leaves. These leaf sub-domains did not themselves show any sign of dissection, and did not express a spectrum of genes associated with leaf dissection, and thus are highly unlikely to represent a truly dissected leaf form; nevertheless, they demonstrate the potential use of growth repression to alter leaf form. When growth suppression was targeted to the non-CUC2 domain, leaf morphology was altered in a predictable manner, with forms reminiscent of those seen in some relatives of Arabidopsis (e.g. some Cochlearia species). These results demonstrate the potential for manipulation of form by growth repression, are consistent with (although do not prove) the hypothesis that the CUC2 domain represents a region of growth suppression (Nikovics et al., 2006), and support the hypothesis that local growth suppression might be part of an endogenous mechanism of leaf morphogenesis.

Previous work on constitutive over-expression of KRP-related proteins has shown that this leads to leaves that are smaller than normal and have deeper indentations than wild-type leaves (Wang et al., 2000; De Veylder et al., 2001). Our data on the effect of induction of AtKRP1 throughout the leaf support these observations (Figure S2). However, when AtKRP1 over-expression was limited to the leaf perimeter, a smoother margin resulted (Figure 6). A change in perimeter shape requires differential growth at points around the perimeter (Figure 1). Thus, imposition of decreased growth around the whole leaf perimeter in E4907>>KRP1 plants should lead to more uniform growth around the perimeter, decreasing the gradients of growth that normally arise. If this happens, a smoother perimeter results (as observed). Why uniform KRP over-expression leads to deeper leaf dissection is difficult to interpret. As previously suggested, one possibility is that response to KRP over-expression is not uniform throughout the leaf, possibly due to a higher endogenous level of positive regulators of the cell cycle (e.g. cyclins) in the emergent serrations/lobes (De Veylder et al., 2001), so that these regions undergo less growth repression in response to a given elevated KRP level. Thus, in the case of uniform KRP over-expression, the growth gradients around the perimeter might actually be amplified, as growth in the inter-serration region is massively repressed whereas growth is maintained to some extent within the serrations. Further investigation of the endogenous spatial/temporal pattern of cell-cycle regulators within the developing leaf may help to resolve this issue. An additional factor influencing the phenotype of 35S::KRP plants may be a decrease in leaf blade growth caused by the reduction in cell number in the inner area of leaf, acting as a drag on growth at the leaf perimeter. How this might interact with potential growth differentials around the leaf circumference is unclear and requires further investigation.

With respect to the immediate downstream targets of the CUC2-related transcriptional network in Arabidopsis, there is no evidence (as far as we are aware) that the relative growth suppression observed in this region involves termination of cell division via AtKRP1 or related genes. Rather, we believe the data presented here are valuable in providing an artificial system in which leaf form can be altered by targeted spatial and temporal control of growth suppression, indicating that growth suppression could act as the endogenous mechanism by which leaf form is generated. Various ‘omic’ techniques are now being used to characterize targets for the transcriptional networks controlling leaf form (Sun et al., 2010), and it will be interesting to see whether the downstream processes identified by such approaches reveal growth-repressing mechanisms. As mentioned above, there are essentially three possibilities: growth increases relatively in the region where a serration/lobe occurs; growth is maintained within the serration/lobe while adjacent tissue undergoes a relative decrease in growth; or a combination of these two options. Our data support the potential importance of growth suppression, and will thus help to guide future investigations in this area.

It is interesting to note that the pattern of vasculature in the leaflet-like organs reported here was appropriate for the shape of the organ (Figure 2d). At the time of initiation of altered morphogenesis, this vascular pattern was not present in the responding tissue, and thus must have formed subsequently, consistent with the view of vascular patterning as a self-regulating system in which pattern accommodates to form (Rolland-Lagan and Prusinkiewicz, 2005). As our data indicate that the tools used to modify leaf shape act downstream of the patterning elements involved in leaf and vascular formation, they may provide a useful background to investigate the relationship between these elements (Dengler and Kang, 2001). Our data also indicate that growth events targeted to the margin can influence leaf shape, supporting the proposed importance of this region of the leaf in aspects of leaf morphogenesis (Reinhardt et al., 2007).

As well as 2D shape, co-ordination of growth is necessary for appropriate 3D leaf architecture. Defects in this mechanism can be observed in brassinosteroid and transcription factor mutants, for example, in which altered growth across the leaf surface leads to abnormal curling and buckling of the lamina (Nath et al., 2003; Savaldi-Goldstein et al., 2007). These experiments provided data supporting earlier (primarily theoretical) work involving mechanical modelling of plant tissue, which indicated that altered gradients of growth across the plane of a leaf lamina automatically lead to altered 3D form. As with previous work on 2D leaf shape, these earlier investigations involved manipulation of upstream regulators rather than downstream effectors of growth. The experiments reported here extend these earlier investigations by showing that targeted manipulation of a downstream growth repressor can indeed be used to alter 3D form in a predictable manner. Thus, the experiments with the E4907>>KRP1 plants showed that an altered pattern of cell size was generated across the plane of the young leaf, consistent with an expected repression of growth at the periphery, while internal cells were still free to grow and divide. The observed buckling of the leaf lamina to form a cup shape was consistent with the predicted outcome of the mechanical models, suggesting that the models do indeed capture an important element of leaf morphogenesis.

Approaches to manipulating and understanding leaf form

The results obtained here provide an indication of the possible impact of growth suppression on leaf shape formation. The endogenous mechanisms regulating leaf shape formation are likely to be far more complex than simple expression of a single growth repressor, and the cellular events leading to tissue growth arrest are likely to be different to those caused by local AtKRP1 over-expression. However, observing how a system responds to artificial perturbation provides an insight into the way in which that system is set up. Indeed, for complex network systems (such as a leaf), such observations may provide insights that are not afforded by simple mutation of an endogenous system in which internal feedback loops within the network may act to dampen any response. In a manner analogous to computational approaches, in which parameter space can be explored to identify factors that robustly influence a particular feature of a system, by artificially manipulating growth it is possible to characterize the spatial and temporal limits within which this parameter acts to influence the system (Scott et al., 2010). Having characterized these limits, it may be possible to ‘reverse engineer’ the system to discover the endogenous pathways that regulate that parameter. The data presented here thus establish the potential role of growth suppression in leaf morphogenesis, and set some spatial/temporal boundaries within which the endogenous system could work.

Experimental procedures

Plant material and growth conditions

All plants used in our experiments were in the Col-0 background. For Dex treatments, seedlings were transplanted onto 0.5× MS medium supplemented with 0.1% v/v DMSO with or without 10 μm Dex for specific times before analysis. Growth conditions were 100 μmol m−2 sec−1 light, a 16/8 h photoperiod and temperature 20/18°C (light/dark). For RT-PCR analysis of CUC2>>KRP1 plants, two samples were generated by dividing 16-day-old plants into a ‘shoot apex’ sample (shoot meristem with leaf primordia smaller than 1 mm) and ‘leaf’ sample (containing leaves ranging in size from 1 to 6 mm).

Light microscopy

Epidermal and mesophyll cells were observed after chloral hydrate or commercial bleach clearing using a BX51 Olympus microscope (Olympus, For GUS histochemical analysis, plants were pre-treated with 90% ice-cold acetone, and further assay was performed as described by Jefferson et al. (1986). The substrate solution contained 5 mm each of potassium ferricyanide and ferrocyanide. For further microscopical observation, GUS-stained plant material was dehydrated by passing through an increasing gradient of ethanol/Histoclear ( solutions and embedded in Paraplast X-tra (Sigma-Aldrich, Material was sectioned using a Leica 2145 microtome (Leica Instruments GmbH, Sections were gradually deparaffinized, hydrated and counterstained using 0.05% w/v safranine-O solution. After mounting in 50% w/v glycerol solution, images were taken using a charge-coupled device (DP71; Olympus) mounted on a light microscope (BX51; Olympus). GFP fluorescence observation was performed using an Olympus BX51 microscope with 470–490 nm excitation and a 515–550 nm barrier filter (narrow GFP band pass) or 330–385 excitation and a 420 nm long pass filter. In order to assess the pattern of GFP fluorescence across the leaf, hand sections were taken after embedding in ultra-low melting point agarose (Sigma-Aldrich). For shape observation, leaves were fixed in ethanol/acetic acid (7:1 v/v) and hydrated with 50% v/v aqueous ethanol. Depending on the leaf size, images were taken using a charge-coupled device (Diagnostic Instruments Inc., mounted on a Leica MZFLIII stereomicroscope, or with a DP71 charge-coupled device (Olympus) mounted on a BX51 light microscope (Olympus). In order to assess leaf perimeter changes in E4907>>KRP1 plants, the central part of the leaf had to be removed to facilitate leaf flattening. For leaf shape changes, observations on at least ten plants were performed per treatment.

RNA extraction, cDNA synthesis and PCR

Total RNA was extracted using TRIZOL (Chomczynski and Sacchi, 1987). For shoot apex samples, 10 μg of RNase-free glycogen solution (Invitrogen, was added to 50 μl of extract to facilitate isopropanol-mediated RNA precipitation. For semi-quantitative RT-PCR analysis, 4 μg of each RNA pool was treated with DNA-free DNAse I (Ambion/Applied Biosystems, followed by first-strand cDNA synthesis using MMLV reverse transcriptase (Promega, For further semi-quantitative transcript analysis, cDNA was amplified using BIO-TAQ™ polymerase (Bioline, using specific primers (Table S1) for the following genes: AtCUC1, AtCUC2, AtCUC3, AtKNAT1, AtSTM, AtLOB, AtBOP1, AtKRP1, AtHis2a and AtHisH4. Products of amplification were collected every three cycles, starting from cycle 26 and ending at cycle 36. Amplification signals from the exponential phase of amplification for each reaction were used for analysis.

In order to detect the miR164 cleavage product, 8-day-old pOpON::KRP1miR164ts plants were transferred onto 0.5× MS medium containing 10 μm DEX. A single-stranded RNA adapter phosphorylated at its 5′ end was ligated to the 3′ end of mRNA extracted from 11-day-old plants using T4 RNA Ligase 1 (Promega). cDNA synthesis was performed using an adapter-specific primer and MMLV reverse transcriptase. The resulting cDNA was used for 3′ RACE amplification with a forward primer specific to the AtKRP1 gene and a reverse primer specific to the adapter. The product of amplification obtained was sequenced in order to confirm the cleavage site.

DNA cloning and transgenic plants

In order to obtain pOpON::KRP1 plants, a KRP1109 deletion fragment (Weinl et al., 2005) was amplified using primers 5′-CACCATGGAATTTGAATCGGCGGTTAAAG-3′ and 5′-TCACTCTAACTTTACCCATTCGT-3′, cloned into the pENTR/D-TOPO® shuttle vector (Invitrogen) and then recombined into the popON2.1 vector (Samalova et al., 2005) using LR clonase II (Invitrogen). For popON::KRP1mi164ts plants, the miR164 binding site (Nikovics et al., 2006) was added upstream of the KRP1109 fragment using the reverse primer 5′-TGGAGAAACAGGACACGTGCTTCACTCTAACTTTACCCATTCGT-3′, with further cloning procedures identical to those described for popON::KRP1 plants.

For the first component of the CUC2>>KRP1 plants, a 3151 bp fragment was amplified from the 5′ CUC2 upstream region using forward primer 5′-CACCGTGGTCGACTAGAGGAAGA-3′ and reverse primer 5′-TAAGAAGAAAGATCTAAAGCT-3′. A genomic clone of the CUC2 promoter in the pBS (KS) vector (Strategene, was used as a template for the PCR reaction. The amplified fragment was subsequently cloned into the pENTR-DTOPO vector, and then recombined into the pBIN-LR-LhGR2 vector (Samalova et al., 2005) using LR clonase (Invitrogen). For the second component of the CUC2>>KRP1 plants, SalI and BamHI restriction sites were added to a KRP1109 fragment using primers 5′-acgcGTCGACATGGAATTTGAATCGGCGGTTAAAG-3′ and 5′-acgcGTCGACATGGTGAGAAAATATAGAAAAGCT-3′, and, after restriction digest, fragments were ligated into a pV-TOP vector (Craft et al., 2005) using T4 DNA ligase (Promega). After transformation into Arabidopsis (Clough and Bent, 1998), homozygous progeny for each component (pBIN-CUC2-LR-LhGR2 and pV-TOP-KRP1) were obtained and crossed. Progeny were self-pollinated, and subsequently selected using the appropriate antibiotic resistance markers for the vector components, and genotyped using specific primers. Homozygous progeny (CUC2>>KRP1) were used for further experiments.

To achieve margin-localized expression of KRP1, the pENTR/D-TOPO®-KRP1109 shuttle vector fragment described above was recombined into the pUAS-pAM-PAT-GW vector (Weinl et al., 2005) and transformed into Arabidopsis (Clough and Bent, 1998). Homozygous transgenic plants pUAS-pAM-PAT-GW::KRP1 were crossed with the E4907 enhancer trap driver line, and the homozygous progeny (E4907>>KRP1) were used for further experiments. The E4907 line was originally identified from a screen of a library generated in the Poethig laboratory (Department of Biology, University of Pennsylvania, Philadelphia, PA), and is available from the Nottingham Arabidopsis Stock Centre.


We thank Ian Moore (Department of Plant Sciences, Oxford University, UK), Arp Schnittger (IBMP, CNRS Strasbourg, France) and Mitsuhiro Aida (Nara Institute of Science and Technology, Japan) for the gifts of vectors and clones, and Scott Poethig (Department of Biology, University of Pennsylvania, Philadelphia, PA, USA) for the original E4907 line. This work was supported by an EU Transfer of Knowledge grant (A.F.) and a Marie Curie FP7 Intra-European Fellowship (A.K.).