Distal expression of knotted1 in maize leaves leads to re-establishment of proximal/distal patterning and leaf dissection

Zea mays (maize) leaves provide a useful system to study how proximal/distal patterning is established because of the distinct tissues found in the distal blade and the proximal sheath. Several mutants disrupt this pattern including the dominant knotted1-like homeobox ( knox ) mutants. knox genes encode homeodomain proteins of the TALE superclass of transcription factors. Class I knox genes are expressed in the meristem and down-regulated as leaves initiate. Gain-of-function phenotypes result from misexpression in leaves. We identified a new dominant allele of maize knotted1 , Kn1-DL , which contains a transposon insertion in the promoter in addition to a tandem duplication of the kn1 locus. In situ hybridization shows that kn1 is misexpressed in two different parts of the blade that correlate with the different phenotypes observed. When kn1 is misexpressed along the margins, flaps of sheath-like tissue form along the margins. Expression in the distal tip leads to premature termination of the midrib into a knot and leaf bifurcation. The gain-of-function phenotypes suggest that kn1 establishes proximal/distal patterning when expressed in distal locations and lead to the hypothesis that kn1 normally participates in the establishment of proximal/distal polarity in the incipient leaf.


INTRODUCTION
Plants produce organs throughout their life span from meristems, groups of self-renewing cells whose derivatives become the roots and shoots of a plant (Veit, 2006). The shoot apical meristem (SAM) initiates the leaves and stem, while the root apical meristem is responsible for generating the root system. The SAM initiates leaves in a defined pattern, called phyllotaxy, and at defined intervals, referred to as plastochrons. A leaf in plastochron 1 (P 1 ) has just emerged from the meristem, while a P 2 leaf is one plastochron older. From observing the position of these leaf primordia, one can predict the position of the next leaf (P 0 ) while it is still part of the meristem. Given that meristem cells are indeterminate and leaf cells are determinate, a major question in plant biology is how cell fate changes during the transitions from meristem to P 0 and then to leaf.
In the last decade, a number of genes have been identified that are expressed in meristems and not in leaves and thus provide insight into the process of leaf initiation. The class I knotted1-like homeobox (knox) genes are expressed throughout the SAM except in the P 0 cells (Jackson et al., 1994). It has been hypothesized that down-regulation of knox genes at P 0 is required for leaf initiation. Indeed, in species with dissected leaves where knox genes are expressed in leaves, knox down-regulation still occurs in the P 0 (Hay and Tsiantis, 2006;Champagne et al., 2007). The importance of knox genes for plant development is highlighted by loss-of-function mutants. Plants carrying mutations in the Arabidopsis SHOOT MERISTEMLESS (STM) gene terminate after the cotyledons form, and further growth occurs from adventitious meristems that may also terminate (Long et al., 1996). The absence of a functional maize knotted1 (kn1) gene also produces a shootless phenotype in certain inbred backgrounds (Vollbrecht et al., 2000). In other backgrounds, vegetative development is normal but plants have reduced axillary branching, and ectopic leaves form in the axils of leaves (Kerstetter et al., 1997;Lunde and Hake, 2009).
Both recessive and dominant mutations have been found that cause misexpression of knox genes in leaves (Hake et al., 2004). The recessive mutants identify regulatory proteins that normally repress knox expression in leaves (Schneeberger et al., 1998;Timmermans et al., 1999;Tsiantis et al., 1999;Byrne et al., 2000;Ori et al., 2000;Scanlon et al., 2002;Ha et al., 2003;Alexander et al., 2005;Xu and Shen, 2008). Gain-of-function knox phenotypes are known in maize, barley and Antirrhinum.. These dominant mutations are due to changes in cis-regulatory sequences of the knox genes and emphasize the complex regulation needed to keep knox genes off in the leaf (Greene, 1993;Greene et al., 1994;Müller et al., 1995;Schneeberger et al., 1995;Mathern and Hake, 1997;Golz et al., 2002;Uchida et al., 2007;Guo et al., 2008). Analysis of the maize dominant knox mutants highlights a consistent theme of proximal tissues displaced into distal tissues. The timing of knox misexpression was hypothesized to be critical to the mutant phenotypes (Freeling, 1992;Muehlbauer et al., 1997).
Here we describe a new kn1 allele, Kn1-DL, which results from alterations in the promoter region. kn1 misexpression at margins of the developing blade leads to flaps of sheath tissue at blade margins. kn1 misexpressed at the midrib and tip of the blade leads to a bifurcated leaf, which is highly unusual for grass species. The Kn1-DL phenotype indicates that kn1 reinitiates new proximal/distal patterns, suggesting it may play a role in establishing proximal/distal polarity during normal leaf development.

Phenotype of Kn1-DL
Maize leaves have a proximal sheath that wraps around the stem, a distal blade that tilts away from the stem, and the ligule/auricle region at the junction of blade and sheath ( Figure 1A, B, G) (Sylvester et al., 1990). The result of knox misexpression is a displacement of proximal tissues, such as sheath, auricle or ligule, into the blade region. This phenotype has been characterized for two other dominant alleles, Kn1-N and Kn1-O (Gelinas et al., 1969;Freeling and Hake, 1985). In Kn1-N, outpockets of tissue, called knots, occur along lateral veins of the blade ( Figure 1F). The cells overlying the veins resemble those found in sheath tissue (Sinha and Hake, 1994). Occasionally, ligule tissue is found running parallel along the veins ( Figure 1E). Kn1-O has a stronger effect on the ligule region, leading to a complete absence of ligule on the first leaf ( Figure 1H) (Mathern and Hake, 1997). In comparison, the ligule of Kn1-N has small gaps at lateral veins ( Figure 1I). On older Kn1-O leaves, knots are found toward the midrib, coinciding with displaced ligule on the adaxial side ( Figure 1C, D).

7
The presence of these hairs in Kn1-DL leaf margins suggests the acquisition of sheath-like cell identity. The direction of curling, toward the adaxial surface, is the same as seen in normal sheath. Flaps are also visible along the margin (Figure 2A, B, E). These are initially less than 3-4 mm (leaf 6) and become larger and more abundant on later formed leaves ( Figure 2A, B). Reorientation of the vasculature occurs at the position of the flaps ( Figure 2E).
In addition to the margin flaps, the vein clearing that is common to other Kn1 alleles is occasionally detectable in lateral veins of the blade, particularly in the most distal part ( Figure   2E, I, K). The "clear veins" have sheath cell identity (Sinha and Hake, 1994).
The Kn1-DL defect is also seen at the midrib. Small knots are occasionally found on young leaves that appear similar to knots on lateral veins in Kn1-N. However, on the last few leaves, a knot forms at the midrib that appears to lead to a bifurcated leaf. In a family of 13 mutants in the B73 background, the last leaf, leaf 20, was either a small flap of sheath tissue (3/13) (not shown) or bifurcated to the base of the leaf (10/13) ( Figure 2F). The bifurcation was always at the position of the midrib, and tissue with sheath-like hairs could be seen along the margin created by the bifurcation ( Figure 2I). Leaf 19 was bifurcated into the blade ( Figure 2G), ending in a large knot, occasionally with a flap of ligule visible inside the knot ( Figure 2J). The depth of this bifurcation ranged from 20 to 80% of the length of the blade.
Leaf 18 was bifurcated 5 to 13% of the length of the blade ( Figure 2H, K), also ending in a knot. The midribs of leaves 15-17 often ended in a small tendril, but were not bifurcated ( Figure 2N). Thus, the bifurcation was more severe as the leaf position approached the tassel.
Interestingly, the margins of leaves that were bifurcated showed fewer or no flaps, but were hairy and sheath-like.
The Kn1-DL defect is also visible on leaves of axillary branches. Two types of axillary branches are found in vegetative nodes of maize plants: the ear, in the axil of ~leaf 14, and tillers, which arise from basal nodes. Ears are enclosed in approximately ten husk leaves. In the B73 background, normal husks are triangular in shape and composed primarily of sheath tissue. In Kn1-DL, the distal margin of all the husk leaves is completely disrupted, and the husks do not come to a point ( Figure 2L, M). In contrast, the subtending leaf in which the ear arises is only mildly affected with a few margin flaps. Tillers are rare in most inbred backgrounds, thus could not be evaluated in the introgressed Kn1-DL plants, but when they did form in mixed backgrounds, each leaf was affected (not shown).
To help determine the type of tissue present in the Kn1-DL flaps, we made handsections through normal and Kn1-DL leaves. In normal blades, veins are centrally located and most cells have high chloroplast content ( Figure 3A). Macrohairs are adaxially located   Sylvester et al., 1990;Candela et al., 2008). In auricle tissue, veins are also centrally located, but only a few cells contain chloroplasts, mainly on the abaxial side adjacent to the epidermis ( Figure 3B). Occasional long hairs can be seen near the adaxial margin ( Figure 3C). In the sheath, the vasculature is abaxial and chloroplasts are found associated with the veins. Large parenchyma cells separate the veins from the adaxial epidermis ( Figure 3D). The margins of sheath have extremely long hairs ( Figure 1B, 3H, 3I).
In Kn1-DL flaps, the tissue resembles sheath or auricle in many ways. The number of cells showing accumulation of chloroplasts is drastically reduced compared to normal blade.
Veins are mostly abaxial ( Figure 3F), though occasionally they are central ( Figure 3G). Hairs are sometimes found on both adaxial and abaxial surfaces ( Figure 3E), but often they are only present on the abaxial side ( Figure 3F, G), as in sheath. We also sectioned through a midrib knot that was coincident with blade termination ( Figure 3J, K). Compared to the midrib in a normal portion of the same leaf ( Figure 3L), the region of the knot has very long hairs, similar to the sheath margin ( Figure 3H, I). These hairs flank a region with hairs similar to the sheath.
The vasculature and photosynthetic cells are highly abnormal.
In summary, Kn1-DL has a major effect on the distal portion of leaves. In early leaves, blade tissue is replaced by sheath/auricle-like tissue at the margins. This transformation results in the appearance of sheath hairs, flaps and curled margins. On leaves initiated late, the midvein becomes progressively more affected and terminates in a tendril or a large knot, and the leaf becomes bifurcated.

The 5' region is altered in Kn1-DL
Most dominant Kn1 alleles result from transposon insertions into the third or fourth intron (Hake et al., 1989;Veit et al., 1990;Greene et al., 1994), but the Kn1-O allele results from a tandem duplication of the gene (Veit et al., 1990). We used DNA gel blots and PCR to determine the cause of the Kn1-DL mutation. Digestion with SacI and hybridization with a probe specific to the third intron showed no difference between Kn1-DL mutants and normal siblings (Supplemental Figure 1). This combination of enzyme and probe would have detected the presence of an insertion in the third intron, where 12 of the previously identified dominant alleles have transposons insertions (Hake et al., 1989;Greene et al., 1994;Ramirez, 2007). In contrast, digesting the DNA with BclI, which cleaves upstream, downstream and within the coding region ( Figure 4A), produced a unique hybridization pattern in Southern blots. Using a probe spanning the two BclI sites within the coding region, we found that the 0.5 Kb fragment that spans the fourth exon and the 3.1 Kb fragment that extends into the 3' end are the same in mutants and normal siblings, but the 9.1 Kb fragment that extends into the 5' end is replaced by two fragments, a 6.2 and >12 Kb fragment ( Figure 4B). The 9.1 Kb fragment is the same size as the band detected with DNA from normal siblings, the progenitor (from which the Kn1-DL mutant originated), and all the inbred backgrounds tested (data not shown). Finding two bands instead of one suggests Kn1-DL contains a partial or complete duplication of the kn1 locus that disrupts the 5' BclI site. We compared the BclI digested DNA from Kn1-DL plants with that from Kn1-O using a probe specific to the third intron. The DNA from Kn1-O generated two bands, one in common with the normal siblings and another band slightly smaller ( Figure 4C). Thus, the duplications found in Kn1-O and Kn1-DL are distinct. To confirm that the duplication is linked to kn1, we carried out DNA gel blot hybridization with 75 Kn1-DL individuals and found that they all produced the same pattern (data not shown), indicating that the duplication is closely linked to the phenotype. Moreover, because fragments in the 3' end of the kn1 locus are identical in normal and Kn1-DL, the duplication is likely to be 5' of the gene.
Because Kn1-DL arose in a line carrying active Mutator (Mu) elements, we used PCR to determine if a Mu element could be identified in the Kn1-DL allele. We were able to amplify a fragment using a Mu primer, which recognizes the terminal inverted repeats common to all Mu elements, and a primer designed from the 5' region ( Figure 4E). This fragment was only detected using DNA from knotted individuals but not from normal siblings or progenitor. We localized the Mu element 297 bp upstream of the start codon. Primers flanking this Mu insertion were able to amplify a fragment in normal siblings and heterozygotes, but not presumed homozygotes ( Figure 4E Mu-containing Kn1 alleles to MuKiller, a line of maize that suppresses Mu activity (Slotkin et al., 2005). MuKiller was able to effectively suppress the knotted phenotype in all other Mucontaining alleles except Kn1-DL (Table 1). Either the Mu insertion at Kn1-DL is unique, or the cause of kn1 misexpression is not only the Mu element, but also the rearrangement resulting from the duplication.

mRNA accumulation in leaves
To determine the size of the mRNA in Kn1-DL, we carried out RNA gel blot analysis. We detected in Kn1-DL leaf primordia a band of the same size as the one detected in meristemenriched tissues of normal or mutant plants ( Figure 4D). Similarly, RT-PCR performed with a primer pair able to amplify a nearly full length cDNA showed a band of the expected size ( Figure 4F). The same primer pair was used to determine when ectopic kn1 transcripts could be detected in leaf tissue of Kn1-DL. When using leaves dissected from 10-day-old seedlings, kn1 cDNA was detected in Kn1-N but not Kn1-DL nor B73 after 33 cycles of PCR, but a faint band was detected in Kn1-DL after 40 cycles of PCR. In our growth conditions, those seedlings had two visible leaves with the third just emerging, thus the tissue harvested included mainly developing leaf primordia three and four. kn1 cDNA was easier to detect with 18-day-old Kn1-DL seedlings, which included leaf 5 and younger, i.e., subsequently initiated leaves, and 26-day-old seedlings, which included leaf 7 and younger. Accumulation To investigate the spatial pattern of kn1 mRNA accumulation, we carried out in situ hybridization. As described previously, the Kn1-N defect is restricted to lateral veins of the leaf blade ( Figure 1F). Our analysis of Kn1-N shoot apices confirmed previous in situ results (Smith et al., 1992;Jackson et al., 1994), detecting the mRNA in lateral veins of blades but not sheaths ( Figure 5B). No signal could be detected in leaves of normal siblings ( Figure 5A).
In order to examine the expression of kn1 in Kn1-DL leaf primordia, we sectioned shoot apices of plants that had just transitioned to flowering, thus allowing us to visualize kn1 accumulation in the last leaves initiated by the SAM, which show the most severe phenotype.
The most basal section in the series in Figure 5 shows expression of kn1 in the SAM but not in leaves, with the exception of a small dot of expression in a margin of the leaf l7 ( Figure   5C). A few sections above ( Figure 5D), strong kn1 accumulation is detected in a tassel branch as well as in many leaves. kn1 expression in leaves 15-19 is visible mainly at the margins.
Expression in leaf 20 is scattered throughout the leaf, including around the midrib. These leaves did not show any kn1 expression at their most basal part ( Figure 5C). Figure 5E shows kn1 expression in two tassel branches and at the margin of leaves 17 and 18, and throughout leaf 19 including the midrib. Leaf 20 is not visible in this section, and thus is a very young leaf. Figure 5F and 5I show expression throughout leaf 18 and at leaf margins. Leaf 19 is no longer visible. In Figure 5G, expression is only detected at margins for leaves 15-17 while leaf 18 primordium is no longer visible. In the last panel of the series, 1-2 mm above the tip of the tassel primordia, kn1 expression is mainly at margins. Thus, kn1 is primarily misexpressed in the distal part of leaf margins except in the upper three leaves, where expression occurs throughout the tip of the leaf primordia.  , 1994). We propose that proximal/distal polarity may also be initiated in the P 0 and that KN1 participates in this process ( Figure 6A, B). kn1 mRNA is present in meristem cells, excluding the P 0 . However, because KN1 protein moves, it is found in cells at the boundary between meristem and P 0 (Jackson, 2002), coinciding with the proximal end of the leaf. Thus, the absence of KN1 in most P 0 cells leads to cell division and differentiation in the leaf primordium, but its presence at the boundary of P 0 and meristem retards these activities, possibly through negative regulation of GA (Hay et al., 2002;Jasinski et al., 2005;Bolduc and Hake, 2009).

If kn1 establishes proximal/distal patterning, what happens when it is missing?
Depending on inbred, kn1 loss-of-function mutants may terminate with a coleoptile or one or two leaves (Vollbrecht et al., 2000). In the embryo, a flattened zone of differentiated cells is found where the meristem would normally reside. The failure to make leaves has been attributed to a failure to maintain the meristem. It is also possible that leaves do not form because kn1 is required to establish the proximal boundary of the leaf. A similar failure to initiate leaves can occur when there is loss of adaxial fate identity following overexpression of genes that specify abaxial fate (Eshed et al., 2001). Indeed, ectopic leaf flaps develop when adaxial cells are found in a field of abaxial cells (Waites and Hudson, 1995;Timmermans et al., 1998;Candela et al., 2008). Considering the kn1 loss-of-function and gain-of-function mutant phenotypes, a similar juxtaposition of proximal and distal cells may be required to elaborate a leaf. The expression of kn1 at the margins of the leaf in Kn1-DL creates a juxtaposition of proximal cells (kn1-expressing) with distal cells (the blade). This juxtaposition reiterates the proximal/distal patterning that normally occurs in the P 0 , leading to a reprogramming of proximal/distal polarity and the formation of leaf flaps ( Figure 6E). kn1 misexpression in the lateral veins of Kn1-N blades allows these cells to adopt more proximal cell identities, but due to the constraints by blade cells on either side, the sheath-like cells grow out of the plane of a leaf, making a knot ( Figure 6D). We speculate that the correct balance of abaxial with adaxial as well as proximal with distal determinants is required for normal leaf initiation, while inappropriate juxtaposition of proximal with distal or adaxial with abaxial cell types leads to ectopic leaf flaps.
In contrast to the kn1 expression patterns in Kn1-N and Kn1-DL, misexpression seen in other dominant knox mutants appears to be continuous with the meristem. In Gnarley, expression extends from the meristem up into the proximal end of early plastochron leaves (Foster et al., 1999). In Roughsheath1 (Rs1), misexpression is seen in P1 and P2 leaves (Schneeberger et al., 1995). In the dominant Lg4 mutant, misexpression of liguleless4 (lg4) was found by RT-PCR in the sheath but not in the blade (Muehlbauer et al., 1999). These three mutants have no ectopic leaf flaps or knots in the blade; rather, the boundary between sheath and blade is displaced toward the distal part of the leaf ( Figure 6C) Timmermans et al., 1998). These narrow leaves are due to smaller groups of leaf initial cells, i.e. fewer cells in which kn1 expression is down-regulated. Occasionally, on more normal leaves, a knot is present in association with a bifurcation of the leaf somewhat similar to Kn1- DL (Timmermans et al., 1998). lbl1 encodes a protein in the trans-acting siRNA pathway that promotes adaxial fate (Nogueira et al., 2007). It is possible that misexpression of kn1 at the midvein interferes with abaxial/adaxial polarity establishment, leading to the tendrils seen at the tips of Kn1-DL leaves. Likewise, the loss of adaxial fate in lbl1 mutants may interfere with proximal/distal establishment, thus causing the bifurcation. Interplay between adaxial/abaxial and proximal/distal axes has been previously documented (Ha et al., 2004;Candela et al., 2008). Misexpression of knox genes leads to leaf dissection in Arabidopsis (Lincoln et al., 1994;Chuck et al., 1996) and increases leaf dissection in tomato (Hareven et al., 1996;Chen et al., 1997). Consistent with those findings, lowering and raising knox expression in Cardamine hirsuta, a member of the Brassica family with dissected leaves, affects leaf shape.
RNAi lines with reduced amounts of the C. hirsuta STM ortholog show a reduction in number of leaf lobes. In addition, C. hirsuta plants that express kn1 through an inducible KN1-GR construct show increased leaflets as well as a shorter proximal distal axis (Hay and Tsiantis, 2006). We propose that the misexpression of kn1 at the distal tip of Kn1-DL is responsible for dissection on the upper leaves ( Figure 6F). Given that ectopic kn1 mRNA increases with time in Kn1-DL, the severity at the midrib is likely due to increasing amounts of KN1 at the tip of www.plantphysiol.org on August 20, 2017 -Published by Downloaded from Copyright © 2009 American Society of Plant Biologists. All rights reserved. the leaf. Although these leaves are unusual, they reveal that grasses are not completely immune to architectural restructuring by knox misexpression.

ACKNOWLEDGEMENTS
We wish to thank members of the Hake lab, especially Devin O'Connor, for their insightful comments, George Chuck for help with in situ hybridizations, and Julie Calfas and David Hantz for their steadfast attention to our plants in the greenhouse. We thank Steve Ruzin for use of the Imaging Center of the Plant and Microbial Biology Department.

Genomic DNA analysis
Isolation of genomic DNA and DNA gel blots was carried out using previously published protocols (Lowe et al., 1992). Probes for Southern blots were generated using PCR with the The identity of amplified DNA was verified by sub-cloning into pGEM T-Easy vector (Promega) and sequencing. The identity of amplified DNA was verified by sub-cloning into pGEM T-Easy vector (Promega) and sequencing.

RNA Analysis
For Northern blot analysis, total RNA was isolated from two to three week-old seedlings using Trizol reagent according to the manufacturer's instructions (Invitrogen). Ten μg of total RNA were glyoxylated, separated by electrophoresis in a 1.2% agarose gel and transferred to nylon membrane as previously described (Smith et al., 1992). Hybridizations were performed with the full kn1 cDNA (Vollbrecht et al., 1991). Following autoradiography, filters were stripped and re-probed with an ubiquitin probe to assess RNA quality and quantity.
For RT-PCR analysis, total RNA was further purified with Qiagen RNeasy Mini Kit following the manufacturer's recommendations for RNA cleanup. Reverse transcriptions were performed at 46°C for three hours using 2 µg of total RNA, SuperscriptIII (Invitrogen) and oligo(dT) primers. PCR were then performed with 1 µl of cDNA using kn1-specific primers E47 (GAGATCACCCAACACTTTGG) and Kn1-3'R (ACATGAGCCGTACCATTAGATTAGG) and actin-specific primers ActinF (AAGTACCCGATTGAGCATGG) and ActinR (GATGGAGTTGTACGTGGCCT).

Histology and in situ hybridization
For histology analysis, hand-sections were mounted in 50% glycerol and viewed immediately using dark-field microscopy. For in situ hybridization, shoot apices of three to four week old seedlings were fixed overnight with FAA (50% ethanol, 5% glacial acetic acid, 3.7% formaldehyde, 0.5% Triton X-100, and 1% DMSO) and embedded in paraffin after staining with 0.5% (w/v) Eosin Y in ethanol as previously described (Jackson, 1992). Tissue sections were pre-treated and hybridized as previously described (Jackson et al., 1994).      kn1 marks the most proximal (P) boundary of the P 0 leaf. Similar to the adaxial (Ad)/abaxial (Ab) axis, the proximal/distal (D) axis is presumably already established in the P 0 . B-F) The expression pattern of knox genes in young leaf primordia is diagrammed on the left and the leaf phenotype at maturity is shown on the right. B) In a normal leaf primordium, knox expression is only at the base, at the site of leaf insertion, the most proximal end of the leaf.

SUPPLEMENTAL MATERIAL
The mature leaf has clearly defined proximal sheath (dark blue), distal blade (orange) and