Lobe-generating centres in the simple leaves of Myriophyllum aquaticum: evidence for KN1-like activity

Abstract

Background and Aims

The mature morphology of most plants can usually be said to consist of three mutually exclusive organs: leaves, stems, and roots. The vast majority of mature morphologies may be easily grouped into one of these mutually exclusive categories. However, during very early stages of development and in many instances from inception, the division between organ categories becomes fuzzy due to the overlap in developmental processes that are shared between the aforementioned mutually exclusive categories. One such overlap has been described at the gene level where KNOXI homologues, transcription factors responsible for maintaining indeterminate cell fate, are expressed in the shoot apical meristem and during early stages of compound leaf development. This study characterizes the occurrence and spatial localization of mRNA of a KNOXI homologue, MaKN1, during the early stages of development in the simple leaves of Myriophyllum aquaticum, an aquatic angiosperm from the family Haloragaceae exhibiting pentamerous whorls of finely lobed leaves.

Methods

A 300-bp KNOXI fragment was sequenced from M. aquaticum and used in an RNA localization study to determine the temporal and spatial expression of KNOXI during the early stages of leaf lobe development in M. aquaticum. The developmental sequence of leaves of M. aquaticum was also described using scanning electron microscopy.

Key Results

Lobe development of M. aquaticum occurs in two very distinct regions at the leaf base in an alternating fashion reminiscent of a distichous shoot system. It was discovered that MaKN1 expression is localized to both the shoot apical meristem and early stages of leaf primordia development (P1–P7). Initially, MaKN1 is expressed ubiquitously throughout primordia (P1–P3); however, as lobes develop, MaKN1 becomes localized to recently emerged lobe primordia, and disappears as lobes develop basipetally.

Conclusions

The pattern of gene expression is indicative of shared developmental processes during early development between shoots, compound leaves, highly lobed simple leaves and unifoliate simple leaves which lack KNOXI expression. These findings are supportive of Arber's less rigid ‘partial shoot’ theory, which conceptualizes compound leaves as having shoot-like elements.

INTRODUCTION

There are a myriad of morphologies represented within the plant kingdom, all of which are typically divided into three groups described as leaves, stems and roots (Goethe, 1790; De Candolle, 1813; Troll, 1937; Kaplan, 2001). A leaf is defined as a lateral structure, borne on an axis, the stem, and exhibits dorsiventral symmetry and a determinate pattern of development (Dengler and Tsukaya, 2001). Both leaves and stem constitute a shoot. Shoots and roots have indeterminate patterns of growth and typically have radial symmetry (Lyndon, 1998). Traditionally, angiosperm morphology has been analysed using this typological paradigm to categorize all observable phenotypes into one of the above categories (Sattler, 1986). However, there are numerous examples of plant morphology that cannot be easily placed into a single category (Rutishauser and Isler, 2001). The leaves of Gaurea are a well-known example of a phenotype that demonstrates both shoots and leaf-like characteristics in a single structure if the three main structural criteria of position, symmetry and determinate/indeterminate growth are taken into account (Fukuda et al., 2003). In this case, the ‘leaves’ are lateral structures that demonstrate an indeterminate pattern of growth (Sattler and Rutishauser, 1992).

Even typical angiosperm leaves show considerable diversity in their patterns of development and overall morphology. Compound leaves are divided into a central rachis which bears leaflets that may themselves be dissected (Goliber et al., 1999). Simple leaves lack the central rachis and are unifoliate, exhibiting a gradient in degree of leaf lobing. In some species this structural division becomes inaccurate from a developmental perspective where leaf primordia may be complex at inception but later engage in secondary morphogenesis to become simple at maturity (Bharathan et al., 2002). The KNOXI family of homeodomain transcription factors offer an example of overlap of organ categories at the molecular level between shoots, compound leaves and simple leaves (Chen et al., 1997; Lacroix et al., 2005).

Class-I KNOX

Class-I KNOX genes are expressed at the developing shoot tip and are responsible for maintaining the indeterminacy of the meristem (Jackson et al., 1994; Granger et al., 1996). Mutations in KNOXI genes result in failure to produce or maintain a shoot apical meristem, and irregularly shortened internodes, respectively (Long et al., 1996). KNOXI was first isolated from mutant maize which developed ectopic meristems on adaxial leaf surfaces (Vollbrecht et al., 1991). If removed from the lamina and cultured on the appropriate media, ectopic meristems developed into mature shoots (Sinha et al., 1993; Chuck et al., 1996). The leaf cells that expressed KNOXI failed to differentiate and remained isodiametric (Kessler and Sinha, 2004). Therefore, the ectopic presence of KNOXI in leaves appears to endow them with shoot-like characteristics (Sinha et al., 1993; Chuck et al., 1996). Maintenance of indeterminate growth through the suppression of cellular differentiation is a key function of KNOXI during shoot development (Bharathan and Sinha, 2001; Kellogg, 2006; Long and Benfey, 2006). Regions around the meristem where KNOXI expression is down-regulated permit the establishment of lateral organ polarity and cell fate, and correspond to the location of the next incipient leaf primordium, P0 (Smith et al., 1992; Sakamoto et al., 2001).

If KNOXI genes promote indeterminate growth, a shoot-like characteristic, it was then unexpected to discover that these genes remain turned on in species, such as Lycopersicon, which have compound leaves. In these species, KNOXI genes are expressed from leaf primordium inception and then down-regulated once leaflet organogenesis is complete (Goliber et al., 1999). It has been found that this presence of KNOXI expression confers a transient state of indeterminacy during leaf development which permits the elaboration of leaf form (Blein et al., 2010; Harrison et al., 2005a). Constitutive ectopic expression of KNOXI during early stages of leaf development can increase the complexity of simple leaves, or intensify inherent complexity resulting in super-compound leaves (Hareven et al., 1996; Müller et al., 2006). From a molecular point of view therefore, KNOXI expression appears to represent a dichotomy in leaf development where the presence or absence of expression results in either a simple or complex lateral organ morphology (Hake et al., 2004).

Complexities in leaves and shoots

If KNOXI expression in compound leaves represents inherent differences in meristematic potential between compound and simple leaves, then it implies that compound leaves and shoots are related on a fundamental level that transcends their traditional classification (Hareven et al., 1996; Hofer et al., 2001a). It has been observed in numerous studies that compound leaves are reminiscent of distichous shoots that lack axillary meristems (Lacroix et al., 2003). At the developmental level, the compound leaf primordia produce leaflets at discrete regions in much the same fashion that leaves are produced around a meristem (Lacroix, 1995). These findings indicate that some compound leaves may represent intermediate morphologies between shoots and simple leaves. If a morphological continuum is present, then a simple yet highly lobed leaf should fall between a compound leaf and a simple leaf in a morphological continuum. Based on this theory, it could be proposed that a highly lobed simple leaf may also express KNOXI genes during its development or exhibit a gradient of KNOXI expression falling somewhere between simple and compound leaf development. To the best of the authors' knowledge, there have been no studies to date done on varying degrees of KNOXI gene expression during leaf development. Quantitative studies on expression gradients during leaf and leaflet development is challenging due to difficulties in isolating leaf primordia tissue from the meristem. RNA in situ hybridization, while not quantitative, allows for the visualization of gene expression patterns in histological sections to determine the presence or absence of gene expression.

Members of the KNOXI family have been identified in all lineages of the plant kingdom, including angiosperms, gymnosperms, ferns and bryophytes; however most of our understanding concerning plant development and morphology comes from the model dicot Arabidopsis thaliana. Numerous examples exist where KNOXI genes are not expressed in the typical pattern, of which the Fabales are an example. Although many species in this family have compound leaves, none of them exhibit KNOXI expression during their development. Instead they express another gene, UNIFOLIATA (UNI), a developmental regulator or transcription factor, which appears to take the place of KNOXI in regulating compound leaf development (Champagne and Sinha, 2004). Conversely, some species which produce simple mature leaves such as Helianthus annuus and Lepidium oleraceum (Brassicaceae) exhibit KNOXI expression during their early stages of development (Bharathan et al., 2002; Tioni et al., 2003). In the subfamily Podostemoideae, where species such as Hydrobryum japonicum and Cladopus doianus demonstrate no typical shoot apical meristem, KNOXI expression is associated with early stages of leaf primordia as they arise below the base of mature leaves (Katayama et al., 2010). These examples suggest that KNOXI expression in leaves is not as straightforward as the simple-compound paradigm and show that further analysis of KNOXI expression in non-model species is warranted to appreciate the diverse morphologies present in the plant kingdom (Cronk, 2001; Bharathan et al., 2002; Nardmann and Werr, 2007).

Myriophyllum aquaticum

To elaborate further on species differences and the morphological relationship between KNOXI expression and leaf development, it would be beneficial to examine KNOXI in a species of plant that exhibits a leaf shape that is visually intermediate between simple and compound leaves. Myriophyllum aquaticum, commonly known as Parrot-feather, is an invasive aquatic angiosperm, found all over the Americas in freshwater streams, and exhibits mature leaves that are simple with an intensely lobed lamina (Fassett, 1972; Sutton, 1985; Rutishauser, 1999). Leaves are produced in whorls of four to six on fleshy stems and may contain upwards of 20 lobes (nine or ten pairs), depending on the environmental conditions (Sutton, 1985). The leaves are heterophyllous and demonstrate varying morphologies depending on whether they develop above or below water. The aerial leaves are fleshy and photosynthetic with fewer lobes compared with immersed leaves, which tend to have a brown, filamentous appearance (Fasset, 1972). This member of the Haloragaceae belongs to order Saxifragales which is morphologically diverse. Consequently, drawing phylogenetic relationships based on leaf morphology is not as informative as expected. However, according to Gerber and Les (1994), dissected leaves in many unrelated submerged angiosperms point to an adaptive function.

The highly lobed yet simple nature of the M. aquaticum leaf is a good system to investigate morphologies that appear to fall between typical compound and simple leaf categories. In the past, developmental analyses were performed on Myriophyllum with regards to the regions of lobe production at the base of the leaf (Jeune, 1975, 1976, 1977). It was discovered that these regions act as potential growth centres for lobes in much the same way that leaves are produced from specific locations around the meristem. The purpose of the present study was to elucidate the role of KNOX during leaf development in M. aquaticum and further determine the nature of lobe growth centres at the base of developing leaf primordia and how these fit in the context of Arber's partial shoot theory (Arber, 1950), which recognized that some elements of leaves, especially their development, make it possible to conceptualize them as ‘partial shoots’, since they share some characters with shoots.

MATERIALS AND METHODS

Probe design

The following degenerate primers were designed using the computer software Clustalw from sequences of 21 knotted-1 like genes available on Genbank: GAY CCN GAR YTN GAY CAR TTY ATG (fwd) and NCK YTT NCK YTG RTT DAT RAA CCA RTT RTT DAT YTG (rev). These primers were used to probe for the presence of KNOX genes in DNA from the mature leaves of M. aquaticum using polymerase chain reaction (PCR). The PCR products were cloned and sent to Genome Quebec (McGill University, Montreal, Quebec, Canada) for sequence analysis. Positive clones were used as templates in conjunction with the computer software Primer3 to generate the following gene-specific primers, AGTTCATGCGAAGGATCGAA (fwd) and TGTAATGCAAGTCCCACCAA (rev). These primers were then used to probe cDNA generated from shoot tip RNA via PCR. Subsequent products were cloned into a TOP PCRII vector (Invitrogen, Burlington, Ontario, Canada), transformed into E. coli and grown on ampicillin/agar plates overnight for colony PCR analysis the following day. Positive colonies were cultured overnight in LB broth with 50 g mL⁻¹ ampicillin and used for glycerol stocks and mini/midi plasmid preparations (Qiagen, Mississauga, Ontario, Canada). Miniprep samples were then sent to Genome Quebec for sequence analysis and insert orientation (sense/antisense) determination.

Plasmids were linearized with the restriction enzyme HindIII (Invitrogen) and RNA antisense probes were transcribed from linearized DNA using T7 RNA polymerase (Roche, Mississauga, Ontario, Canada) incubated at 37 °C for 2 h. Negative control probes were produced from plasmids containing a 350-bp fragment of intergenic DNA isolated from lobster (Homarus americanus) (courtsey of Lobster Science Center, Atlantic Veterinary College, Charlottetown, Prince Edward Island, Canada). After incubation, 1 µL was run on a 1 % agar gel to assess the quantity and quality of the probe produced. Plasmid templates were digested with RQ1 DNase (Promega, Nepean, Ontario, Canada) in a reaction mix which was incubated at 37 °C for 1 h. Template integrity and probe concentration was estimated by running 1 µL on a 1 % agarose gel. According to the manufacturer (Roche Applied Science, Laval, Quebec, Canada) each transcription reaction performed with 1 g of plasmid template can be expected to produce approx. 10 g of RNA template.

RNA in situ hybridization

In situ hybridization was used as a method for localizing KNOXI gene expression within shoot tip sections. The protocol described here is derived from Kramer (2005). Myriophyllum aquaticum (Vel.) Verdc. plants were grown in Conviron growth chambers at an ambient temperature of 22 °C and a light/dark cycle of 16 h/8 h. Shoot tips were dissected from plants and fixed overnight in 4 % paraformaldehyde at 4 °C. Tissue was then equilibrated in PBS, dehydrated and stored overnight in 95 % ethanol, and then placed in 100 % ethanol at room temperature. Tissue was then transferred through 1 : 1 HemoDe (Fisher, Nepean, Ontario, Canada) and ethanol solution, and then to 100 % HemoDe solution for 30 min at room temperature and then 55 °C. Tissue and HemoDe solution was gradually infiltrated with paraplast paraffin chips for 1 week. Tissue was then embedded in paraffin moulds and stored at 4 °C.

In situ preparation

Paraffin blocks were sectioned to a thickness of 7 µm on a manual rotary microtome and sections were allowed to adhere to probe on plus slides (Fisher) overnight at 42 °C. Slides were removed from the slide warmer the next day and transitioned through the series of prehybridization solutions, as in Kramer (2005), briefly described as follows. Slides were incubated in fresh citrisolve for 20 min and then rehydrated. Slides were incubated in 2× SSC and 0·2 m HCl each for 20 min, and then rinsed in DEPC H₂O. Slides were incubated in protein kinase K for 30 min at 37 °C and then rinsed in 0·2 % glycine to stop the reaction. Slides were rinsed in 1× PBS and then incubated in 4 % PFA and triethanol amine and acetic anhydrate for 10 min each with PBS washes before and after. Slides were then rehydrated to 100 % ethanol and allowed to air dry. Once completely dry, slides were placed in a moist hybridization chamber, incubated with 200 µL of hybridization solution (without RNA probe) at 55 °C for 1 h. Approximately 300 ng of probe in 50 % formalin was added to each slide, covered in parafilm strips, and incubated at 55 °C for 18 h. Slides were then soaked in 0·2× SSC prewarmed to 55 °C for 2 h and then incubated in RNase A for 30 min at 37 °C. Slides were rinsed in NTE buffer and then incubated in 0·2× SSC for another hour. Slides were then successively blocked in Roche blocking solution and then in BSA plus Triton X-100 (Block 2) for approx. 1 h each. Anti-digoxigenin AP fab fragments were mixed in Block 2 solution to a ratio of 1 : 1250. Slides were coated with 400 µL of antibody solution and allowed to incubate in a dark, humid container for 2 h at room temperature. Slides were then washed in three consecutive 20-min washes of Block 2 solution and then incubated overnight at 4 °C.

Slide development and analysis

Slides were moved to a fresh Block 2 solution for 20 min at room temperature, and through two subsequent washes in Buffer C of 20 min each. After buffer washes were complete, Western Blue substrate (Fisher) was carefully spread over the slides with a sterile pipette tip in 200-μL aliquots. Slides and substrate were then incubated in the dark at room temperature, and monitored on an hourly basis for colour development for up to 24 h. Once slides were judged to be fully developed, they were rinsed for 2 min in TE buffer and then dehydrated in an ethanol series and soaked in citrisolve (Fisher) for 5 min. Slides were then removed from citrisolve and permanently mounted with glass coverslips using Histoprep (Fisher) as a mounting medium. Slides were then observed using a standard light microscope and pictures were captured using the accompanying digital imaging program.

Scanning electron microscope tissue preparation and observation

Shoot tips were quickly dissected from plants and placed in ice-cold FAA fixative overnight. Tissue was then washed twice in 50 % ethanol for 30 min each, and then sequentially moved to 70 % and 80 % ethanol washes at room temperature for 1 h each, and finally stained overnight in a 0·5 % alcoholic solution of acid fuchsin. The next day, tissue was washed twice in 100 % ethanol for 1 h each, and dissected in 100 % ethanol under a stereoscope until the desired level of leaf development was revealed. Tissue was then placed in microporous specimen capsules (Marivac, Lakefield, Quebec, Canada) and dried using a Ladd model #28000 critical-point dryer. Specimens were then mounted on pin type aluminum specimen stubs using adhesive tabs and silver paint for grounding, and then coated with 600 Å of gold–palladium using a Denton vacuum desk II sputter-coater. Specimens were viewed using a Cambridge Stereoscan 604 scanning electron microscope equipped with a SEMICAPS digital imaging system.

RESULTS

KNOX gene isolation and sequence analysis

A nucleotide fragment of approx. 300 bp was cloned using gene-specific primers designed from a fragment of Myriophyllum aquaticum DNA previously isolated using degenerate primers. Once sequenced, the fragment was blasted on Genbank and was revealed to have up to 80 % nucleotide similarity to KNOXI genes found in other species of plants. When translated into an amino acid sequence and aligned with other KNOXI proteins, the M. aquaticum putative KNOX gene exhibited the presence of highly conserved domains typical of all KNOX genes. When proceeding from the N terminus to the C terminus, the putative protein sequence codes for eight amino acids of the Meinox domain, 24 non-conserved amino acids, the GSE and ELK domains, and 24 amino acids corresponding to the beginning of the Homeobox (Fig. 5). The M. aquaticum gene fragment was dubbed MaKN1 and placed on Genbank under the accession number EU203679.

Fig. 5.

Protein alignments of MaKN1 and ten other KNOX homologues. Percentage similarities of sequences are show in red and blue which correspond to >90 % and >50 %, respectively. Black corresponds to <50 % similarity between amino acid residues. Sequence similarities can be seen which correspond to the Meinox, GSE, ELK and homeobox domains. Arrowheads correspond to lines containing MaKN1 putative protein.

Scanning electron microscopy: shoot tip morphology

Figure 1 shows scanning electron micrographs depicting the three-dimensional morphology of the shoot tip. Leaf primordia appear in pentamerous alternating whorls which originate from the flank of the meristem (Fig. 1A, B), although the number of primordia produced per whorl actually may vary between five and six. At the P1 stage, primordia appear as simple bulges in the peripheral zone of the meristem (Fig. 1A). At stage P2, primordia have expanded into a conical leaf buttress lacking any lobes (Fig. 1C). The first lobes appear at plastochrone 3, and are initiated at two well-defined regions at a relatively fixed location near the base of the leaf (Fig. 1D). By plastochrone 4 (Fig. 1E), primordia have developed several lobes in a basipetal fashion where proximal lobes are younger (more recently developed) than those towards the distal portion of the leaf. It is also apparent at this stage that the lobes in each lobe pair develop in an alternating rhythm with respect to each other. This phenomenon becomes more apparent in older primordia (Fig. 1F) where lobe insertions are not at the same level on the leaf axis. Throughout leaf development, lobes retain their initial orientation and symmetry, flattened in the same plane as the main axis of the leaf primordium (Fig. 1F) and do not become concave as leaflets do in compound leaves. Between plastochrones 5 and 6, small dome-like protuberances which will eventually become trichomes, appear in the axils of lobes on the abaxial side of the leaf (Fig. 1G). By plastochrone 7 almost all the lobes of the leaf have been formed (Fig. 1H). At P7, cellular differentiation becomes evident at the tip of the leaf as bulb-like cells form on the epidermis of the distal-most lobes (Fig. 1I). This differentiation progresses towards the base of the leaf in subsequent primordia until the whole surface of the leaf is covered in spherical cells (used for trapping air around the leaf).

Fig. 1.

Scanning electron micrographs of M. aquaticum shoot tips. Primordia are represented by ‘P’ followed by their plastochrone number. (A) and (B) Frontal and overhead views of the meristem (M) and primordia (P). (C) Close-up view of P2 primordium. (D) Close-up view of P3 primordium; arrowheads indicate location of lobe-generating centres. (E) P4 primordia; arrowheads indicate alternating lobe development. (F) Close-up view of lobes of a P6 primordium; arrowheads indicate alternating lobe insertion. (G) P5 and P6 primordia showing trichome development in lobe axils (arrowheads). (H) P7 primordia showing complete lobe morphogenesis. (I) Close-up view of P7 leaf tip; arrowhead indicates epidermal differentiation. Scale bars: (A, B, E–G) = 30 µm; (C, D) = 15 µm; (H) = 150 µm; (I) = 75 µm.

Light microscopy: shoot tip anatomy

In median section the meristem appears dome-like and is approx. 100 µm in diameter at the peripheral zone. Provascular strands (PV) in the developing stem are apparent by plastochrone 3 and become flanked by large empty spaces of aerenchyma (Ar) in more mature stem tissue (Fig. 2A). Leaf primordia may appear as either long unbroken protuberances corresponding to the central midrib portion of the leaf, or as a fragmented series of circular sections corresponding to cross-sections through leaf lobes (Fig. 2A). In cross-section, the meristem appears as a central circle of undifferentiated cells surrounded by alternating whorls of leaf tips corresponding to primordia older than plastochrone 3. Cross-sections of leaf primordia appear as a central midrib flanked on either side by sections through progressively younger lobe primordia (Fig. 2D). Lobe sections that are closest to the mid rib are from more distal (older) lobes and are closer to the base of each lobe. Lobe sections farther away from the midrib are from more proximal (younger) lobes and are closer to the tip of each lobe. Figure 2C shows a longitudinal section of a P7 leaf primordium to highlight lobe morphology and provide a visual comparison for upcoming in situ sections.

Fig. 2.

Histological sections of M. aquaticum shoot tips stained with toluidine blue (A, C, D) and negative control RNA in situ hybridization (B). (A) Longitudinal section through meristem (M), leaf midribs (MR) and lobes (Lb). Provascular strands (PV) and aerenchyma (Ar) are evident in more mature stem tissue. (B) Negative in situ hybridization control section using lobster gene template as a probe. Hybridization signal is absent from both lobes (arrows) and meristem (M). (C) Paradermal section through a leaf primordium (P7). Veins are depicted by (V) and lobes by (Lb). (D) Cross-section through the meristem. Sequential plastochrones are represented by corresponding numbers, leaf lobes by an asterisk and apical meristem by ‘M’. Scale bars: (A, B) = 200 µm; (C) = 50 µm; (D) = 100 µm.

Several types of negative controls were used during in situ experiments to ensure specificity of probe hybridization. Sense probes were initially used as a negative control during hybridization experiments; however, they proved to hybridize in much the same pattern as antisense probes. To determine if sense probe binding was due to non-specific binding of digoxigenin antibodies, no probe control slides were run alongside antisense and sense control slides. No probe slides failed to develop any hybridization signal, indicating that signal from sense negative controls arose during probe hybridization. Other sources have reported similar difficulties with sense probe hybridization due to faulty T7 polymerase binding during transcription (Müller et al., 2001; Tioni et al., 2003). To determine if this was the case during this particular experiment, a 350-bp probe from a lobster gene fragment was used as an alternate negative control. When hybridized under the same conditions as antisense probes, the lobster probe failed to show any hybridization signal (Fig. 2B) indicating that observed signal specifically represents KNOXI expression.

KNOX in situ localization: longitudinal sections

In situ hybridization was repeated a total of five times with five shoot apices per experiment. Results were found to be similar for all apices examined. Figure 3 shows shoot tips of M. aquaticum in longitudinal sections probed for KNOX gene expression using RNA in situ localization. Sites of antisense probe hybridization are observed as a blue stain. As sections progress through the meristem (Fig. 3A–C), overall patterns of putative KNOX expression can be observed. KNOXI expression was observed evenly throughout the meristem (Fig. 3A–C, E). Down-regulation of KNOXI expression at the site of P0 was not observed in this study (Fig. 3A–C, E). Provascular strands exhibit maKN1 expression in tissue subtending leaf primordia (Fig. 3D, arrowhead) and leaf traces between primordia P1 and P8 (Fig. 3B, C). Leaf-trace vascular tissue appears to extend from the site of leaf insertion, through internodal tissue, to connect with vasculature of the next lower vascular ring (Fig. 3B, arrowhead). Plastochrones 1–3 (P1 to P3) exhibit probe hybridization evenly throughout their respective primordia (Fig. 3A–C). The spatial pattern of gene expression changes as development proceeds. Expression at plastochrone 4 (P4) appears to be located primarily in the tip and on the adaxial side of the midrib (Fig. 3A–C). As sections progress towards the outer lobed margins of the primordia, expression becomes more intensely and evenly expressed throughout the tissue. By plastochrones 5 and 6, hybridization signal is no longer present in the central midrib region of the leaf and appears primarily in the lateral lobes. In Fig. 3A, regions of the P5 lobes can be seen appearing in the section and exhibit intense hybridization signal. In Fig. 3B, P5 tissues appear to exhibit less hybridization signal. As sections pass through the marginal lobes and into the central portion of the primordia, signal fades until it is completely absent (Fig. 3C).

Fig. 3.

Serial longitudinal sections through a single shoot apical meristem (M) exhibiting putative KNOX gene expression (blue stain) using antisense probes. Primordia are represented by ‘P’ followed by their respective plastochrone number. Leaf-trace vascular tissue appears to extend from the site of leaf insertion, through internodal tissue, to connect with vasculature of the next lower vascular ring (Fig. 3B, arrowhead). Provascular strands exhibit maKN1 expression in tissue subtending leaf primordia (Fig. 3D, arrowhead); hybridization signal is more strongly expressed in the tips of the younger lobes at the base of the leaf (arrows). Scale bars: (A–C) = 200 µm; (D) = 500 µm; (E) = 50 µm.

Expression in plastochrones higher than P6 is more readily observed in sections through the marginal region of the shoot tip (Fig. 3D) where the outer-most region of the stem is in section. In the central region of the P7 primordia, very little expression is evident and it is apparent that hybridization signal is absent from older lobes at the tip of the leaf and is more strongly expressed in the tips of the younger lobes at the base of the leaf (arrows) (Fig. 3D).

KNOX in situ localization: cross-sections

Serial cross-sections offer a complementary perspective of KNOXI expression in the shoot tip of Myriophyllum aquaticum. In Fig. 4A, sections begin at the P8 node until the apical dome of the meristem is reached (Fig. 4D). By examining serial sections of the mature stem, it can be seen that gene expression varies between nodal and internodal regions of the stem. At the P8 node in Fig. 4A, there appears to be very little hybridization signal in either the lower lobes of P9 or at the leaf bases of P8. As sections proceed into the P7 internode in Fig. 4B and C; however, expression is evident in the outer regions of the stem as well as in the provascular cylinder (Fig. 4B, arrowhead). At this point, expression is also apparent in the lobes of the P8 primordia. Expression remains absent in all sections of P9 primordia. In Fig. 4C, the location of the expression pattern in the outer tissues of the stem becomes isolated to pockets of tissue between the leaf bases and in the lobes of P7 primordia. At this point, lobes of P8 still exhibit expression; however, as serial sections proceed up the primordia, expression becomes weaker until it is completely absent at the level of the nodal region of plastochrone 6 (Fig. 4C).

Fig. 4.

Serial cross-sections beginning at plastochrone 8 (A) proceeding towards the shoot apical meristem (D). Primordia are represented by their respective plastochrone number. Expression is evident in the outer regions of the stem as well as in the provascular cylinder (Fig. 4B, arrowhead). Scale bars = 200 µm.

In sections close to the meristem (Fig. 4D), the difference in pattern of expression between nodes and internodes is less obvious as the hybridization signal becomes more evenly expressed in the developing pith and equally distributed throughout the apical dome. Cross-sections (e.g. Fig. 3C) show similar patterns of expression in leaves as seen in longitudinal sections where plastochrones 1 and 3 exhibit hybridization signal evenly throughout the primordia and plastochrones P4 or older show signal primarily in lobes.

DISCUSSION

Even though mature plant organs may be classed in one of the three fundamental organ categories (leaf, stem and root), this may overlook developmental characteristics that could qualify the organ for placement within a different category. In other words, similar mature forms do not necessarily arise from the same developmental pathway (Wilkins, 2002; Jaramillo and Kramer, 2007). This inherent ambiguity in the way homology is conceptualized underlines the importance of exploring development in non-model species and establishing homology at multiple biological (molecular, structural and functional) levels (Rutishauser and Moline, 2005; Kirchoff et al., 2008). In this study, KNOXI gene expression was consistently found in the meristem of all M. aquaticum shoot tips examined. This pattern is typical of KNOXI expression in other species as well (Jackson et al., 1994; Scofield and Murray, 2006). The fidelity of this pattern of localization led to the initial suggestion that KNOX genes such as STM could be used as molecular markers for determining meristem identity during plant development (Hake, 1996; Hirayama et al., 2007). However, as shown in this study and many others, KNOXI expression is as much in evidence in developing leaves as it is in the meristem (Janssen et al., 1998b; Bharathan et al., 2002; Harrison et al., 2005b). In M. aquaticum, KNOXI expression was not found to be down-regulated at the site of the incipient leaf primordia (P0), a phenomenon which seems to be unique to compound-leaf development (Chen et al., 1997; Reiser et al., 2000). MaKN1 expression was also observed in the developing vasculature of the stem, subtending the site of leaf insertion, and in leaf trace primordia. This pattern of expression has been observed in numerous other studies, the function of which remains unclear (Smith et al., 1992; Jackson et al., 1994; Jannson et al., 1998a). It has been hypothesized that the expression in the peripheral developing stem tissue subtending leaf primordia is derived from meristematic tissue as opposed to the more internal stem tissue which originates from the rib zone (Jackson et al., 1994). KNOXI expression during compound leaf and shoot development indicates that there are shoot-like processes at work during the early stages of development in compound leaves (Sinha, 1997; Hofer et al., 2001a; Lacroix et al., 2005; Brand et al., 2007; Hirayama et al., 2007). It may therefore be more accurate to suggest that, rather than the physical structure of the meristem, KNOX genes could represent a function of indeterminate growth acting within both the meristem and compound leaf primordia (Janssen et al., 1998a; Hay and Tsiantis, 2006; Brand et al., 2007).

Although they have been identified in many different species, there are no recorded data on KNOXI gene expression in the order Saxifragales previous to this study. Degenerate primers were designed to encompass part of the homeodomain and a portion of the ELK domain resulting in a fragment of approximately 300 bp that was found to have 80 % sequence similarity with other KNOXI genes recorded in the online NCBI database, Genbank. A single species of plant may have multiple KNOXI homologues which are expressed during plant development. Arabidopsis thaliana has four KNOXI homologues in its genome while Oryza sativa has six (Jouannic et al., 2007). Although they all appear to share functional redundancy, they exhibit specific regions of expression within the shoot tip (Hareven et al., 1996; Reiser et al., 2000; Groot et al., 2005). The specific regions of gene expression are typified by the expression domains seen in the two Arabidopsis KNOX genes STM and KNAT1, and LeT6 which is the KNAT1 homologue in Lycopersicum esculentum (Reiser et al., 2000). When examining the patterns of expression of KNOX in M. aquaticum, it is evident that it appears to exhibit patterns reminiscent of different types of expression. There is expression present at the meristem, the developing stem and developing leaf primordia (Figs 3 and 4). These observations indicate that the M. aquaticum putative KNOX probe could be binding to the RNA of more than one KNOXI gene. This is entirely possible when considering the region of the KNOX gene the RNA probe was cloned from. The RNA probe includes the ELK domain and a region of the homeodomain, both of which are heavily conserved between KNOX homologues. To distinguish between KNOX homologues within M. aquaticum, a region of the gene should be cloned that is a less conserved, such as the region upstream of the MEINOX domain. This would avoid non-specific binding between KNOX homologues and allow for the determination of specific patterns of expression for individual genes.

KNOXI gene expression has been studied in an evolutionarily diverse array of species including Acetabularia (green algae), monocots such as Zea mays, Ruscus aculeata (Asparagaceae) and Elaeis guineensis (palms), and many dicot species such as Lycopersicon esculentum, Arabidopsis thaliana and now Myriophyllum aquaticum. By comparing KNOXI expression between species such as these, it has become apparent that the function of KNOXI genes in maintaining indeterminate growth and its correlation with increasing morphological complexity has been well conserved (Bharathan et al., 2002; Tioni et al., 2003; Harrison et al., 2005b). For the most part, KNOX expression is localized to the shoot apical meristem, and is either down-regulated at P0 during simple leaf development or is up-regulated again at P1 in primordia destined to be compound (Scofield and Murray, 2006). The partial shoot theory as proposed by Arber (1950) and described as a morphological continuum by Sattler (1986), suggests that rather than being morphologically mutually exclusive, shoots and leaves develop via a collection of developmental processes which posses the potential to be expressed in either homologous (homotopy) or non-homologous organs (homocrasy). The overlap of KNOXI expression exhibited between leaves and shoots does not necessarily equate to evolutionary homology, but it does indicate that structural relationships between organs are not as straightforward as their mature morphologies might at first indicate. It is clear that the elaboration of plant form requires the expression of KNOXI (Kim et al., 2003; Jasinski et al., 2007). This can be seen in LeT6 overexpression mutants of tomato exhibiting super-compound leaves as well as in expression studies in Arabidopsis thaliana and its close, compound -leaved relative Cardamine hirsuta (Hareven et al., 1996; Hay et al., 2006). Wherever KNOXI is expressed dissection occurs, and once it is suppressed, differentiation develops regardless of whether this occurs in the leaf or an apical meristem. This indicates that the appearance of dissection and KNOXI expression, as in the leaves of M. aquaticum, is correlated with indeterminacy, whether transient or continuous.

Myriophyllum aquaticum produces lobes from specific locations along the leaf axis in a manner that is reminiscent of leaflet development in compound leaves (Lacroix et al., 2003; Jeune and Lacroix, 2009). The lobes of M. aquaticum are produced basipetally and appear to originate from two very distinct centres at the base of the leaf in an alternating fashion. Although tissue sections can only represent static snapshots of the dynamic process that is plant development, the expression of KNOXI exhibited at each plastochrone may be extrapolated to be representative of the changes in gene expression that each leaf undergoes during its entire development. This is further supported by Jeune and Lacroix (2009) where leaf developmental parameters were analysed from 40 shoots over the span of 20 plastochrones. Leaves followed an exponential increase in size from one plastochrone to the next, indicating that leaf development is relatively constant between plastochrones. It can therefore be seen in Myriophyllum that a single leaf primordium expresses KNOXI up until it reaches plastochrone 9. After this point, all the lobes of the leaf primordium are in place, signalling the end of the morphogenetic phase of leaf development (Jeune and Lacroix, 2009) and transient indeterminacy of the leaf lobes ceases. During this period of time, the overall pattern of expression of KNOXI appears to change as primordia develop. When the primordium is a simple buttress and has yet to initiate lobes, KNOXI is expressed ubiquitously throughout the primordium. With the onset of lobe development at plastochrone 3, KNOXI disappears from the central portion of the leaf and becomes localized to lobes between plastochrones 4 and 8. This pattern is reminiscent of KNOXI expression as observed in the leaves of other dissected leaf species (Hareven et al., 1996; Bharathan et al., 2002).

These basal leaf centres appear to have organogenic potential as noted in the past by Jeune (1975, 1976, 1977). The growth centres were further characterized recently in a study by Jeune and Lacroix (2009) in a quantitative analysis of leaf development in M. aquaticum. It was found that, in 40 shoot tips, the growth centres and subsequent lobe development were highly localized at a region of the leaf primordia approximately one lobe width from the base of the leaf. These studies indicate that the alternating production of lobes could be thought of as reminiscent of the production of leaf primordia on a distichous shoot where each nodal region contains a single leaf and successive leaves are arranged 180° from each other (Jean, 1994). In other words, the same processes that result in distichous leaf patterning at the shoot apical meristem could also be functioning during lobe development. Further analysis into these shoot-like qualities of M. aquaticum leaves could include exploring the developmental processes that turn growth centres on and off, allowing them to produce lobes in a distichous fashion. However, in conjunction with the current analysis on KNOXI expression, a different picture emerges. KNOXI expression correlates with the acquisition of indeterminate growth. If this is so, then the presence of KNOXI expression in lobes makes them more reminiscent of transient meristems which fail to develop lateral elements. It is possible that if KNOXI expression was prolonged or overexpressed in leaf lobes, the lobes themselves would develop lateral elements or an increased complexity as occurred in such studies of tomato.

One possible study would be to explore the effects of growth hormones, such as auxin, on the effects of the meristematic centres at the base of the leaf. Where auxin has been studied in model species, it has been found to be closely tied to controlling the position of lateral organs around the shoot (Reinhardt et al., 2003; Quint and Gray, 2006). In the compound-leaved relative of Arabidopsis, Cardamine hirsuta, it has been found that KNOXI will actually co-accumulate with auxin at the site of growth foci to promote leaflet production, which further illuminates differences in KNOXI expression between simple and compound leaves (Barkoulas et al., 2008). If this process is repeated in M. aquaticum, it would be interesting to explore the growth centres at the base of the leaf primordia with regards to the correlation between auxin accumulation and the pattern of lobe production. Another area of future research would be to explore the organogenetic capacity of the lobe-producing centres at the base of the leaf. If the two regions of lobe production act as ephemeral meristematic centres for lobe production, it might be expected that if one or both of these centres were destroyed it would affect lobe development. This experiment was previously attempted on a rudimentary level by mechanically destroying the centres with fine needles (Jeune, 1976). A more-refined experiment using microsurgical laser ablation may produce more reliable results (Reinhardt et al., 2004).

The fact that KNOXI is expressed in the meristem, early leaf primordia, and then later in lobes is suggestive of a shared developmental process between these three structures (Sinha et al., 1993; Hofer et al., 2001b). The partial shoot theory would describe this phenomenon as representative of a morphological continuum present between all plant structures (Sattler and Rutishauser, 1992). Shared developmental processes, however, do not necessarily indicate that dissected leaves and stems are partially homologous. Leaves and stems may have originated as mutually exclusive organ categories that have, over the course of evolutionary history, undergone convergent evolution to become more similar with respect to KNOXI expression. It has been shown that the ancestral angiosperm had a simple leaf morphology and that dissected leaf morphology was derived secondarily an average of 29 times (Bharathan et al., 2002). This suggests the possibility that different plant species may have evolved compound leaves via different molecular mechanisms, indicating that homology of development could be ambiguous (Boyce, 2010). Therefore perhaps some plants evolved complex leaves with shoot like characterisitcs but not necessarily all. From a dynamic point of view, however, consider M. aquaticum as an entire structure rather than a compilation of non-homologous organs. With this perspective in mind, it becomes evident that in M. aquaticum, KNOXI expression is localized to terminal tips of the plant structure (i.e. meristem, young leaf primordia and lobes) and appears to be less strongly expressed in more developed regions. All three of these structures are developmentally reminiscent of each other in that each initially develops as an undifferentiated, more-or-less radial projection where KNOXI is likely to be promoting indeterminate growth to varying degrees. Therefore, if one were to forget categorical names such as shoot and leaf, one might suggest that the above three structures represent repetitions of the same basic process at different developmental levels (Arber, 1950; Lacroix et al., 2005). This, of course, is not the entire picture because there are many genes and developmental processes active during leaf development and differentiation that are not present during shoot development which ultimately results in different mature structures (Tsiantis and Hay, 2003; Shani et al., 2006). However, KNOXI is expressed during the initial stages of organogenesis before differentiation has occurred in the leaf (before plastochrone 7), which suggests that, although leaves and shoots have different mature forms, their early developmental stages are more closely related. The presence of KNOXI genes during early stages of leaf development serves to illustrate a morphological continuum between previously supposed non-homologous structures (Sattler and Rutishauser, 1996; Hofer et al., 2001b).