1

American Journal of Botany 98(8): 1–10. 2011.

American Journal of Botany 98(8): 1–10, 2011; http://www.amjbot.org/ © 2011 Botanical Society of America

Although plants are often thought of as completely sessile organisms, many aspects of plant life do involve movement ( Simons, 1992 ). For example, circumnutation ( Simons, 1992 ), closing and opening of stomata, righting of tree branches ( Bowling and Vaughn, 2008 ), and coiling of tendrils ( Meloche et al., 2007 ; Vaughn and Bowling, 2009 ) involve true movement, although even those motions are relatively slow compared to many movement responses in zoological systems. Exceptions to this slower movement are the very rapid release of pollen from bunchberry, the closing of Venus fl ytrap leaves and the ballistic dispersal of seeds from capsules such as in Impatiens , Hura , and other species ( Simons, 1992 ). In these cases, the movement is so rapid (0.1 s) that it is diffi cult to capture these phenomena accurately even with high-speed digital cameras. Movement of anthers of white mulberry has been documented as the fastest biological movement on the planet: half the speed

of sound ( Taylor et al., 2006 ). Recently, Skotheim and Mahadevan (2005) described a framework of the scales and speeds of movement that are possible in biological movements. These authors categorized plant movement into three categories: swelling/shrinking, snap buckling, and explosive fracture, with the latter two responsible for the most rapid plant movements. Although this theory explains and classifi es some of the mechanics of these movement phenomena, how the plant is able to produce cell types and structures that can accomplish this rapid movement is still largely unknown.

Ballistic seed dispersal is one of the more effective ways in which weed seeds may be distributed compared to the more passive methods used by many species ( van der Pijl, 1972 ; Benvenuti, 2007 ). For example, jewelweed ( Impatiens capen-sis ) may eject seeds as far as 2.0 m from the mother plant, en-suring that its propagules will be spread over a wider area and/or lessen competition between seedling and the parent plant ( Hayashi et al., 2009 ). In some species, such as bittercress, the ballistic dispersal of seeds serves as an herbivory defense mechanism as well ( Yano, 1997 ). As a potential herbivore en-counters a capsule, the touch of the herbivore sets off the dis-persal mechanism, springing the insect herbivores far away. Thus, ballistic seed dispersal not only serves as an effective way in which to disperse seed but also to protect them from would-be herbivores.

In this study, siliques of the ballistic species bittercress (also known as hairy bittercress, the common name approved by the Weed Science Society of America) are compared to the much more passive dispersal mechanisms in the siliques of the related and well-studied Arabidopsis siliques ( Spence et al., 1996 ). Se-lection of mutant silique phenotypes in Arabidopsis has allowed workers to demonstrate the importance of each of the layers in the silique walls in the ability of the silique to open and disperse

1 Manuscript received: 24 September 2010; revision accepted: 9 May 2011. A.J.B. was supported in part by the ARS Research Associate program to

K.C.V. Ted Whitwell generously supplied bittercress seed, and Mark Czarnota supplied photos of bittercress plants in weedy situations. Timothy Huggins provided excellent technical assistance. Production of the CCRC series monoclonal antibodies was supported by NSF grants DBI-0421683 and RCN-0090281. Mention of a trademark, vendor or proprietary product does not constitute an endorsement by USDA. Thanks are extended to several researchers at the Plant Cell Wall Gordon Conference, who provided excellent suggestions as to experiments to perform and suggestions for interpretation.

4 Present address: 4358 River Road South, Salem OR 987302 USA 5 Author for correspondence (e-mail: AJBowling@dow.com); present

address: Dow Agrosciences, 9330 Zionsville Rd., Indianapolis, Indiana 46268 USA

doi:10.3732/ajb.1000374

THE MECHANISM FOR EXPLOSIVE SEED DISPERSAL IN CARDAMINE HIRSUTA (BRASSICACEAE) 1

Kevin C. Vaughn 2,4 , Andrew J. Bowling 2,5 , and Katia J. Ruel 3

2 Crop Production Systems Research Unit, USDA-ARS, P. O. Box 350, Stoneville, Mississippi 38776 USA; and 3 Research Center on Plant Macromolecules, CERMAV-CNRS-UPR 5301, BP 41 38041 Grenoble, France

• Premise: Although many highly successful weed species use a ballistic seed dispersal mechanism, little is known about the mechanics of this process. Bittercress ( Cardamine hirsuta ) siliques are morphologically similar to Arabidopsis siliques, but they can project their seeds up to 5 m, while Arabidopsis seeds are dispersed by gravity. Comparison of these species should enable us to determine which structures might be responsible for ballistic seed dispersal.

• Methods: Sections of Arabidopsis and bittercress siliques were immunolabeled with antibodies raised against a variety of polysaccharide epitopes.

• Results: In bittercress, the second endocarp layer (enB) of the valve had strongly asymmetrical cell wall thickenings, whereas the analogous cells in Arabidopsis were reinforced symmetrically and to a lesser extent. Additionally, an accumulation of mucilaginous pectins was found between the fi rst and second endocarp (enA and enB) layers in the bittercress valve that was not present in Arabidopsis . However, in both species, highly de-esterifi ed homogalacturonan was lost in the dehiscence zone (at the carpel/replum interface) as the siliques matured, thus allowing for separation of the valve at maturity.

• Conclusions: Ballistic seed dispersal in bittercress may involve the contraction of the outer pericarp tissue against the highly asymmetrically thickened enB cells, which are hypothesized to bend in one direction preferentially. The stress generated by the differential drying of the inner and outer layers of the valve is released suddenly as the adhesion between the cells of the dehis-cence zone is lost, leading to a rapid coiling of the valve and dispersal of the seeds.

Key words: Arabidopsis ; ballistic seed dispersal; bittercress; Brassicaceae; dehiscence ; immunocytochemistry; silique.

GALLEY PROOF

2 American Journal of Botany [Vol. 98

in PBS, the primary rabbit antibodies were diluted 1 : 40, and the secondary antibody was replaced with protein A-gold conjugated to 15-nm colloidal gold (EY Laboratories) and diluted 1 : 20 in PBS-BSA. Controls included a no pri-mary antibody control and a boiled primary antibody.

Immunogold-transmission electron microscopy — Sections with pale gold-silver refl ectance (~100 nm) from the same block faces used for light micro-scopic immunocytochemistry were obtained with a diamond knife on a Reichert Ultracut ultramicrotome and were collected on gold 300 mesh grids. The grids were then fl oated on the following solutions and times: PBS-BSA, 30 min; primary antibody, either neat or diluted 1 : 8 in PBS-BSA, 3 h; PBS-BSA, 4 × 2.5 min on each drop; secondary antibody coupled to 15 nm gold (EY Labora-tories, San Mateo, California, USA) diluted 1 : 20 in PBS-BSA, 1 h. Grids were then washed extensively in distilled water and poststained in uranyl acetate (2 min) and lead citrate (30 s) before observation with a Zeiss EM10CR elec-tron microscope operating at 60 kV. Negatives were scanned with an Epson 700 scanner, and the images were reversed in Photoshop (Adobe, San Jose, California, USA).

RESULTS

Whole plant descriptions — Plants of bittercress develop quickly after germination and after ~30 d of growth begin to produce fl owers and young siliques, with up to 147 siliques/plant under our conditions. Approximately 10 d after the siliques have started to develop (directly after fl owers have shed), they reach a stage where the silique can be triggered to disperse their seeds explosively by touching the silique, even though the seeds are not quite mature at the earliest stages of this ability. Nearly mature siliques, while being examined with a dissecting micro-scope, often exploded as the light dried the silique ever so slightly, sometimes with just the side toward the light of the dissecting scope dehiscing alone ( Fig. 1A ). The siliques that explode naturally seem to do this in two different modes. In some, carpels remain attached at the top of the remnant stylar material and coil like a rolltop desk back upon the remnant sty-lar material ( Fig. 1B ), whereas other siliques detach completely, leaving a naked stylar remnant ( Fig. 1C ). These carpels that detach from the replum coil more loosely, similar to a vine. Despite the outward similarity of the siliques of bittercress and Arabidopsis , the bittercress siliques open explosively, resulting in a carpel that is highly coiled after ballistic seed release. In contrast, the carpels of the Arabidopsis silique remain in an un-coiled position even after the siliques dehisce and after the seeds have been completely dispersed. This only allows for relatively passive seed dispersal in Arabidopsis compared to the bittercress.

Anatomy of the siliques — The Arabidopsis silique has been well described in the literature, and the morphology of the bit-tercress silique is very similar prior to dehiscence ( Fig. 2 ). When observed in cross section, the two fused carpels form a cavity (locule) in which the seeds reside. The carpels are linked along the central structure called a replum. At either side of the area where the replum is attached to the carpels, a zone of dehiscence occurs where layers of smaller cells occur at the site of silique opening; these are associated with highly thickened cells at either side of the replum ( Fig. 2B ). In these characteristics of the overall silique organization and dehiscence zone, the siliques of Arabi-dopsis and bittercress are virtually identical, with the exception of the overall larger dimensions of the bittercress silique.

In the carpels, two internal layers are differentiated from the others. The enB (second endocarp) layer is a layer of highly thickened cells. In the case of Arabidopsis , the thickenings are

the seeds ( Lilegren et al., 2004 ; Mitsuda and Ohme-Takagi, 2008 ; Ogawa et al., 2009 ). Because the siliques of bittercress and Arabidopsis are so similar in every other respect, we were able to draw conclusions as to the nature of cell wall changes that occur during silique development, especially in those char-acters that were unique to bittercress in allowing for explosive seed dispersal.

MATERIALS AND METHODS

Plant material — Seeds of bittercress ( Cardamine hirsuta L., a generous gift of Ted Whitwell, Horticulture Department, Clemson University, Clemson, South Carolina) were germinated in a soilless potting mix consisting of 3 parts ground pine bark, 3 parts peat and 1 part perlite in a Conviron Growth Chamber (Winnipeg, Manitoba, Canada) under constant illumination. Seedlings were watered with a dilute fertilizer solution. Seeds of Arabidopsis thaliana Colum-bia strain were obtained originally from Lehle Seeds (Tuscon, Arizona, USA) and were germinated and grown under identical conditions. Siliques of Arabi-dopsis plants were collected from 9 to 15 d after fl owering, which represent a range of development from relatively mature siliques to siliques with seed ready to be dehisced ( Louvet et al., 2006 ). On bittercress plants, three types of silique were collected: those that could not be stimulated to touch open and oc-curred just above those that could on the infl orescence, those that could be touch-stimulated to open, and valves from siliques that had naturally opened. These represented similar stages to those observed in the 9 – 15 d fl owering stages of Arabidopsis .

Microscopy — Capsules were immersed in 3% (v/v) glutaraldehyde in 0.05 mol/L PIPES buffer (pH 7.4), and pieces of the siliques were excised by cutting with a razor blade in a drop of the fi xative on dental wax. The samples were transferred to 20 mL scintillation vials of the same glutaraldehyde solution, and fi xation continued for 2 h at room temperature. After this, the samples were washed twice in PIPES buffer at 4 ° C and dehydration with ethanol, at 25% in-crements, was commenced. After a fi nal change of 100% ethanol, the samples were transferred to a − 20 ° C freezer. Addition of LR White (Polysciences, War-rington, Pennsylvania, USA) resin occurred at 25% increments over the next 4 d with a fi nal exchange of 100% resin also at − 20 ° C for 1 d. After this incubation, the vials were rocked on a shaker for 24 h at room temperature before heat po-lymerization, as described by Bowling and Vaughn (2008) . This protocol rapidly polymerizes the LR White resin and greatly improves antigenicity of the sec-tions. Sections were mounted on acrylic stubs to cut cross sections of the silique and were sectioned with a Reichert Ultra-Cut ultramicrotome using a Delaware HistoKnife for light microscopic (0.35 µ m) sections or with a Delaware ultrami-crotome knife for thin (~100 nm) sections. Light microscopic sections were stained with 1% (w/v) toluidine blue in 1% (w/v) sodium borate for ~30 s.

Immunocytochemistry — Immunogold-silver light microscopy — Groups of sections were transferred from the HistoKnife boat with a wire loop to chrome-alum coated slides to which a circle (radius ~1 cm) from a wax pencil was drawn. The sections were dried on a slide warmer and transferred to a slide incubation chamber, and a drop of 1% (w/v) bovine serum albumin (BSA) in 0.02 mol/L phosphate-buffered saline (pH 7.2; PBS) was applied to the slide for 30 min to block nonspecifi c binding sites. After the blocking stage was over, the solution was decanted and 100 µ L of antibody either undiluted or diluted 1 : 8 in PBS-BSA was added to the ring created by the wax pencil to each slide, and incubation was carried out for 3 h. After the incubation, three exchanges of PBS-BSA were given to each slide, and 100 µ L of either goat anti-rat or goat anti-mouse IgG coupled to 15-nm gold diluted 1 : 20 in PBS-BSA was added to each slide and the sections allowed to incubate for 1 h. After that incubation, the slides were washed with a stream of distilled water from a squirt bottle to re-move excess chlorides that would interfere with the silver intensifi cation steps. Either GE BioSciences or Pella Silver intensifi cation kits were used, with an equal mixture of the A & B components, with reactions occurring for up to 16 min. Control experiments included no primary antibody or a boiled primary antibody that was inactivated. Images were obtained with a Zeiss Axioskop microscope using a QColor3 digital camera and QCapture Pro software. Images were back-ground fl attened and contrast and brightness adjusted with the program ImageJ. Panels were assembled and labeled with the GIMP.

For anti-lignin localizations, the protocols for the polysaccharide monoclo-nals (above) were followed excepting that the blocking step was 3% (w/v) BSA

GALLEY PROOF

3August 2011] Vaughn et al. — Explosive seed dispersal in bittercress

always occur on the locule side of the wall and the greatest thickening toward the locule as well. The enA cells are the cells innermost in the carpel and are parenchymatous. In the case of the Arabidopsis silique, the enA layer degenerates before silique

not as great as that in the bittercress, and they are distributed evenly all around the cell. In the bittercress enB layer, the thicken-ings are in an inverted V or coupled teardrop shape in thin sec-tion with thickenings arranged in pairs ( Fig. 2C ). The thickenings

Fig. 1. Characteristics of the siliques of bittercress. (A) Nearly mature silique that was removed from the plant and then placed under the dissecting scope. The light from the dissecting scope has dried one side of the silique, causing an explosive dispersal and coiling of the carpel wall. (B) Plants on which the carpel has remained attached to the stylar remnants, which have coiled back onto themselves, like a rolltop desk (arrows). (C) Plants of bittercress on which the carpels have completely separated from the stylar material (replum) and the seeds have been well dispersed. The pale stylar remnants (arrow-heads) are all that remains attached to the plant. (D) Closer view of a highly coiled carpel wall. Scale bars = 1 mm.

GALLEY PROOF

4 American Journal of Botany [Vol. 98

Immunocytochemistry — To determine the possible reasons for the differences in silique explosiveness, we probed semithin and thin sections with a battery of antibodies to cell wall con-stituents. Although all of the antibody localizations that were attempted in these studies are reported in Table 1 , here we just describe those antibody localizations relevant to the dispersal.

In both bittercress and Arabidopsis , a zone of dehiscence oc-curs along the site where the carpels and replum meet, and in both species, highly de-esterifi ed homogalacturonans labeled with JIM5 and LM19 are absent or weak in these areas just

opening. In bittercress, the cells remain more prominent and a mucilage/polysaccharide aggregation (based upon its similarity in toluidine blue staining with the mucilage in the seed coats) appears to be stored in the intercellular space between the enA layer and the cells of the enB. The staining of the mucilage was variable from section to section and from different areas within the carpel. Even in bittercress, the enA cells appear to be dilated or distorted often bulging into the locule, indicating degenera-tion or distortion from the tension in the silique even before the seed could be discharged ballistically ( Fig. 2C ).

Fig. 2. Light microscopic anatomy of the bittercress silique, at a stage just before being able to dehisce naturally. (A) Cross section through the silique material reveals carpels (c) that together with the central replum (r) form a locule where the seed (s) is held. The outer layer of the seed coat has conspicuous mucilage deposits (*). (B) Higher magnifi cation image through the dehiscence zone, showing the interface between the carpel (c) and the replum (r). The actual point of dehiscence is marked by arrowheads at external and internal sides of the silique. At this site, a number of very small cells occur right at the dehiscence site and a layer of thickened cells occurs in the adjoining replum tissue. (C) Section through the carpel layers reveals the distinct endocarp B (enB) layer where the cell walls proximal to the locule are highly thickened in an inverted V-shaped pattern, while the walls on the distal side are not thickened. Large accumula-tions of mucilage (*) has forced the enA layer cells to loose contact with the enB cells in many cases. Scale bars in (A) = 50 µ m; (B and C) = 25 µ m.

GALLEY PROOF

5August 2011] Vaughn et al. — Explosive seed dispersal in bittercress

Also, the bulk of the Arabidopsis seed coat mucilage does not react with the LM16 antibody ( Fig. 5A ). In contrast, the bitter-cress seed coat mucilage seems to be highly enriched in (modi-fi ed) arabinans (LM16; Fig. 5B ), arabinogalactan proteins (AGPs; labeled with JIM13, Fig. 5C ; and CCRC-M7; not shown), and RGI recognized by CCRC-M22 ( Fig. 5D ), that does not label Arabidopsis seed coat mucilage. Some faint label is observed on the bittercress seed coat mucilage with the antihomogalacturonan antibodies, generally bordering the larger accumulation of arabi-nan/AGP. The columella area of the mucilage cells (the central region where mucilage is not present) is labeled with most of the antibodies that label primary walls in other tissues of the silique. Lignin antisera (G specifi c as well as S and G-S lignins) also strongly label the outermost portion of the columella region, but not the mucilage itself in these cells.

To examine the relationship between the intercellular muci-lage in bittercess more closely, we probed thin sections with the same antibodies used in the light microscopic analysis and ex-amined them via transmission electron microscopy. When sec-tions are labeled with any of the mucilaginous pectin antibodies, the label occurs right up to the highly thickened ends of the enB cells, indicating that the mucilage is tightly associated with the thickenings ( Fig. 6A ). Conversely, label with the xylan anti-bodies is strictly confi ned to the wall thickening ( Fig. 6B ) and does not label the mucilage area although label is present right up to the periphery of the thickening.

DISCUSSION

Similarities and differences between bittercress and Arabi-dopsis siliques — One advantage of studying ballistic seed dispersal in the Brassicaceae is having the well-studied and nonballistic Arabidopsis silique as a model because so much is known about the roles of the various tissues in the silique from the investigation of Arabidopsis mutants (e.g., Lilegren et al., 2004 ; Arsovski et al., 2009 ). In many respects, the siliques of bittercress and Arabidopsis are very similar. The arrangement of locule, seed, replum, and carpels are virtually identical, save the size of the siliques. In the dehiscence zone, the anatomy is virtually identical, and both species initiate the dehiscence pro-cess by the loss of highly de-esterifi ed pectins in the cells of the dehiscence zone. Thus, there are many structures and processes that are quite similar between the two species, despite the radi-cal differences in dispersal mechanisms.

Actually, even in the areas of difference, the differences are still amazingly subtle for such different behaviors in terms of seed dispersal. All of these relate specifi cally to differences in the carpel, specifi cally the enA and enB layers. In bittercress, the enA layer persists for a longer time in the development of the capsules, whereas in Arabidopsis the enA layer degenerates rapidly. In the bittercress silique, a large accumulation of muci-laginous pectins occupies the intercellular space between the enA and enB layers. In the Arabidopsis silique, only a weak labeling of homogalacturonans is noted, perhaps residual from the breakdown of the enA layer. Cells in the enB layer of bittercress have a unique pattern of thickenings in an inverted V- or conjoined teardrop-shaped pattern that is present only on the locule side of the enB cells. These thickenings, but not other areas of the enB layer cell wall, are strongly labeled with anti-bodies to xylan and lignin. The mucilage from the intracellular space is in intimate contact with these wall thickenings. In contrast to the bittercress enB layer, the Arabidopsis enB layer cells are

before opening of the silique but are present in other cells throughout the section ( Fig. 3A, B ). Both of these antibodies label the middle lamellae strongly, indicating that the loss of highly de-esterifi ed pectins in the middle lamellae is responsi-ble for the separation of the valves from the replum in both species. Antibodies that recognize a more esterifi ed epitope (JIM7) or recognize a wide range of HG epitopes (CCRC-M38) do not show the loss of reaction in this area, or they show only small reductions in labeling ( Fig. 3C, D ). Thus, both species are similar in the manner in which the silique is opened: loss of middle lamellae highly de-esterifi ed homogalacturonans.

The carpel walls of the two species are quite different, how-ever. In Arabidopsis , the xylan antibodies LM10 and LM11 strongly labeled the thickened cells in the dehiscence zone and the vascular strands and the enB cells, all rather uniformly ( Fig. 4A ). In contrast, the labeling with these same antibodies on sections of bittercress siliques resulted in a strong labeling on the vascular tissue as in Arabidopsis, but in the highly thickened cells in the enB layer the label was distributed just on the highly thick-ened teardrop-shaped areas facing the locule ( Fig. 4B ). The wall thickenings in bittercress also labeled strongly with antibodies to xyloglucan (CCRC-M1, CCRC-M88, CCRC-M93, CCRC-M108, and LM15), AGP (JIM13), and lignin (G, S, and S-G lignin) (not shown). In addition to this difference in the enB layer, in the bittercress silique, large accumulations of material accu-mulate in the intercellular mucilage space between the enB and enA layers ( Fig. 4C, D ) that is not observed in the Arabidopsis silique. This mucilage accumulation reacts strongly with anti-bodies that recognize all of the homogalacturonans (JIM5, JIM7, LM20, CCRC-M38) found in mucilage and other secretions.

Besides the mucilage in the carpel of the bittercress, mucilage is also a constituent of the seed coat of both species. The mucilage of the Arabidopsis seed coat is known to be composed primarily of a species of rhamnogalcturonan I (RGI), which reacts with the CCRC-M36 monoclonal antibody ( Young et al., 2008 ).

Table 1. Antibodies used in this study and references to these antibodies.

Antibody/antiserum Reference

John Innes and Leeds monoclonals JIM5 Clausen et al., 2003 JIM7 Clausen et al., 2003 JIM8 Knox et al., 1991 JIM13 Knox et al., 1991 LM1 Smallwood et al., 1995 LM5 Jones et al., 1997 LM6 Willats et al., 1998 LM10 McCartney et al., 2005 LM11 McCartney et al., 2005 LM16 Verhertbruggen et al., 2009b LM18 Verhertbruggen et al., 2009a LM19 Verhertbruggen et al., 2009a LM20 Verhertbruggen et al., 2009a Complex carbohydrate center monoclonals CCRC-M1 Freshour et al., 1996 CCRC-M7 Freshour et al., 1996 CCRC-M13 CCRC website a CCRC-M22 CCRC website a CCRC-M38 W. Willats, personal communicationLignin antiserum Gzl (G lignin) Ruel et al., 1994 G-S lignin Joseleau and Ruel, 1997 S lignin Joselau et al., 2004

a CCRC website: http://www.ccrc.uga.edu/~mao/wallmab/Antibodies/antib.htm

GALLEY PROOF

6 American Journal of Botany [Vol. 98

valves from the two species in Fig. 4A, B ). The fl at valve of bittercress will allow curling to the outside much more easily than will the cupped valve of Arabidopsis .

How is the force generated for ballistic seed dispersal in bittercress? — One possibility is that the bending force is pro-vided by the shrinkage of the mesocarp tissue of the carpel as the silique matures and dries out, like other dehiscent brassicas ( Spence et al., 1996 ). The mesocarp cells of bittercress are not reinforced with lignins or xylan that would prevent or limit their shrinkage upon drying, so as these cells dry, they will tend to shrink and thereby exert a tension on the rigid thickenings of the adjacent enB layer. Furthermore, due to the nearly complete loss of HG that was seen in the dehiscence zone of siliques that had not yet sprung ( Fig. 3A, B ), these cells would be very easily separated, because this same loss of de-esterifi ed pectins was found to be suffi cient for abscission in leaves of Impatiens ( Bowling and Vaughn, 2011 ).The mucilage layer located be-tween the enB and enA layers in bittercress would retain moisture longer than the outer pericarp tissues, so it would prevent

thickened, but not to the extent of the bittercress enB cells, and the thickenings are uniform throughout the cell walls (i.e., they are circular in cross section).

What is the function of the distinctive, V-shaped secondary wall thickenings that run down the length of the cells on the side of the cell closest to the locule in bittercress? — A linear object with a V-shaped cross section will take much less force to bend toward the open end vs. the spine. Similarly, the cells of the enB layer in bittercress can be predicted to bend much more easily toward the open end of the “ V ” and away from the apex. In the bittercress, this means that the carpels will bend much more easily to the outside of the silique than toward the inside. So the role of the thickened cells of the enB layer in the bitter-cress may be 2-fold: one to provide longitudinal stiffness to resist shrinkage and the other to provide a bending direction preference. In contrast, the circular (in cross section) thicken-ings of enB layer in the Arabidopsis valve would resist bending in any direction. Furthermore, the valve of bittercress is fl at, while in Arabidopsis it is cupped (compare the shape of the

Fig. 3. Immunocytochemistry with various antihomogalacturonan antibodies of the dehiscence zone region of bittercress. Both (A) JIM5 and (B) LM19 antibodies, which recognize highly de-esterifi ed homogalacturonans (HGs) commonly found in the middle lamellae, reveal a loss of reactivity in the dehiscence zone (between arrowheads). However, (C) JIM7 and (D) CCRC-M38 antibodies, which label HGs with either more highly esterifi ed or a broad range of esterifi cation states, respectively, show no substantial loss of reactivity in this zone. Scale bars = 25 µ m.

GALLEY PROOF

7August 2011] Vaughn et al. — Explosive seed dispersal in bittercress

the enB layer. However, we believe that the tension in the bit-tercress valve is generated by the contraction of the outer peri-carp tissue upon drying, which is more consistent with the general loss of moisture from these organs as they reach matu-rity, not by drying of the siliques and then rehydration of the inner mucilage layer, as would be required by this model.

Is there a general mechanism for coiled structures? — In other systems in which a curling or coiling motion of a structure is induced, two dramatically different cell types or wall regions are juxtaposed to each other ( Witzum and Schulgasser, 1995 ). One type of wall or wall component is one that changes size with moisture content. This type of wall component is expanded when fully hydrated, but eventually dries and shrinks dramati-cally upon maturity of the tissue. The second cell type has a highly rigidifi ed cell wall or cell wall layer that resists bending. The drying and shrinkage of the moisture-sensitive cell type/wall

shrinkage of the cells on the inner surface of the valve (which would hinder the curling action). This retention of moisture may also help keep the lignifi ed enB layer fl exible enough to curl. Another possibility is that this layer is simply functioning as a (greatly expanded) middle lamellae for holding the enB layer together under the tension forces generated as the carpel dries and subsequent curling action of the valve as it springs away from the replum. This may be why bittercress valves were observed not to be able to curl after thourough drying in an oven ( Hayashi et al., 2010 ); the outer layers must be dry while the inner layer is still moist. Otherwise, the tensions may be more in balance between the inner and outer pericarp tissue, while the enB becomes too rigid to bend easily.

An alternative model for the generation of force in the bit-tercress valve was suggested by Hayashi et al. (2010) , whereby the valve curling is driven by expansion of the mucilage layer upon absorption of water against the highly thickened cells of

Fig. 4. Immunocytochemistry of (A) Arabidopsis and (B – D) bittercress siliques. (A) An Arabidopsis silique probed with the LM10 antibody that recognizes xylan. Vascular tissue, thickened cells in the replum (r) and thickened cells in the endocarp B (enB) layer (arrowhead) are labeled. Note that the secondary wall thickenings of the enB layer in Arabidopsis are uniform. (B) Bittercress silique cross section labeled with LM10 reveals a similar labeling pattern to the Arabidopsis siliques, except that secondary wall thickenings of the the enB layer (arrowhead) display a characteristic inverted V-shape on the side proximal to the locule. (C) A cross section through the carpel of bittercress reveals the strong labeling of the mucilage (m) between the endocarp A (enA) and enB layers with the JIM5 antibody. (D) A similar section to (C), but probed with the CCRC-M38 antibody reveals similar strong labeling of the mucilage (m). Scale bar in (A, B, D) = 50 µ m; (C) = 25 µ m.

GALLEY PROOF

8 American Journal of Botany [Vol. 98

have shown that the average is more like 0.25 m ( Bachman and Whitwell, 1994 ). These differences in the distance of dispersal may refl ect differences in silique drying, developmental stages of the parent plant, or seed load within the silique, but such vari-ability would allow for a more balanced spread of seed, both near and more distant from the parent plant. Similar distances of seed dispersal were observed for jewelweed ( Hayashi et al., 2009 ) and in another species of Cardamine . In the case of jew-elweed, the plants require relatively moist conditions so that, although the seed would be deposited a good distance from the parent plant, it is also likely that the seed would be deposited on another site with similar favorable growth conditions to what the parent plant was exposed. In fact, bittercress seedlings seem to occur in patches in nursery crops and will solidly fi ll nursery pots if left uncontrolled. Bittercress seems to occur in more eco-logical niches than jewelweed, but the ability to move propagules to other close sites has made it a major pest in nurseries and lawns, where it can quickly spread at many favorable sites. In addition, an individual plant may produce up to 5000 seeds ( Bachman and Whitwell, 1994 ) and would quickly make the bittercress plant a formidable weed problem, considering a

layer puts pressure on the adjoining rigid layer to bend. In the highly coiled pseudo-elaters of hornworts ( Carafa et al., 2005 ; Kremer et al., 2004 ) and in the G fi bers of trees and vines ( Meloche et al., 2007 ; Bowling and Vaughn 2008 , 2009 ), this force is accomplished by a single cell type, with highly hydrated wall layers (containing AGPs or pectins) directly in contact with wall layers that are highly thickened (containing xylans and lignin). As the layers with AGPs or pectins shrink, they put tension on the lignifi ed rigid layers and exert tension. The result is that the cells must change morphology to relieve the tension, and so the characteristic coiled shape of tendrils and pseudoe-laters is generated.

Advantages to ballistic seed dispersal — Ballistic seed dis-persal offers some obvious advantages, although it might not be what would be most benefi cial for some species. Of course, one of the advantages is that the plant does not require any other agent, biotic or abiotic, to enhance seed dispersal. Thus, bitter-cress seed may be dispersed even in the absence of wind, rain, or animals. Ballistic seed expulsion from bittercress siliques can be as much as 5 m from the parent plant, although other studies

Fig. 5. Seed coat labeling in (A) Arabidopsis and (B – D) bittercress seeds with (A, B) the LM16, (C) M22, and (D) JIM13 antibodies. The seed coat mucilage of bittercress is labeled strongly with antibodies that recognize modifi ed arabinans, arabinogalactan proteins (AGPs), and rhamnogalacturonan I (RGI). Bar = 50 µ m.

GALLEY PROOF

9August 2011] Vaughn et al. — Explosive seed dispersal in bittercress

Bachman , G. , and T. Whitwell . 1994 . Hairy bitttercress ( Cardamine hirsuta ) seed production and dispersal in the propagation of landscape plants. Proceedings of the Southern Nursery Association Research Conference 39 : 299 – 302 .

Benvenuti , S. 2007 . Weed seed dispersal strategies in the agricultural environment. Weed Biology and Management 7 : 141 – 157 .

Bowling , A. J. , and K. C. Vaughn . 2008 . Immunocytochemical studies of tension wood: Gelatinous fi bers contain more than just cellulose. American Journal of Botany 95 : 655 – 663 .

Bowling , A. J. , and K. C. Vaughn . 2009 . Gelatinous fi bers are widespread in tendrils and vines. American Journal of Botany 96 : 719 – 727 .

Bowling , A. J. , and K. C. Vaughn . 2011 . Leaf abscission in Impatiens (Balsaminaceae) is due to loss of highly de-esterifi ed homogalac-turonan in the middle lamellae. American Journal of Botany 98: 619 – 629 .

Carafa , A. M. , J. G. Duckett , J. P. Knox , and R. Ligrone . 2005 . Distribution of cell wall xylans in bryophytes and tracheophytes: Insights into basal interrelationships of land plants. New Phytologist 168 : 231 – 240 .

generation can occur in 30 – 60 d and in favorable climates growth year round. Another major advantage of ballistic seed dispersal is the potential antiherbivory characteristics of these plants. Yano (1997) has shown that caterpillars may be shot sev-eral meters from the site of silique explosion. This may be one of the few instances where plants actively fi ght off the would-be herbivore! Thus, with its combination of large numbers of seeds, quick generation times, year -round growth, and mechanisms to enhance dispersal and discourage predation, bittercress is well poised to be a weed of signifi cance in many situations.

LITERATURE CITED

Arsovski , A. A. , T. M. Popma , G. W. Haughn , N. C. Carpita , M. C. McCann , and T. L. Western . 2009 . AtBXL1 encodes a bifunc-tional β - d - xylosidase/ α - l -arabinofuranosidase required for pectic arabinan modifi cation in Arabidopsis mucilage secretory cells. Plant Physiology 150 : 1219 – 1234 .

Fig. 6. Bittercress endocary B (enB) layers probed with (A, C) CCRC-M38 and (B, D) LM10. (A, C) Label with the CCRC-M38 antibody is confi ned to the mucilage layer (m) although it is contiguous with the highly thickened (t) inverted V-shaped walls of the enB layer. (B, D) LM10 antibody labeling is restricted to the thickened cell wall areas (t) and does not label the mucilage (m). Bar = 1.0 µ m.

GALLEY PROOF

10 American Journal of Botany [Vol. 98

Clausen , M. H. , W. G. T. Willats , and J. P. Knox . 2003 . Synthetic methyl hexagalacturonate hapten inhibitors of anti-homogalacturonan monoclonal antibodies LM7, JIM5 and JIM7. Carbohydrate Research 338 : 1797 – 1800 .

Freshour , G. , R. Clay , M. S. Fuller , P. Albersheim , A. G. Darvill , and H. G. Hahn . 1996 . Development and tissue-specifi c structural altera-tions of the cell-wall polysaccharides of Arabidopsis thaliana roots. Plant Physiology 110 : 1413 – 1429 .

Hayashi , M. , K. L. Felich , and D. J. Ellerby . 2009 . The mechanics of explosive seed dispersal in orange jewelweed ( Impatiens capensis ). Journal of Experimental Botany 60 : 2045 – 2053 .

Hayashi , M. , S. P. Gerry , and D. J. Ellerby . 2010 . The seed dispersal catapault of Cardamine parvifl ora (Brassicaceae) is effi cient but un-reliable. American Journal of Botany 97 : 1595 – 1601 .

Jones , L. , G. Seymour , and J. P. Knox . 1997 . Localization of pectic ga-lactan in tomato cell wall using a monoclonal antibody specifi c to (1 → 4)- β - d -galactan. Plant Physiology 113 : 1405 – 1412 .

Joseleau , J. P. , O. Faix , K. I. Kuroda , and K. Ruel . 2004 . A polyclonal antibody directed against syringylpropane epitopes of native lignins. Compte Rendus Biologie Paris 327 : 809 – 816 .

Joseleau , J.-P. , and K. Ruel . 1997 . Study of lignifi cation by nonin-vasive techniques in growing maize internodes (An investigation by Fourier transform infrared cross-polarization-magic angle spinning 13C-nuclear magnetic resonance spectroscopy and immunocyto-chemical transmission electron microscopy). Plant Physiology 114 : 1123 – 1133 .

Knox , J. P. , P. J. Linstead , J. Peart , C. Cooper , and K. Roberts . 1991 . Developmentally regulated epitopes of cell surface arabinogalactan-proteins and their relation to root tissue pattern formation. Plant Journal 1 : 317 – 326 .

Kremer , C. , F. Pettolino , A. Bacic , and A. Drinnan . 2004 . Distribution of cell wall components in Sphagnum hyaline cells and in liverwort and hornwort elaters. Planta 219 : 1023 – 1035 .

Liljegren , S. J. , A. H. K. Roeder , S. A. Kempin , K. Gremski , L. Ostergard , S. Guimil , D. K. Reyes , and M. F. Yanofsky . 2004 . Control of fruit patterning in Arabidopsis by INDEHISCENT. Cell 116 : 843 – 853 .

Louvet , R. , E. Cavel , L. Guiterrez , S. Guenin , D. Roger , F. Gillet , F. Guerineau , and J. Pelloux . 2006 . Comprehensive expression profi ling of the pectin methlyesterase gene family during silique de-velopment in Arabidopsis thaliana. Planta 224 : 782 – 791 .

McCartney , L. , S. E. Marcus , and J. P. Knox . 2005 . Monoclonal antibodies to plant cell wall xylans and arabinoxylans. Journal of Histochemistry and Cytochemistry 53 : 543 – 546 .

Meloche , C. G. , J. P. Knox , and K. C. Vaughn . 2006 . A cortical band of gelatinous fi bers causes the coiling in redvine tendrils: A model based upon cytochemical and immunocytochemical studies. Planta 225 : 485 – 498 .

Mitsuda , N. , and M. Ohme-Takagi . 2008 . NAC transcription factors NST1 and NST3 regulate pod shattering in a partially redundant man-ner by promoting secondary wall formation after the establishment of tissue identity. Plant Journal 56 : 768 – 778 .

Ogawa , M. , P. Kay , S. Wilson , and S. M. Swain . 2009 . ARABIDOPSIS DEHISCENCE ZONE POLYGALACURONASE 1 (ADPG1), ADPG2 and QUARTET2 are polygalacturonases required for cell separation during reproductive development in Arabidopsis. Plant Cell 21 : 216 – 233 .

Ruel , K. , O. Faix , and J. P. Joseleau . 1994 . New immunogold probes for studying the distribution of the different lignin types during plant cell wall biogenesis. Journal of Trace and Microprobe Techniques 12 : 247 – 265 .

Simons , P. 1992 . The action plant. Blackwell, Oxford, UK. Skotheim , J. M. , and L. Mahadevan . 2005 . Physical limits and design

principles for plant movements. Science 308 : 1308 – 1310 . Smallwood , M. , H. Martin , and J. P. Knox . 1995 . An epitope of rice

threonine- and hydroxyproline-rich glycoprotein is common to cell wall and hydrophobic plasma membrane glycoproteins. Planta 196 : 510 – 522 .

Spence , J. , Y. Vercher , P. Gates , and N. Harris . 1996 . ‘ Pod shatter ’ in Arabidopsis thaliana, Brassica napus and B. juncea. Journal of Microscopy 181 : 195 – 203 .

Taylor , P. E. , G. Card , J. House , M. H. Dickinson , and R. C. Flagan . 2006 . High-speed pollen release in the white mulberry tree, Morus alba L. Sexual Plant Reproduction 19 : 19 – 24 .

van der Pijl , L. 1972 . Principles of dispersal in higher plants. Springer, Berlin, Germany.

Verhertbruggen , Y. , S. E. Marcus , A. Haeger , J. J. Ordaz-Ortiz , and J. P. Knox . 2009a . An extended set of monoclonal antibodies to pectic homogalacturonan. Carbohydrate Research 344 : 1858 – 1817 .

Verhertbruggen , Y. , S. E. Marcus , A. Haeger , R. Verhoef , H. A. Schols , B. V. McCleary , L. McKee , H. J. Gilbert , and J. P. Knox . 2009b . Developmental complexity of arabinan polysaccharides and their pro-cessing in plant cell walls. Plant Journal 59 : 413 – 425 .

Willats , W. G. T. , S. E. Marcus , and J. P. Knox . 1998 . Generation of a monoclonal antibody specifi c to (1 → 5)- α - l -arabinan. Carbohydrate Research 308 : 149 – 152 .

Witztum , A. , and K. Schullgasser . 1995 . The mechanism of seed expul-sion in Acanthaceae. Journal of Theoretical Biology 176 : 531 – 542 .

Yano , S. 1997 . Silique burst of Cardamine scutata (Cruciferae) as a phys-ical inducible defense against seed predatory caterpillars. Researches on Population Ecology 39 : 95 – 100 .

Young , R. E. , H. E. McFarlane , M. G. Hahn , T. L. Western , G. W. Haughn , and A. L. Samuels . 2008 . Analysis of the golgi appara-tus in Arabidopsis seed coat during polarized secretion of pectin-rich mucilage. Plant Cell 20 : 1623 – 1638 .

GALLEY PROOF