Protoplasma (1997) 199:173-197

PROTOPLASMA
9 Springer-Verlag 1997
Printed in Austria

Covisualization in living onion cells of putative integrin, putative spectrin, actin,

putative intermediate filaments, and other proteins at the cell membrane and in
an endomembrane sheath

Christophe Reuzeau**, Keith W. Doolittle, James G. McNally, and Barbara G. Pickard*

Biology Department, Washington University, Saint Louis, Missouri

Received February 18, 1997

Accepted July 2, 1997

Summary. Covisualizations with wide-field computational optical- Keywords: Actin; Adhesion sites; Allium cepa; Endomembrane
sectioning microscopy of living epidermal cells of the onion bulb system; Integrin; Spectrin.
scale have evidenced two major new cellular features. First, a sheath
of cytoskeletal elements clads the endomembrane system. Similar Abbreviations: DiOC6 3,3'-dihexyloxacarbocyanine iodide; GF
elements clad the inner faces of punctate plasmalemmal sites inter~ Gaussian filtering; ML maximum likelihood (algorithm or method);
preted as plasmalemmal control centers. One component of the endo- PBS phosphate-buffered saline
membrane sheath and plasmalemmal control center cladding is anti-
genicity-recognized by two injected antibodies against animal spec-
trin. Immunoblots of separated epidermal protein also showed bands Introduction
recognized by these antibodies. Injected phalloidin identified F-actin
The endomembrane system of the interphase plant
with the same cellular distribution pattern, as did antibodies against
intermediate-filament protein and other cytoskeletal elements known
cell, like that of the better-studied endomembrane
from animal cells. Injection of general protein stains demonstrated system of animal cells, is extensive and is dynamical-
the abundance of endomembrane sheath protein. Second, the endo- ly responsive to prevailing conditions. The features of
membrane system, like the plasmalemmal puncta, contains antigen a major component, the endoplasmic reticulum (ER),
recognized by an anti-J31 integrin injected into the cytoplasm. Previ- are well reviewed by, e.g., Hepler et al. (1990) and
ously, immunobiots of separated epidermal protein were shown to
have a major band recognized both by this antibody prepared against
Gunning and Steer (1996); pertinent original papers
a peptide representing the cytosolic region of [31 integrin and an anti- include those by Quader and Schnepf (1986), Quader
body against the matrix region of [31integrin, The latter antibody also et al. (1987), Allen and Brown (1988), Craig and Sta-
identified puncta at the external face of protoplasts. It is proposed ehelin (1988), Knebel et al. (1990), Lichtscheidl and
that integrin and associated transmembrane proteins secure the endo- Url (1990), Quader (1990), and Oparka et al. (1994).
membrane sheath and transmit signals between it and the lumen or
There is evidence that the ER is intricately connected
matrix of the endoplasmic reticulum and organeltar matrices. This
function is comparable to that proposed for such transmembrane with the major cytoskeletal polymers F-actin and
linkers in the plasmalemmal control centers, which also appear to myosin (Allen and Brown 1988, Lichtscheidl et al.
bind cytoskeleton and a host of related molecules and transmit sig- 1990, Liebe and Menzel 1995, Quader and Liebe
nals between them and the wall matrix. It is at the plasmalemmal 1995) and sometimes with microtubules (Hepler et al.
control centers that the endoplasmic reticulum, a major component
1990; see also Gunning and Steer t 996: plate 31), and
of the endomembrane system, attaches to the plasma membrane.
that it has links to the plasma membrane (e.g., Quader
*Correspondence and reprints: Biology Department, Washington
et al. 1987, Lichtscheidl and Url 1990, Pont-Lezica et
University, Saint Louis, MO 63130-4899, U.S.A. al. 1993, Oparka et al. 1994). It connects cells via the
E-mail: pickard@ allenlab.wustl.edu plasmodesmata (Grabski et al. 1993, Oparka et al.
**Current address: Department of Plant Biochemistry, University of 1994). It serves as an important reservoir for calcium
Lund, Lund, Sweden. ions which helps to regulate their level in the cyto-
174 C. Reuzeau et al.: Integrin and cytoskeletal proteins in onion cells

plasm (Hepler et al. 1990). By definition, the endo- "correctly" with its carboxyl terminus in the cyto-
membrane system wraps each organelle. The ER plasm. Second, when the cell is mechanically stressed
serves as a surface on which the machinery of protein the endomembranes shift from an "extended" to a
synthesis can organize. Its lumen and vesicles derived "constricted" state, revealing some of the endomem-
from it convey proteins and carbohydrates to the brahe and {3x integrin colocalized in puncta at the
external face of the plasma membrane. Understanding inner face of the plasma membrane. Third, all five of
the endomembrane system appears to be an important the examined cytoskeletal antigenicities/elements
key to understanding cell activities. also colocalize with the endomembranes. We surmise
Yet, our view of the endomembrane system is evi- that cytoskeleton clads the endomembranes with a
dently quite incomplete. Historically, there has been structure we name the endomembrane sheath. Com-
little alternative to the potentially disruptive methods paring the distribution of the puncta seen at the inner
of chemically fixing or physically freezing plant cyto- face of the plasma membrane with 13t integrin puncta
plasm, and efforts to localize proteins have often previously reported (Gens et al. 1996a; see also Gens
relied on high-magnification examination of very et al. 1997) at the outer face, we surmise that the
small regions of the potentially disrupted cytoplasm. puncta represent adhesion sites.
Moreover, although there is growing interest in the It is clear that these observations and surmises have
possible diversity of cytoskeletal proteins in plants implications for the regulation of cellular activities.
(Reuzeau and Pont-Lezica 1995), for a long while the We will incorporate them into a model for metabolic
focus of research has been primarily on actin and regulation and cytoplasmic streaming (Reuzeau et al.
microtubules plus a few associated proteins (e.g., 1997b). We will also speculate on some consequences
Goddard et al. 1994, Penman 1995). of the stress-induced shift of the endomembranes, and
For the same reasons, our view of the plasma mem- on some mechanisms of stress resistance.
brane has been restricted. The concept of plasmalem- Preliminary reports of this work have been presented
real control centers or adhesion sites functionally (Pickard et al. 1994, 1995; Reuzeau et al. 1995a, b, c).
comparable to those of animals is just emerging, as
considered, e.g., in another paper in this series (Gens Material and methods
et al. 1996a; see also Gens et al. 1997), and the nature Antibodies and fluorochromes
of cytoskeletal proteins that associate with the plasma All fluorochromes were obtained from Molecular Probes, Inc.,
membrane is just beginning to be explored. Here, it is Eugene, OR.
useful to note, we define cytoskeleton in a broad All antibodies utilized were of the IgG class. Anti-[51 integrin was a
gift from Eugene Marcantonio, Columbia University, New York, NY
sense to mean most protoplasmic proteins whose pri-
and Richard O. Hynes, Howard Hughes Institute Fellow at Massa-
mary function is architectural. chusetts Institute of Technology, Cambridge, MA. It was a polyclon-
This paper participates in the accelerating trend to use al rabbit IgG raised against the well-conserved 39-residue sequence
electronic microscopy to examine adhesion sites, of chicken [51 integrin at the terminus extending into the cytoplasm
endomembranes, and the cytoskeleton in plant cells. (Marcantonio and Hynes 1988). Monoclonal antibody reacting with
all classes of intermediate filaments was a gift from Mary Jane Saun-
In particular, it reports on results with wide-field
ders, University of South Florida, Tampa, FL, who prepared it from
computational optical-sectioning microscopy, a
hybridoma cells (American Type Culture Collection TtB 131)
method with a level of spatial resolution and signal (McNulty and Sannders 1992). All other antibodies were from Sigma
discrimination suitable for global examination of the Chemical Co., St, Louis, MO; they were: anti-human (c~ and [5) spec-
distribution and codistribution of low-abundance trin (ascites fluid, clone SB-SP1) (S 3396); polyclonal rabbit anti-
cytoskeletal proteins in living epidermal cells from human (c~ and [5) spectrin (S 1515); anti-chicken vinculin (ascites
fluid, clone V[N-11-5) (V 4505); anti-chicken talin (ascites fluid,
the inner side of mature onion bulb scales. In describ-
clone 8d4) (T 3287).
ing the distribution of proteins that seem on several All antibodies were affinity-purified with a 5-ml Hitrap protein G
grounds to correspond to the cytoskeleton-anchoring column from Pharmacia Biotech, Inc., Piscataway, NJ. The column
transmembrane protein 61 integrin and the cytoskele- was equilibrated with 20 mM sodium phosphate buffer at pH 7.0.
tal proteins spectrin, intermediate filaments, and F- Then, approximately 400 [xg of antibody was diluted into 3 ml of the
same buffer and applied to the column. The column was washed with
actin, as well as antigenicities putatively correspond-
30 ml of the buffer, and it was checked that no protein appeared in
ing to vinculin and talin, three important findings the final fractions of effluent. The column was then washed with 20
stand out. ml of 0.1 M glycine-HCl at pH 2.7 to release IgG. The eluate was
First, ~1 integrin is abundant within the cell and colo- collected in 1-ml fractions, each of which was neutralized immedi-
calizes with the endomembrane system, oriented ately with 80 vtl of 1 M Tris at pH 9.0. The 280 nm absorbance of
C. Reuzeau et al.: Integrin and cytoskeletal proteins in onion cells 175

each tube was assessed in order to determine which contained the times with a mixture of 250 mM NaC1, 0.05% Tween 20, and 30 mM
IgG. The buffer was then shifted to 0.2 M sodium carbonate at pH Tris at pH 8.0. Product was released into Laemmli's (1970) sample
9.0 by passage of sample and solution through a Pdl0 column, buffer by boiling the beads in it, and SDS-PAGE was then carried out
Sephadex G25, from Pharmacia. as above.
Antibodies and non-immune mouse serum (Sigma I5381) were con- Western bIots were prepared by either the semi-dry method, as
jugated with isomer 1 of FITC or with Texas red sulfonyl chloride, described earlier, or with a liquid system (Towbin et al. 1979), and
using kits and protocols from Molecular Probes. The extent of con- stained as described earlier (Gens et al. 1996). The spectrin prepara-
jugation was checked by measuring absorbance ratios at 495 nm/280 tions and controls were treated with a 3/5000 dilution of either
nm for FITC and 596 nm/280 nm for Texas red. The ratios ranged monoclonal anti-spectrin or polyclonal anti-spectrin. Bound antibod-
between 0.5 and 4.0. Conjugated IgGs were concentrated using Cen- ies were visualized by alkaline phosphatase color assay as described
trex UF-2 10K MWCO centrifugal ultrafilters (Schleicher and Schu- previously (Gens et al. 1996).
ell, Keene, NH) and refrigerated until use.
Preparation and injection of cells
Depletion of antibodies by authentic antigen
Onion bulbs were obtained and stored as previously described (Ding
A sample of the FITC-conjugated monoclonal anti-(ct and 13) spectrin and Pickard 1993, Gens et al. 1996). Rectangles of epidermis were
was depleted of specific binding capability by incubation for 12 h at gently peeled from the inner face of bulb scales, and floated (cuticle
4 ~ with cyanogen bromide-activated Sepharose beads from Sigma side up) for 1 h in dim light on a bath containing 50 mM mannitol (or
coupled according to Sigma protocol with pure spectrin from human more if the cells were to be osmotically challenged), 10 mM MES-
erythrocytes (Sigma S 3644), followed by removal of the beads. Tris at pH 5.7, 10 mM KC1, and 10 mM CaCI2. (The purpose of the
A sample of the Texas red-conjugated anti-13I integrin was depleted 50 mM mannitol was to maintain cell turgor at a level comparable
of specific binding capability by incubation with an immunoprecipi- with that when the sheet was fieshly pulled from the bulb scale.)
tation product from onion cells, as follows. Monoclonal anti-chicken Next, if fibers were large and abundant at the cortical face of a sheet,
131 integrin (Sigma I 8638, clone W1B10), raised against the domain they were gently scraped away with a spatula. Peels were refloated
of that protein once considered to be exclusively extracellular, was for at least 0.5 h on fresh solution. Their cells might then be injected,
reacted with protein A agarose beads (Sigma) and cross-linked with but more often peels were centrifuged to amass a more readily inject-
dimethyl pimelidate (20 mM; from Sigma). Crude onion epidermis ed volume of cytoplasm at the centrifugal pole of those cells with
extract was obtained using the Hepes medium of Ito et al. (1992) as tapered ends. For this centrifugation, the peel was laid with the corti-
previously described (Gens et al. 1996). The extract was incubated cal face against a double sheet of Parafilm laboratory film (American
with the beads for 12 h at 4 ~ the mixture was centrifuged at 2000 g National Can, Neenah, WI) mounted on a glass microscope slide;
for 15 min, the pelleted beads were washed three times with a mix- around the edge of the film had been built a narrow wall consisting
ture of 250 mM NaC1, 0.05% Tween 20, and 30 mM Tris at pH 8.0. of two more layers of the film. Bath was provided inside the wail;
The immuno-reacted protein was eluted from the beads with then, a second glass slide was emplaced as a lid and secured with two
100 mM glycine at pH 2.7, and the buffer was changed to phosphate- or three bands of Scotch Magic brand tape (3M, St. Paul, MN). The
buffered saline (PBS) by use of a centrifugal ultrafilter. The PBS- unit was slipped into a 50-ml clear polypropylene centrifuge tube and
buffered protein was gently agitated for 12 h at 4 ~ with Texas red-
about 20 ml of bath was added until the upper edge of the tissue strip
conjugated polyclonal anti-integrin prepared against a peptide corre- was just submerged. The tube was capped and centrifuged for at least
sponding to the cytoplasmic domain of 131 integrin (Marcantonio and 30 rain with an acceleration ranging, along the length of the tissue,
Hynes 1988).
from 25 to 30 g. Because the small increment of cytoplasm forced to
the tips of the cells fully returned to the rest of the cell within 15 min,
Purification and characterization of spectrin-like antigen from onion it was necessary to inject immediately. We could find no evidence of
The first step in obtaining spectrin-like antigen was to obtain epider- cellular damage by such centrifugation, in agreement with a study by
mal sheets and to extract them under conditions as specified by Gens Quader et al. (1987).
et al. (1996) using two different buffers. The first was the Hepes For injection, an epidermal sheet was laid, cuticle up, on a drop of
buffer ofIto et aI. (1992), and the second was the PBS buffer of Eno- bath on a 24 • 50 • (0.16-0.19) mm "cover glass", and this was
moto-Iwamoto et al. (1993). The crude extracts were cleared by mounted on the stage of a Nikon Diaphot microscope provided with
treating 4 h with anti-IgG agarose beads (Sigma) and removing the Hoffman modulation contrast optics (objective lenses • 0.4 NA
sediment. and • 0.5 NA; Modulation Optics Inc., Greenvale, NY) and
Initially, proteins in crude extract were separated by native poly- equipped for fluorescence. Working in very dim, usually red, light, a
acrylamide gel electrophoresis (Chrombach and Rodbard 1971), labeling solution was loaded by capillary uptake into the tip of a
with 6% gel, or by sodium dodecyl sulfate-polyacrylamide gel elec- glass pipette pulled from a capillary tube. Any given solution con-
trophoresis (SDS-PAGE) (Laemmli 1970), with 10% gel. Molecular si~t~ . . . . .~,~to three fluorescent stains dissolved in bath solution; the
mass standards for the former method were the 272 kDa trimer and concentrations were 0.46 g/l for antibodies (concentration based on
545 kDa tetramer of urease and the 66 kDa monomer and 132 kDa measurement of TCA precipitate with a Micro BCA protein assay
dimer of bovine serum albumin, all from Sigma. Pre-stained molec- reagent kit supplied by Pierce Chemical Co., Rockford, IL), 6.6 ~tM
ular mass standards for SDS-PAGE were from a low-range kit pro- for rhodamine phalloidin, 7 mM for Lucifer yellow, 3.6 mM for Cas-
vided by Bio-Rad Laboratories, Inc., Hercules, CA. cade blue, and 2.5 mM for Cascade blue conjugated to dextran with
In later experiments an extra, immunopurification, step was inter- molecular weight 3000. The pipette was back-filled with silicone
jected. Crude extract was combined with anti-IgG agarose beads to fluid [melting point bath oil, 2 stokes (= 2 • 10-4 m:/s) at 25 ~
which either monoclonal anti-spectrin or non-immune mouse serum Sigma M 6884] and mounted in a chuck connected to a hydraulic
had been coupled. After incubation, the beads were washed three injector similar to that described by Oparka et al. (1991). Using a
176 C. Reuzeau et al.: Integrin and cytoskeletal proteins in onion cells

micromanipulator, the pipette tip was positioned at a corner of a cell. (1996). Jos6-Angel Conchello (Conchello and McNally 1996) devel-
An instant before entry, the pipette contents were pressurized to a oped a regularized version of ML for restoring images such as that of
value judged likely to buck cellular turgor pressure (275-350 kPa), Fig. 5 A; it is available with the fundamental ML algorithm and user-
monitoring the pipette pressure with a digital meter. The pipette tip friendly interfaces at World Wide Web site http:]/ibc.wustl.edu:
was slipped into the cytoplasm at the tip of a cell and the hydraulic 80/bcl/.
pressure was elevated enough to drive a small volume of the fluoro-
chromes into the cell (350-700 kPa). A little bath solution was added
to the tissue to prevent drying, and the pipette was then left in posi-
tion for 20-30 rain before removal to encourage sealing of the mem-
Results
brane and wall at the point of withdrawal. In some cases, the tissue States of the endomembrane system
was then floated for about 3 min in 8.3 gM DiOC6 in bathing medi-
Extensive observation of the ER and organelles after
um and washed in DiOC6-free bath. (Uncentrifuged, uninjected, or
uncentrifuged and uninjected DiOC6 controls might be prepared at floating epidermal peels on solutions of the widely
the same time.) Finally, a second "cover glass" was placed over the used lipophilic stain DiOC6 confirmed and extended
preparation. It was transported in a light-proof box to the computa- previous findings (e.g., Quader et al. 1989, Quader
tional optical-sectioning microscope, and mounted cuticle side away 1990, Quader and Fast 1990, Liebe and Quader 1994,
from the inverted lens for photography. When scanning cells for pho-
Liebe and Menzel 1995, Quader and Liebe 1995,
tography, every effort was made to find fields without cytoplasmic
streaming either in the volume to be imaged or in transvacuolar Fuhrmann et al. 1990, Stickens and Verbelen 1996)
strands above it. that the state of the endomembrane system depends
on its recent history of stimuli and stresses. Here we
Microscopy deal with mechanically stressed cells, and the differ-
Wide-field computational optical-sectioning microscopy was select- ent configurations of the cells following different
ed over the more commonly used laser confocal microscopy because degrees of stress have been utilized to visualize distri-
the former method (1) requires far less excitation light and (2) can
butions of proteins associated with the plasmalemma
utilize all available emitted light, and hence can provide better
images of small dim fluorescent objects which are readily bleached and endomembranes. Typically, mechanical stress has
(Carrington et al. 1989). been applied by injecting fluorochromic markers at
The microscope and data acquisition system are described and refer- relatively low or high pressures, but for the baseline
enced by Gens et al. (1996). As before, CCD camera wells of 6.8 Bm series of observations made with DiOC6 cells were
width were binned 4 x 4. However, optical sections were obtained
stressed by pressing or bending the epidermal peels to
with three different oil immersion objective lenses and three corre-
sponding cubic voxel sizes: (1) an Olympus (Lake Success,
deform them. For all cells, an additional possible
NY) X40, 1.0 NA Dplan Apo UV lens, with a cubic voxel size set at component of sensory input is gradual decrease in
X = Y = Z = 0.67 ~tm, (2) a Nikon (Garden City, NY) X60, 1.4 NA water potential and oxygen level in the bath during
Plan Apo lens with X = Y = Z = 0.45 ~tm, or with an Olympus X 100, observation on the microscope stage. While there is
1.3 NA Dplan Apo UV lens with X = Y = Z = 0.27 ~tm. Processing continuous gradation from unstressed to heavily
by Gaussian filtering (GF), restoration by 1000 iterations of the
stressed cell states, for convenience of discussion this
maximum likelihood algorithm (ML), and projection of optical sec-
tions to make stereo pairs (in each case with displacement angle c~ = continuum will be divided into three categories:
_+4 ~ or two-dimensional images are also discussed by Gens et al. unstressed, mildly stressed, and highly stressed.

Fig. 1 A-F. Some common states of the endomembrane system in onion epidermal cells. A - E Single sections from five cells showing endo-
membranes stained exclusively with DiOC6 applied from outside the cell; the objective lens was either X40, 1.0 NA or • 1.4 NA; image
stacks were restored with ML and then sections were excerpted and processed by GF. A - C Membranes in characteristic "low-stress" configu-
rations. A Characteristic cortical ER with polygons, some webbed with small cisternal lamellae. B Broadly stranded regions of ER with associ-
ated and interspersed polygonal "fenestrae". C A "sheet" or cisternal lamella of intermediate size (center of image) with larger sheets slightly
out of focus at the top and bottom of the panel, plus a thin, isolated streaming track and numerous small figures some of which are probably the
more corpuscular organellar components of the endomembrane system. D and E Membranes in characteristic "high-stress" configurations, with
small, sometimes irregular, puncta (primarily vesicles and organelles) connected by fine strands (tubules). F1 Larger presentation of the endo-
membrane system in a "high-stress" configuration in a cell treated from the outside with DiOC6 (pseudocolored green) and injected with anti-
vinculin (pseudocolored red), as photographed with a X 100, 1.3 NA lens and restored with ML. The yellow overlap color is extensive, with
numerous round objects showing oranger. These appear as rings in a monocolor 2-D projection of the 64-section anti-vinculin image stack
alone. Section-by-section viewing of that stack permitted determination that at least many of the rings seen in projection are hollow vesicles and
that most of the vesicles and smaller spots of the image are connected by threads interpreted as fine tubules; this is illustrated for a small re-
gion near the left edge of F1 by displaying selected sections (1, 5, 9, 11, 13, 15, 17, 19, and 21; F2-10) with a nonlinear yellow scale (color bar
in F10). Bars: 20 ~tm
178 C. Reuzeau et al.: Integrin and cytoskeletal proteins in onion cells

The unstressed state is characterized by large cisternal ble with the • 1.4 NA lens and was readily over-
lamellae and extensive regions of reticulum, with a looked even with the • 100, 1.3 NA lens when the
great deal of cytoplasmic streaming. Because neither resolving capacities of the computer system were par-
large lamellae nor active streaming strands were read- tially devoted to multichromatic imaging as required
ily imaged with our system, only extensive reticula for covisualization or when a 3-D stack of sections
will be further considered. was projected (Fig. 1 F1), even if tubules were readi-
The mildly stressed state is characterized by cortical ly evident when sections were examined serially (Fig.
reticula which are usually partially webbed (Fig. 1 F2-10). The projection of Fig. 1 F1 shows fluores-
1 A); by sheets and strands with fenestrae (Fig. 1 B); cence from DiOC6 in green and from anti-animal vin-
and by small cisternal lamellae and fine tubules (Fig. culin in red, with the essentially total overlap signi-
1 C). The small lamella in the center of Fig. 1 C fied by yellows, whereas the selected and enlarged
appears dim in comparison to overlying organelles sections from the left middle portion of Fig. 1 F1 dis-
and patchy cisternae, and less conspicuous than the play anti-vinculin separately using a nonlinear yel-
tube or strand slanting across the image. It is likely low-scale in order to evidence both dim and bright
that the strand was recently streaming; limited and features in the same view. (The antibody image was
relatively slow movement of organelles and small ER selected for individual display because, as was gener-
strands is common in mildly stressed cells. As illus- ally the case when an antibody and DiOC6 were pro-
trated, extensive views of the plasmalemma are rare vided together, the antibody formed the sharper
because it is occluded by the large expanse of endo- image.) The many hollow spheres of uniform size
membrane features. might possibly be Golgi vesicles, and the frequent
The highly stressed state is characterized by the association of these vesicles with faint tubules is con-
absence of large ER cisternae and the consequent sistent with the possibility (Ayala 1994, Morr6 and
conspicuousness of cytoplasmic organelles, tiny plas- Keenan 1994) that some or all of them may be
malemma-associated "puncta", and thin tubules (Fig. swollen, coated termini of tubules. The anti-vinculin
1 D, E). The larger organelles such as plastids and seemed to stain the hollow spheres more strongly than
mitochondria are often cloaked with a little extra the rest of the endomembrane (note the reddish-
membrane, so that their outlines may be somewhat orange color of these bodies in Fig. 1 F1), and it must
irregular. Given the imaging parameters selected for be emphasized that anti-vinculin is shown in Fig. 1 F
the majority of observations of this paper, the thin simply as an interesting marker to delineate the fea-
tubules are close to the limits of resolution. They tures of the endomembrane system. Because vinculin
appear to be extensive, and to interconnect both many was not a protein central to the present study, there
of the fine membrane "puncta" and many of the cyto- was not a major effort to evaluate whether or not a
plasmic organelles. The extensive network of fine vinculin-like antigen could be isolated from the onion
tubules in a highly stressed cell was not usually visi- cells even though the occurrence of apparent

Fig. 2 A-E. Colocalization of injected integrin and spectrin antibodies. AI Represented in red is a Texas red-conjugated polyclonal antibody
prepared by Marcantonio and Hynes against a peptide corresponding to the well-conserved cytoplasmic domain of 131integrin, and in green is a
fluorescein-conjugated monoclonal antibody against the peptide sequence common to ct and 13 spectrin. The image was obtained with a X60,
1.4 NA objective lens, restored with ML, and projected to effect a stereo pair. A2 Section 1 (of 64 sections) in the raw image stack, transecting
the cell and hence cutting approximately perpendicular to the plasma membrane. A3 Section 1 after restoration of the image stack by ML. A4
Section 43 from the raw image stack, focused in the vicinity of the cortical face of the plasma membrane. A5 Section 43 after restoration of the
stack by ML. B-E Controls for A; recording parameters for all red and green images match the respective parameters for red and green images
in A. For the raw sections, display parameters correspond to those used in A2 and A4. For the processed images, processing and display param-
eters match those for A1 except that the projections in D3 and E3 are simple rather than stereo. B and C Raw optical sections of a cell inject
-ed with depleted Texas red-conjugated polyclonal anti-integrin and of a cell injected with depleted fluoreseein-conjugated monoclonal anti-
spectrin, respectively. The level of the sections in the cells is comparable to that in A2. D1-3 Depleted Texas red-labeled anti-integrin coinject-
ed with non-depleted fluorescein-labeled anti-spectrin. D1 Raw section of a representative stack photographed in red light and pseudocolored
red. D2 Same section recorded in green light and pseudocolored green; level of cell transection is comparable to that in A2. D3 Overlay of sim-
ple projections of the two stacks after restoration. E l - 3 Coinjection of two aliquots of non-immune IgG proteins, identical except for labeling
with either Texas red or fluoresceiu. A single section is shown raw in E1 (red pseudocolor) and E2 (green pseudocolor) as photographed, respec-
tively, with the same filters and exposure times used for anti-integrin and anti-spectrin in A1 and A4; the level of the section in the cell is com-
parable to that in A2. E3 Overlay of simple projections of restored green and red image stacks. Bars: 20 gm
c, Reuzeauet al.: Integrinand cytoskeletalproteins in onion cells 179

sequence similarity in the Arabidopsis database of volume ratio will be relatively high for extensive net-
expressed sequence tags suggests that such an effort works of thin tubules typical of stressed cells com-
might be successful, pared with simple cisternal sheets often seen in
Although the ER is more conspicuous in unstressed or unstressed cells.
mildly stressed cells than in highly stressed cells, it The remainder of the images will show covisualiza-
should not be concluded that it is less well represent- tions of applied fluorescent indicators of an integrin
ed in the latter. The surface area and the surface-to- and selected cytoskeletal elements in relation to the
180 C. Reuzeau et al.: Integrin and cytoskeletal proteins in onion cells

plasmalemma and the endomembrane system. Be- A priori, it is possible that when two antibodies, two
cause low stress and high stress bring out different stains, or paired antibody and stain are injected into
features, covisualizations at both levels of stress will the same cell, binding of one might alter the pattern of
enable a more thorough examination of the extent of antigen distribution recognized by the other. As a par-
association of the indicators and the membranes than tial check against this possibility, anti-spectrin was
would be possible by examination of only one state. injected at relatively high pressure without anti-inte-
grin or other fluorescent probe. Punctate patterns
similar to that of Fig. 2 A1 were observed (data not
[3; Integrin and spectrin are similarly distributed
shown), indicating that the pattern is not an artifact of
Occurrence and pattern of colocalization in heavily the coinjection.
stressed ceils
When the polyclonal antibody against the cytoplas- Overview of three types of controls
mic domain of ~1 integrin and the monoclonal anti-
[31 Integrin and spectrin are two proteins critical for
body against a common domain of c~ and [3 spectrin
the present study. Thus it is important to determine
are coinjected into cells of the inner epidermis of
whether the antibodies have recognized antigens in
onion bulb scales, their distributions overlap exten-
onions comparable to those animal antigens against
sively. Such colocalization is illustrated for a heavily
which they were prepared, and to this end three types
stressed cell in the stereo projection of Fig. 2 A1,
of control have been carried out. First, fluorescence-
where the anti-integrin is pseudocolored red, anti-
tagged non-immune IgG proteins have been injected
spectrin is pseudocolored green, and the color signi-
into onion cells. Second, onion proteins recognized
fying overlap ranges from yellow-green through
by the antibodies have been purified and electro-
bright yellow to orange depending on the respective
phoretically characterized. Third, cells have been
intensities of the green and red. All features of the
injected with preparations of ~1 integrin and spectrin
image appear in the overlap color range, though the
antibodies following depletion of their capacity for
relative contributions of red and green vary some-
specific antigen recognition. These three kinds of pro-
what. The stereo projection of the ML-restored data
cedures are described below.
of Fig. 2 A1 permits examination of the distribution
of the two antibodies throughout the sampled volume
of the onion cell cytoplasm, but as with all processed Immunoglobulin controls
images it is important to verify that the general fea- The distribution pattern of Fig. 2 A did not result
tures of the final product are observed in the raw from nonspecific recognition by immunoglobulins.
image. Thus, Fig. 2 A2 and A4 shows representative For this determination, one aliquot of purified nonim-
sections of the raw image stack, and these same sec- mune IgG proteins was fluorochromically tagged by
tions are shown in Fig. 2 A3 and A5 after ML restora- the same procedure used for the IgG antibody to ani-
tion of the stack. Essentially all the features of the mal integrin while another aliquot was tagged by the
processed data are evident in the raw data, but the procedure used for anti-animal spectrin, and these
light blurred by the objective lens has been restored, were injected into onion epidermal cells singly and in
to considerable extent, to its points of origin. A few combination. The data are represented in Fig. 2 E. A
irregular rings appearing in the vacuolar region of representative single section is shown unprocessed in
Fig. 2 A3 remain because the original blur cones from Fig. 2 El (red pseudocolor) and Fig. 2 E2 (green
the brightest objects in the stereo projection of Fig. pseudocolor) as photographed, respectively, with the
2 A1 have not been fully restored to their origins, and same filter systems and exposure times used for anti-
corresponding blur remains evident in the vicinity of integrin in Fig. 2 A2 and for anti-spectrin in Fig.
these bright points in Fig. 2 A1. 2 A4. The section transects the injected cell at a level
The polyclonal antibody against animal 131 integrin comparable with that of the section in Fig. 2 A2; also
and the monoclonal antibody against spectrin recog- it is displayed at the same intensity scales. The reason
nize a discrete pattern of puncta on the plasmalemma the red and green views are separated is that the latter
and small patches within the thin layer of cytoplasm happens to be much brighter and therefore obscures
in the heavily stressed onion epidermal cell of Fig. the former when overlaid. In Fig. 2 El, the cytoplasm
2 A, a pattern which has been reproduced in numer- is seen as a faint blur, with the region around the
ous experiments. nucleus (which is relatively distended with ER and
C. Reuzeau et al.: Integrin and cytoskeletal proteins in onion cells 181

organelles) showing more intensely. In Fig. 2 E2, the antibodies. In overview, purifications by extraction
cytoplasm and nucleus are a little brighter, and followed by separation with native PAGE and
modest fluorochrome aggregation is hinted along the Western blotting permitted recognition by polyclonal
lower edge of the cell. The faint diffuse staining of the anti-spectrin of one very conspicuous band and a
cytoplasm along the wall and stronger but nonetheless second smaller but distinct band plus some very faint
diffuse staining of the perinuclear cytoplasm con- bands. Extracts were further purified using beads
firms that the injected IgG was successfully targeted coated with monoclonal anti-spectrin and the product
to the cytoplasm, and indeed it is to present this case was separated by SDS-PAGE. Western blotting then
that an image containing the nucleus was selected. permitted recognition by the monoclonal anti-spectrin
However, unlike the cytoplasm of the vivid image of two distinct bands, one relatively large and one
section of Fig. 2 A2, which contains both specific relatively small, plus some very faint bands.
structures in the plane of focus and out-of-focus light In more detail, purification by extraction was carried
from such structures in other sections, the cytoplasm out using either PBS or Hepes buffer. Western blots of
of the IgG controls in Fig. 2 E1 and E2 is essentially the extracted proteins, separated by native PAGE and
featureless. Thus, the injected fluorochromically identified with polyclonal anti-human spectrin, are
tagged control IgG has diffused randomly through the shown for PBS and for Hepes. Clearly, the Hepes
bulk of the cytoplasm but has failed to recognize par- method was the more successful (Fig. 3 A, lane 2). A
ticular structures except, possibly, around the nucleus large band is seen at a position corresponding to mol-
(where the layer of cytoplasm is typically thick and ecular mass of 210-230 kDa, a small sharply-defined
densely filled with organelles). This conclusion is band occurs at a position corresponding to about
reinforced by the image in Fig. 2 E3, an overlay in 580 kDa, and a blurry band occurs in the vicinity of
which the green and red image stacks were processed, the 380 kDa region. These molecular mass marker
projected, and displayed identically to those of Fig. comparisons serve only to identify the positions of
2 A but which nevertheless shows no discrete features bands in the experimental lanes, of course, since elec-
except for a shell of stain around the nucleus and a trophoretic separation on native gels is not simply
very few spots at the level of the section in Fig. 2 E1 correlated with molecular mass. The PBS extract
and E2. (Fig. 3 A, lane 1) does not show the large band in the
As a further control, it was checked that injected 210-230 kDa marker zone, but does show a distinct
polyclonal anti-spectrin identified patterns closely band at the 580 kDa marker site and also shows a
similar to those identified by the monoclonal anti- faint smear at the 380 kDa marker zone. The faint
spectrin (data for punctate array not shown, but see broad bands in the 380 kDa region of both lanes may
Fig. 5 B). possibly represent a species of spectrin, or may repre-
sent degradation products of the sharp bands at the
580 kDa marker site. In lanes 3 and 4 Fig. 3 A shows
Partial characterization of proteins controls with PBS and Hepes extracts treated in the
In a companion paper Gens et al. (1996) have already same way as lanes 1 and 2 except that the primary,
described how purification of the integrin-like anti- polyclonal, antibody against spectrin was omitted.
genicity was accomplished by reaction of onion Thus, the antibody recognition of bands in lanes 1 and
extract with monoclonal antibody against the domain 2 is specifically due to the polyclonal anti-human
of 13~ integrin which, in focal adhesions of animal spectrin rather than to a miscellaneous reaction by the
cells, protrudes into the extracellular matrix. Figure 2 secondary antibody.
of that paper shows Western blots with two bands The large band at the 210-230 kDa position in lane 2
(105-110 and 115-125 kDa) of immunoprecipitated of Fig. 3 A was not evidenced when replicate blots
protein that react again with the precipitating anti- were tested with the monoclonal anti-human spectrin,
body but are also recognized by the antibody of Mar- nor were the smaller bands (data not shown). Howev-
cantonio and Hynes prepared against the cytoplasmic er, a more searching test for antigenicity was carried
domain of {31 integrin. Gens et al. (1996) concluded out as follows. The extracts were further purified with
that the bands from onion cells very likely represent affinity beads constructed with the monoclonal anti-
protein homologous to 131 integrin from animal cells. human spectrin, the product was separated by SDS-
In the present paper, spectrin-like antigenicity has PAGE, gets were blotted, and blots were assayed with
been identified in onion extract using two different the monoclonal anti-spectrin. Results are shown in
182 C. Reuzeau et al.: Integrin and cytoskeletal proteins in onion cells

Fig. 3. Western blots of electrophoretic gels of spectrin antigenicity from onion epidermal cells, identified with anti-spectrin followed by second-
ary antibody and alkaline phosphatase color assay. A Recognition by polyclonal anti-human spectrin on blot of native PAGE (6% gel) of total
cell extracts. 1 and 2 Crude PBS and Hepes extracts, respectively. On the left, specially selected molecular mass markers indicate position of
antigens, recognizing that, as always in native gels, the migration of experimental proteins is not monotonically governed by molecular mass.
Distinct bands occur in both lanes at about the 580 kDa marker zone, and additionally a heavy band occurs in 2 at about the 210-230 kDa mark-
er zone. 3 and 4 Crude PBS and Hepes extract controls, respectively, with secondary antibody and color assay alone. B Recognition by mono-
clonal anti-spectrin on blot of SDS-PAGE (10% gel) of material separated from total cell extracts by monoclonal anti-spectrin beads. 1 Control,
with product from affinity beads provided only with buffer. A dark band of antibody occurs between molecular mass markers 53.2 and 84, and
some material of lower molecular mass is seen. 2 Product from affinity beads provided with PBS extract. In addition to the antibody bands dis-
played in the control lane, distinct antigenicity recognized by the monoclonal anti-spectrin occurs at about 100 kDa and less distinct antigenic-
ity at about 150 kDa. Tiny hints of antigenicity occur in the original blot per se at somewhat higher (ca. 170 kDa) and lower (ca. 90 kDa) mo-
lecular mass. 3 Product from affinity beads provided with Hepes extract. In addition to the antibodies of the control lane, faint antigenicity
occurs at about 100 kDa and very faint antigenicity is seen in the original blot per se at about 150 kDa. 4 and 5 No antigen was detected when
PBS and Hepes extracts, respectively, were reacted with beads to which nonimmune mouse IgG had been bound

Fig. 3 B. L a n e 1 s h o w s c o n t r o l p r o d u c t f r o m the r e c o g n i z e d b y the m o n o c l o n a l a n t i - s p e c t r i n on a

b e a d s alone, with a large b a n d at a b o u t 60 k D a f r o m W e s t e r n b l o t f o l l o w i n g S D S - P A G E separation. A t -
the b e a d a n t i b o d y i t s e l f a n d a s m a l l c o n t r i b u t i o n at 23 testing to the s p e c i f i c i t y o f these antigens, c o n t r o l
k D a l i k e l y due either to the light c h a i n o f the b e a d l a n e s 4 and 5 (Fig. 3 B) e v i d e n c e d no b a n d s w h e n
a n t i b o d y or to d e g r a d a t i o n p r o d u c t s o f the b e a d anti- n o n i m m u n e m o u s e I g G was b o u n d to b e a d s and P B S
body. A v e r y faint l i k e l y d e g r a d a t i o n p r o d u c t , or p o s - and H e p e s extracts were r e s p e c t i v e l y m i x e d with the
s i b l y s o m e o f the light c h a i n o f I g G , is also p r e s e n t at b e a d s , w a s h e d , r e l e a s e d , e l e c t r o p h o r e t i c a l l y separat-
a b o u t 36 kDa. L a n e 2 s h o w s the p r o d u c t f r o m affinity ed, blotted, and tested as for lanes 2 and 3.
b e a d s c o m b i n e d w i t h P B S extract, w h i c h in a d d i t i o n Clearly, then, the a n t i b o d i e s a g a i n s t a n i m a l integrin
to the c o n t r o l b a n d s s h o w s a d i s t i n c t b a n d at a b o u t and spectrin are a b l e to r e c o g n i z e a set o f d i s t i n c t i v e
100 k D a a n d a light b a n d at a b o u t 130 kDa. L a n e 3, a n t i g e n s e x t r a c t e d f r o m o n i o n cells. A s a s t r o n g l y
the affinity p r o d u c t f r o m H e p e s extract, s h o w s a b a n d s u p p o r t e d w o r k i n g h y p o t h e s i s , n o t o n l y the anti-{31
at 100 k D a , and the o r i g i n a l blots s h o w e d a faint b a n d integrin but also the anti-(cz and ~) spectrin r e c o g n i z e
at 130 k D a w h i c h d i d not p h o t o g r a p h well. ( A d d i t i o n - s p e c i f i c o n i o n c o u n t e r p a r t s to antigens o f the a n i m a l
ally, in the o r i g i n a l W e s t e r n blots, lanes 2 and 3 b o t h families.
h a d d i s c e r n a b l e b a n d s at 170 k D a , but these w e r e t o o
faint to p h o t o g r a p h . ) Thus, at l e a s t t w o p r o t e i n s pres- Depleted antibody controls
ent in b o t h the P B S and H e p e s extracts can be affini- Cells w e r e i n j e c t e d w i t h the d e p l e t e d a n t i - i n t e g r i n
t y - p u r i f i e d with m o n o c l o n a l a n t i - s p e c t r i n b e a d s a n d (Fig. 2 B) and d e p l e t e d a n t i - s p e c t r i n (Fig. 2 C).
C. Reuzeau et al.: Integrin and cytoskeletal proteins in onion cells 183

Fig. 4 A, B. Stereo views of a cell injected with Texas red-conjugated anti-[31 integrin at moderate pressure. The images were obtained with a
• 1.4 NA objective lens, restored by ML, and shown with intensity encoded in a nonlinear yellow scale. In order to represent well both the
nucleus and the plasmalemmal puncta and endomembranes best seen near the periclinal wall of the cell, two overlapping 64-section image
stacks were taken, one at a median depth (A) and one at a shallow depth (B). A Nucleus, which is surrounded by a sheet of ER; and organelles
and patches of lamellae are abundant in the cytoplasm pooled at the tapering end of the cell. B Portion of the cell nearer the inner periclinal wall,
with puncta near the cell membrane conspicuous. Bar: 20 ~tm

Panels 2 B and 2 C were prepared with parameters centrated in a typical punctate array. Illustrating this,
identical to those for the integrin and spectrin anti- Fig. 2 D1 shows one section of a representative stack
b o d y image shown in Fig. 2 A2, and illustrate raw with depleted Texas-red labeled anti-integrin photo-
optical sections at approximately the same level of graphed with a red filter system and p s e u d o c o l o r e d
cell transection as in that panel. Insofar as these and red, and Fig. 2 D2 shows the same section recorded in
the remaining sections o f the two stacks were uni- green light emitted by fluorescein-tagged non-deplet-
formly dark, the fluorochromically tagged antibodies ed anti-spectrin and p s e u d o c o l o r e d accordingly; the
which identified structures in the cytoplasm of Fig. level of cell transection is comparable to that of Fig. 2
2 A 1 were evidently relatively specific for their A3. Whereas the red photograph has only a faint, dim
intended antigens. glow, the green one evidences extensive structure.
Additionally, to check that there were structures sus- W h e n the two stacks were processed in the same way
ceptible o f recognition in the cells injected with anti- as for Fig. 2 A1, projected, and overlaid (Fig. 2 D3),
body-depleted solution, depleted anti-integrin was there was an array of punctate objects in green but
coinjected with non-depleted anti-spectrin. As antici- very few of these were colocalized with red objects.
pated, the depleted solution dimly tagged at m o s t a Thus, not only are antigens recognized by antibodies
very few spots in the cell, whereas the antibody con- against animal ~1 integrin and spectrin similarly dis-
184 C. Reuzeau et al.: Integrin and cytoskeletal proteins in onion cells

tributed in the interior of onion cells, but also it is with ER in a moderate-stress conformation, with
plausible that the antigens are in fact homologs of ani- numerous organelles visible, and with many plasma-
mal [3~ integrin and spectrin. The next set of experi- lemma-associated puncta in view as well. The stereo
ments asks with what cellular feature or features the pair of Fig. 4 A, projected from a stack of sections
spots containing the antigens may be identified. taken near the midplane of the cell, shows a thin shell
of stain around the nucleus. Such a perinuclear stain-
lntegrin and spectrin colocalize with ing pattern is highly characteristic of the ER, which
the endomembrane system generally wraps the nucleus in one to several lamellar
layers. In the large pool of cytoplasm near the pointed
The anti-J31 integrin and anti-spectrin of Fig. 2 A1 are
tip of the cell are many bodies which are undoubtedly
codistributed as tiny spots around the periphery of the
organelles, some perhaps associated with small irreg-
cell and as larger bodies in the thin layer of cytoplasm
ularly formed patches of lamellar ER. Figure 4 B is
between the plasma membrane and vacuole mem-
from an overlapping stack photographed with the
brane. Most of the tiny spots appear as if intimately
same X, Y coordinates but lower on the Z axis. Where
associated with the plasma membrane. This appear-
the cytoplasmic pool overlaps with Fig. 4 A, numer-
ance is particularly strong where the plasma mem-
ous organelles are visible (some redundantly). Addi-
brane is transected at the top of the image stack.
tionally, and importantly, where a good view of the
Because the cytoplasmic layer and its endomem-
periclinal wall is permitted there are many tiny punc-
branes can be thin, in images such as that of Fig. 2 A1
ta at the surface of the cell membrane.
it is not possible to discriminate whether spots close
Figure 5A illustrates injected spectrin antibody asso-
to the periphery are (1) directly associated with the
ciated with a large-meshed reticulum that is readily
plasma membrane, (2) associated with endomem-
recognized as ER. Five sections are presented from
brane linked with the plasma membrane, or (3) asso-
the restored stack (Fig. 5 A2-6). By comparing these
ciated with endomembrane nestled close to the
sections, it can be seen that the entire sloping plasma
plasma membrane. However, further experimental
membrane of the cell, which is photographed near an
approaches contribute relevant information.
acute end, is underlain by the reticulum.
The anti-spectrin mesh visualized in Fig. 5 A has
Integrin and spectrin in mildly stressed cells large polygonal open areas, with relatively thick diag-
In some injected cells, polyclonal anti-J31 integrin and onals ranging up to about 8 ~tm. However, finer anti-
monoclonal anti-spectrin assumed patterns plainly spectrin mesh was also seen. For example, Fig. 5 B
characteristic of the endomembrane system of mildly shows an extremely delicate cortical mesh with barely
stressed cells. The occurrence of such patterns was resolved tubules and a wide range of mesh diagonals.
correlated with a relatively low injection pressure. Figure 5 C shows a section from another image with a
Figure 4 illustrates integrin antibody injected in a cell fine-meshed reticulum, in this case identified by

Fig. 5 A-E. Reticular displays of injected anti-spectrins, and covisualizations. A Injected Texas red-labeled monoclonal anti-spectrin identifies
a reticulum of relatively large mesh size in a "low stress" cell. Photographed with a • 100, 1.3 NA objective lens. A1 Section number 45 in a
64-section raw image stack; intensity is encoded by a nonlinear yellow scale as indicated. A2-6 Sections 18, 31, 45, 51, and 56; the color bar is
in A6. Because standard ML tends to emphasize bright points along lines of varying intensity rather than continuity of the lines, the stack has
been restored with a regularized, intensity-penalized ML, followed by GF to enhance edges and contrast. The continuity of the mesh is readily
evident by comparing these sections; it is visualized even better by sequential viewing of the 64 planes of the image stack, and its contour can
be visualized by stereo projection. However, the out-of-focus light in these images is far from fully restored to its points of origin; and the rea-
son can be understood by examining the amount and distribution of blur in the sample raw section AI. It is because such extensive blur, typical
in the vicinity of large bright areas, interfered with projections that the data are presented simply as a panel of sections. B A finer reticulum is
identified by monoclonal anti-spectrin in a "high stress" cell, with numerous vesicles and puncta interconnected by fine tubules. C Polyclonal
anti-spectrin also identifies a reticulum; in this cell it is of relatively fine mesh size. For both B and C a • 100, 1.3 NA objective lens was used;
for B a single section of an ML-restored stack is shown after GF, and for C, because of high background blur, a single section has been pro-
cessed only by GF. Color bars provide intensity scales. D Covisualization of spectrin antigenicity with DiOC6, marker for the endomembrane
system. Stereo projection of ML-restored image obtained with a • 1.4 NA objective lens. In this "high stress" cell, the injected monoclonal
anti-spectrin (pseudocolored red) and exogenously applied DiOC6 (pseudocolored green) show extensive overlap (yellow) in staining the endo-
membrane system, which appears punctate at this magnification. E A stereo pair showing overlap of injected anti-spectrin (green) with anti-vin-
culin (red); ?440, 1.0 NA objective lens. Bars: 20 p~m
C. Reuzeau et al.: Integrin and cytoskeletal proteins in onion cells 185
186 C. Reuzeau et al.: Integrin and cytoskeletal proteins in onion cells

injected polyclonal anti-spectrin. Here, the relatively occurring or has occurred recently; limited streaming
regular polygonal open areas tend to have maximal was sometimes imaged despite efforts to avoid it.
diagonals of about 4 ~tm. As for the coarser mesh of The resemblance of various images of cells injected
Fig. 5 A, the finer mesh patterns of Fig. 5 B and C are with rhodamine phalloidin alone to those labeled with
highly characteristic of the ER. integrin and spectrin antibodies and with DiOC6 was
reinforced by covisualizations. Figure 6 D and C
Covisualization of anti-spectrin and endomembranes exemplifies, respectively, overlaps with anti-spectrin
In general, then, the punctate and reticular patterns and DiOC6. The cell in Fig. 6 D was photographed
visualized with injected anti-integrin and anti-spec- again with a higher-magnification objective lens as
trin resemble those seen (cf. Fig. 1) with applied shown in Fig. 6 B 1, with an enlarged detail in Fig.
DiOC6. In order to confirm that the arrays of individ- 6 B2, in order to emphasize cotocalization on barely
ually injected integrin and spectrin antibodies such as resolved tubules, small vesicles, and puncta or par-
illustrated in Figs. 4 and 5 A-C represent the distrib- ticles associated with them. Such vesicles with punc-
ution of the endomembrane system, Texas red-conju- ta were not always present, but (cf. Fig. 7 A, C) were
gated monoclonal anti-spectrin was injected into cells seen stained with several of the antibodies utilized.
of tissue treated with DiOC6. In Fig. 5 D, a represen-
tative stereo projection with stains represented by red lntermediate filaments coat the endomembranes
and green pseudocolor shows extensive overlap of the Colocalization with endomembranes
two fluorochromes. (It is far greater than appears at Demonstration of intermediate filaments in various
first glance; note that most of the green spots have a plants (McNulty and Saunders 1992, Yang et al. 1992,
yellow tinge and a central yellow region. Separate Fairbairn etal. 1994, Mizuno 1995) obviates an
projections not here shown enabled a more searching urgent need to demonstrate intermediate-filament
pixel-by-pixel comparison.) antigenicity in onion extract, and, moreover, the anti-
body chosen for covisualization is well documented
F-actin "coats" the endomembrane system (Pruss et al. 1981; compare McNulty and Saunders
Rhodamine phalloidin, a stain specific for F-actin 1992) as specifically identifying a highly conserved
(Haugland 1992), was injected (cf. Schmit and Lam- region in animal intermediate-filament proteins. Fig-
bert 1990, Cleary et al. 1992, Cleary 1995; cf. Traas ure 7 E and the higher-magnification image in Fig.
et al. 1987, Liebe and Menzel 1995) into a number of 7 A show that injected antibody can recognize protein
cells to compare F-actin distribution with the distri- in plant cytoplasm, and that it colocalizes extensively
bution patterns of integrin, spectrin, and the endo- with plasmalemma-associated puncta and with the
membrane system. This stain labeled meshes of a various tubules, lamellae, and organelles of the endo-
wide range of sizes as well as endomembranes in membrane system.
other configurations. An example of a fine mesh is
shown in Fig. 6 A, in which a single section captured Localization within the nucleus
an extensive portion of the periphery of an unusually Because plant intermediate filaments have been at
flat cell. In addition to the fine cortical mesh, bright least as extensively studied in the nucleus (McNulty
spots are present; examination of the entire stack of and Saunders 1992) as in the cytoplasm, Fig. 7 B1-5
sections, particularly when rotated 90 ~ for inspection is presented to verify that the antibody recognizes
along X, Z planes (data not shown), indicates that both the periphery of the nucleus and structures in the
these bodies occur deeper in the cytoplasm, consistent interior. In particular, a novel (cf. McNulty and Saun-
with the interpretation that these are organelles such ders 1992) feature of this nuclear image is that the
as mitochondria. Indeed, the shape and size of these transcription zone (Wachtler and Stahl 1993) of each
bodies is comparable with those stained in separate of the two nucleoli stains brightly, whereas the pro-
experiments with dilute rhodamine 123, known as a cessing zone (ibid.) is free of stain. Small bright
stain with preference for mitochondria (data not spots, fibers, and loosely-bounded regions are stained
shown). Mitochondria and other organelles are alrea- in the main body of the nucleus as well, and these are
dy known to associate with F-actin (e.g., Lichtscheidl all the more conspicuous because they contrast with
et al. 1990). The bright streaks in Fig. 6 A are small very dark zones or interstices. The cytoplasmic face
regions where organelle movement or streaming is of the nuclear membrane may possibly be well sup-
C. Reuzeau et al.: Integrin and cytoskeletal proteins in onion cells 187

Fig. 6 A-D. F-actin. A A single section showing reticular distribution of F-actin, labeled with injected rhodamine phalloidin; fine streaming
strands and F-actin-coated organelles are also present. • 100, 1.3 NA objective. The color scale encodes intensity. B Cell photographed with a
• 100, 1.3 NA objective lens. In B1 (enlarged detail in B2), small "puncta" can be distinguished from larger, less regular forms (evidently
organelles and associated endomembrane) and from little vacuoles. The latter are rimmed with small puncta. When all the sections of the raw
stack or the restored stack are viewed sequentially, fine tubules seem to connect some of the other features, but due to the occurrence of noise
along the voxel rays of this simple projection of the restored stack, only a hint of these tubules remains. C Use of a • 60, 1.4 NA objective lens
shows a greater portion of the cell of B, emphasizing colocalization of the stains, but detail is diminished and little can be discerned except
numerous small spots, larger more amorphous masses, and a shell of endoplasmic reticulum surrounding the nucleus. (Because of its low mo-
lecular weight, some phalloidin has leaked into adjacent cells; DiOC6 of course stains the entire tissue.) D Colocalization (yellow) of injected
rhodamine phalloidin (red) and anti-spectrin (green); • 1.4 NA objective lens. Small puncta in varying shades of yellow and ovoid
organelles in bright yellow are scattered around the periphery of this "high stress" cell. Bars: 20 ~xm
188 c. Reuzeau et al.: Integrin and cytoskeletal proteins in onion cells

plied with intermediate filaments, as expected is shown in the representative images of Figs. 1 F and
(McNulty and Saunders 1992), but because interme- 5 E. As already remarked for Fig. 1 F, this image well
diate filaments clad the ER (Fig. 7 A, E) and the ER illustrates the occurrence of scattered small, uniform,
forms a shell around the nucleus, much of the anti- hollow vesicles that might be Gotgi (Fig. 1 F). While
body label at the nuclear periphery is doubtless on the the abundance of these vesicles in onion epidermis, a
ER. storage tissue, m a y often be low, it is likely that only
with strong staining such as achieved here with anti-
Two other endomembrane-associated antigens vinculin can the hollowness of the vesicles be docu-
Talin and vinculin seem not to have been described mented under the utilized conditions of microscopy.
previously in plants, but similar partial sequences Another feature exemplified in an anti-vinculin image
occur in the Arabidopsis database of expressed (Fig. 5 E) is an arrangement of fluorochrome-labeled
sequence tags. It is of interest to look for these pro- puncta in polygons or "rings" associated with the
teins because in animal cells they can be associated plasma membrane. These are clearly distinguishable
with integrin and actin at adhesion sites (e.g., Gumbi- from spherical vesicles or vacuoles. Such rings were
ner 1993), and, indeed, talin m a y possibly form an seen in cells treated with a variety of stains and anti-
indirect connection between integrins and actin fila- bodies, being c o m m o n and sometimes very large in
ments via vinculin (Gumbiner 1993). cells which had dried a little during a long photogra-
phic session or which had been subject to bathing
Talin medium of relatively high osmotic strength (data not
shown). In the latter case, particularly when DiOC6 or
As a first indication of talin antigenicity in a plant, the
anti-spectrin was provided in excess so that the posi-
distribution of injected Texas red-labeled anti-animal
tion of the plasma membrane was clearly evident, it
talin is shown in representative images in Fig. 7 C and
was sometimes seen that the cell m e m b r a n e had
D. In the former it tends to colocalize with DiOC6,
pulled away from the plasma membrane locally but
and in the latter with anti-spectrin.
had remained tethered at the spots. Rings of compara-
It will be essential to perform an extensive series of
ble diameter were sometimes seen at the surface of
controls to check for specificity of recognition. One
protoplasts treated with anti-j31 integrin and other
confirmatory control already accomplished is the
antibodies to reveal putative adhesion sites (Gens
injection of I g G serum, as depicted in Fig. 2 E. A
et al. 1996: fig. 4 d); here it was suggested that puta-
likelihood that antigen could be isolated and charac-
tive links connecting adhesion sites had broken, with
terized is indicated by the reaction of the monoclonal
the tensions due to remaining links tugging the neigh-
antibody with dot blots from onion extract (data not
boring spots into a relatively stable ring.
shown). Clearly, description of appropriate antigenic-
The injected I g G control in Fig. 2 E suggests that fur-
ity is an important desideratum.
ther checks for possible specificity of recognition
would be rewarding. Attempts to isolate the protein
Vinculin might also be worth while, as vinculin-like antigenic-
Colocalization of injected anti-animal vinculin with ity has been seen in dot blots of onion extract (data
plasmalemmal puncta and the endomembrane system not shown).

Fig. 7 A-It. Antibodies against intermediate filaments and talin, plus general controls. A Texas red-labeled antibody against intermediate fila-
ments covisualized with DiOC6 in a "high stress" cell; • 100, 1.3 NA objective lens. BI-B5 Selected sections (18, 31, 40, 47, 62) transecting
the nucleus shown in A, but from a separate stack of sections higher in the cell. The color scale is the same as for A except for expansionto bring
out nuclear structure. C Texas red-conjugated anti-talin and DiOC6, showing extensive overlap; • 100, 1.3 NA lens. (The red color of anti-talin
tends to predominate on the tiny puncta and on some of the larger forms, but it is unknown whether this is reproducible.) D Anti-talin and anti-
spectrin, "high stress" cell in stereo view, • 1.4 NA objective lens. E Antiqntermediate filaments and DiOC6; stereo view, • 1.4 NA
objective lens. A nucleus is visible, and some patches of endomembraneare relatively large, as is typical of moderatelystressed cells. Both D
and E show extensive stain overlap. F Injected Cascade blue, a hydrazide. GI Injected Lucifer yellow, a hydrazide; G2 same image stack but
photographed to record rhodamine coinjected with dextran. H Cascade blue coinjected with dextran (HI) and rhodamine coinjected with phal-
loidin (H2). Bar for G1-H2 in H2. Bars: 20 ~tm
C. Reuzeau et al.: Integrin and cytoskeletal proteins in onion cells 189
190 C. Reuzeauet al.: Integrin and cytoskeletalproteins in onion cells

Minimal staining of the plasma membrane tant large neutral fluorescent compound should
Spectrin and associated proteins have been extensive- remain soluble in the cytosol. When injected into the
ly studied in animal cells as molecules which under- cytoplasm, Cascade blue dextran failed to identify the
coat the cell membrane (e.g., Bennett and Gilligan endomembranes, causing the cytoplasm to fluoresce
1993). In some onion cell images the transected posi- with little hint of subcellular features (Fig. 7 H1)
tion of the plasma membrane was indicated by a faint except for the central vacuole and occasional lesser
thin line, and when present in covisualizations such a vacuoles (generally discernable only by studying an
line was generally seen with both fluorochromes. image stack section by section).
Figures 2 A, 5 E, and, especially, 6 C illustrate such Comparable demonstrations were performed with
staining. Such images raise the possibility that rho- 3 kDa dextran tagged with rhodamine. Figure 7 G2 is
damine phalloidin and the fluorochrome-labeled anti- a section from an image stack that illustrates the dif-
bodies might lightly but specifically stain the plasma fuse fluorescence emitted from the cytoplasm of cells
membrane over its entire surface. However, although injected with this dextran.
it is important to emphasize that the possibility is not The validity of these diffuse images of fluorochrome-
excluded, it is equally important to appreciate that it tagged dextran is emphasized by examining paired
is not proven. Aromatic substances such as the fluo- images of protein stains that were coinjected. For
rochromes tend to be soluble in membranes, and for example, Fig. 7 G1 and G2 shows an identical section
this reason not only simple fluorochromic stains but viewed with filter systems for injected Lucifer yellow
also fluorochrome-labeled antibodies might associate hydrazide and rhodamine dextran, respectively, and
to some extent with the plasma membrane. It is also Fig. 7 H1 and H2 shows an identical section with
possible that refraction and optical piping of light by Cascade blue dextran and rhodamine phalloidin.
the celt wall might create an appearance of plasma Clearly, the failure of the neutral dextran conjugates
membrane labeling, although there is no obvious rea- t o identify the patterns of the endomembrane system
son why such physical effects would not light up the was not due to any impropriety of imaging, for the
entire thickness rather than just the inner surface of paired protein stains show the endomembrane system
the wall. in characteristic fashion.

Considerable cytoplasmic protein may associate

with membranes Discussion
The apparent association of several cytoskeletal ele- The codistributed proteins
ments with the cytoplasmic face of the endomem- Integrin and spectrin in onion cells
brahe system provokes the question: might a consi- Evidences from protein isolation and from micros-
derable fraction of "cytoplasmic" protein be associat- copy combine to suggest that ~ integrin and spectrin
ed with membranes? To provide an answer, two flu- not only exist in mature onion bulb scale epidermal
orescent hydrazides traditionally used for wide- cells, but also are situated in fairly close proximity to
spread, indiscriminate tagging of protein (Haugland each other.
1992) were injected into onion cytoplasm. Figure 7 F With respect to ~ integrin antigenicity in onion cells,
and G1 illustrates that these two hydrazide stains, the double-identification of a conspicuous band on
Cascade blue and Lucifer yellow, "light up" the mem- Western blots by two antibodies against domains
brane/endomembrane system. known for the animal protein to be situated outside
and inside the plasma membrane, respectively, is con-
Visualization of space not occupied by sistent with more limited findings for other plant cells
the endomembrane system (Gens et al. 1996). Apparent molecular weights for
Considering the number of antibodies and protein the integrins depend on redox state as well as on
stains labeling the endomembrane system, it is impor- extent of glycosylation, and thus the only meaning
tant to perform controls addressing whether apparent that can be attached to molecular weight at this time is
detection of the endomembranes might be an artifact the electrophoretic migration of the band into an
of the imaging methodology. The reactivity of Cas- approximately reasonable (Cheresh and Mecham
cade blue with protein can be eliminated by linkage 1994) size zone. Since the well conserved ~l integrin
via the hydrazide group to 3 kDa dextran. The resul- is never found in animal cells except as a heterodimer
C. Reuzeau et al.: Integrin and cytoskeletal proteins in onion cells 191

with another, less well conserved, integrin (Cheresh As deduced from comparisons with DiOC6 staining,
and Mecham 1994), the putative occurrence of a the second category of localization within onion cells
integrin in onions implies the occurrence of at least is on the ER and all the major organelles of the cyto-
one companion integrin. Thus, understanding of plant plasm. Indeed, by far the greatest amounts of both
integrins is very tentative: other integrins should be putative ~ integrin and putative spectrin occur in and
sought, and a full description of genes and proteins of on the endomembrane system. Because the endo-
the integrin family is badly needed. membrane stain DiOC6 stains the plasmalemmal
With respect to the at least three antigens apparently puncta, perhaps even here at the face of adhesion sites
representing the spectrin family in onions, three sub- putative spectrin and putative integrin may be associ-
stantial bands on Western blots are identified by poly- ated with tiny tubules or cisternae of ER. Staining of
clonal antibodies against animal protein, and two are membranes by DiOC6 is considered (e.g., Haugland
recognized by the monoclonal antibody. It is notable 1992, Oparka and Read 1994; but compare Terasaki
that in animals the spectrin family has numerous and Reese 1992) to depend on the voltage drop across
representatives (e.g., Bennett and Gilligan 1993, them; given the likelihood that the several kinds of
Hartwig 1994). Spectrin antigenicity has already been membranes support different voltage drops under dif-
reported for Western blots from other plant extracts ferent levels of stress (but regarding ER, see Li et al.
(Wang and Yan 1988, 1991; Michaud et al. 1991; de 1995), it is surprising what extensive colocalization
Ruijter and Emons 1993; Faraday and Spanswick with anti-spectrin is seen not only in the images pre-
1993; Sikorski et al. 1993), though generally without sented in the figures but also in the collection of
extensive controls. The molecular size of c~ and ~ ani- images they represent. Perhaps variation in intensity
mal spectrins is typically about 220-280 kDa (Hart- of DiOC6 staining is usually within the range of
wig 1994), and this is not far from the size of two detection and would be more apparent with careful
bands isolated by virtue of affinity for the monoclon- quantitation, perhaps the variability that is seen does
al anti-human (c~ and ~) spectrin or of one of the indeed represent electrophysiological states, perhaps
bands identified by the polyclonal antibody (Fig. 3). the DiOC6 is at a saturating level in which variable
As for the putative plant integrin family, genes and susceptibility of the membranes is not conspicuous,
proteins of the putative plant spectrin family deserve or perhaps adherance of surface cytoskeleton paral-
description. lels susceptibility of membranes to DiOC6 staining
Antibodies to spectrin and to the cytoplasmic domain Whatever the explanation of this mystery may prove
of ~1 integrin injected into onion cells colocalize to be, it will modulate rather than negate the rather
extensively, within the resolution of the visualization general conclusions of this paper.
methods utilized, in two kinds of places. The first is at For animal cells, [31 integrin has been most intensive-
puncta on the inner face of the plasma membrane. It ly studied in its role as a characteristic protein of
may be supposed that these puncta correspond to various kinds of cell adhesion sites. Nevertheless, the
those visualized at the outer face of the plasma mem- bulk of ~1 integrin in animal cells has for some years
brane when antibody against the external domain of been known to be situated in the endoplasmic reticu-
~1 integrin is applied to onion protoplasts (Gens et al. lum with the cytoplasmic domain correctly positioned
1996). Similar puncta have been visualized recently (Eugene Marcantonio pets. comm.), though little has
on the outer surface of protoplasts from BY2 tobacco been published on this topic (see, however, Pardi
cell cultures (Gens et al. 1997), and what appear to be et al. 1995).
large "caps" of such puncta (possibly aggregated on In animal cells, c~ and ~ spectrin are well known to
stressed cells) have been seen earlier (Schindler et al. join with other cytoskeletal proteins to form a rein-
1989). Accumulating evidence (e.g., Wayne et al. forcing mesh under the cell membrane (e.g., Bennett
1992, Pont-Lezica et al. 1993, Pickard 1994) makes it and Gilligan 1993). It is quite possible that putative
reasonable to assume that these puncta represent spectrin antigenicity is similarly distributed at the
adhesion sites comparable to those of animal cells inside face of the plasma membrane of onion epider-
(e.g., Geiger and Ayalon 1992, Gumbiner 1993, mal cells, but if so it is either an isoform not well rec-
Jockusch et al. 1995). Less prejudicially with respect ognized by the antibody we have injected or else its
to presumed homology, we have named such sites distribution is less prominent than at adhesion sites
plasmalemmal control centers (Pickard and Ding and on the cytoplasmic face of the endomembrane
1993, Pickard 1994). system. Previous visualizations by de Ruijter and
192 c. Reuzeauet al.: Integrin and cytoskeletalproteins in onion cells

Emons (1993) for several plant cells of fluorochrome- arrays of a myosin seen by Miller et al. (1995). The
tagged anti-animal spectrin in the periphery of the cell apparent close association of F-actin with the cell
have been considered to indicate webbing under the membrane is also usefully compared with the associa-
plasma membrane comparable to that of animal cells. tion seen by A. L. Cleary during plasmolysis of Tra-
(Controls for this report consisted of treatments with descantia cells (Gunning and Steer 1996: plate 16).
non-immiJne serum.) However, in the fixed cells and Colocalization of actin with the entire endomembrane
at the resolution employed, the abundant label could system in onion epidermal cells may be viewed as
equally well have been in plasmalemmal puncta or on consistent with a variety of emerging evidences and
the endomembranes, consistent with our observations interpretations in literature. At the least, essentially
in onion. Bands of spectrin antigenicity were found all authors have considered that actin associates
by Faraday and Spanswick (1993) following gel elec- closely with the ER, an interpretation reviewed and
trophoresis of extract from highly purified rice plas- championed by Hepler et al. (1990) and more recent-
ma membrane vesicles; of course, the extent to which ly presented by Gunning and Steer (1996).
this association with the plasma membrane was dis- However, at first view some aspects of literature may
tributive or punctate cannot be determined with this seem slightly discordant with the present findings. It
method without further analysis. is a common interpretation that F-actin occurs as sim-
Additionally, in some animal cells, [3 spectrin anti- ple microfilaments or bundles of them, more or less
genicity has been found associated with the Golgi independent of the ER but interacting with it. This
compartment of the endomembrane system (Beck interpretation is based on a variety of kinds of obser-
et al. 1994). This may provide a parallel with the vations. Using enhanced video interference contrast
onion cells. In the work of de Ruijter and Emons microscopy, Allen and Brown (1988) occasionally
(1993) on plant cells, organelles thought to be plastids viewed fibrous bundles in unstained living cells, iden-
were labeled in some cell types, as were "nuclei" and, tifying them as actin by virtue of optical properties
rarely, cytoplasmic streaming strands. Organelles are slightly different from the typical tubules of ER along
certainly labeled by injection of anti-spectrin into which they ran parallel. With fluorescent micros-
onion cells, but nuclear label would be difficult to copies and a variety of methods to prepare cells,
evaluate because of ensheathing ER. Consistent with workers who have used either rhodamine phalloidin
the present data, the labeling of nuclei and cytoplas- or fluorescent anti-actin have more readily visualized
mic strands reported by de Ruijter and Emons may F-actin as apparently independent fibers of various
actually represent the ER. diameters, lengths and arrangements: among the
numerous authors are Thimann et al. (1992), who
Actin fixed Arena coleoptile cells prior to staining; Traas
Actin is a protein extensively studied by plant biolo- et al. (1987) who compared application of phalloidin
gists. In animals, it commonly binds at the cytoplas- to living cells by electroporation, to cells extracted
mic face of adhesion sites (e.g., Wang et al. 1993, Hitt with stabilizing buffer, and to cells prepared by a
and Luna 1994, Kieffer et al. 1995, Wang and Ingber standard method of fixation, Quader (1990) who
1995); and spectrin, which occurs there too, is a wide- stained fixed ceils, Staiger et al. (1994), Cleary et al.
spread member of the large group of actin-binding (1992), and Cleary (1995), who imaged fibers after
proteins (e.g., Hartwig 1994, Hitt and Luna 1994, injecting low levels of rbodamine phalloidin into
Pollard and Almo 1994). Although covisualization by living cells, Liebe and Menzel (1995) who developed
computational optical-sectioning microscopy does an improved method of fixing cells with the aid of
not test molecular contiguity, images of onion epider- microwaves.
mal cells showing colocalization of putative spectrin Using material fixed by freezing at high pressure for
with F-actin in plasmalemma-associated puncta are electron microscopy, Lichtscbeidl et al. (1990) ob-
consistent with the associations at adhesion sites in served individual actin fibrils in contact with the
animal cells. The conspicuousness of actin, putative ER, mitochondria, plastids, vacuoles, and other
spectrin, and other proteins in the putative adhesion organelles in Drosera tentacle cells. Bundles of
sites in stressed onion epidermal cells is remindful of microfilaments, sometimes splaying conspicuously,
the punctate arrays of F-actin seen by Thimann et al. often appeared to be wrapped with ER. In some
(1992) in cells of oat coleoptiles treated with very experiments, identity of the microfilaments was con-
high levels of calcium ion as well as of punctate firmed with immunogold-labeled antibodies. As fully
c. Reuzeauet al.: Integrinand cytoskeletalproteinsin onioncells 193

appreciated by these authors, only filaments at the maintain an active voltage drop (e.g., Li et al. 1995).
surface of an ultrathin section are well discriminated; Sixth, even if stained with a level of rhodamine phal-
filaments weaving through the embedding material of loidin low enough to preserve a natural distribution of
the section will be under-estimated because the F-actin, the extensively coated planar reticula and
refractive index of the embedding material so closely lamellae and particularly the very fine tubular net-
resembles that of the protein (cf. Penman 1995). works of the ER are difficult to visualize with con-
Thus, although the authors tended to think of separate ventional fluorescence microscopy. For example, the
filaments and ER interacting, with a tendency for ER reticulum stained by rhodamine phalloidin shown in
to clad filaments rather than vice versa, their major Fig. 6 A was not seen without a CCD camera, and the
emphasis was on the close association of filaments optimal plane for photography was not known until
and all the endomembranes. retrospective examination of the optical sections of
In general agreement with Lichtscheidl et al. (1990), the stack, one by one. A second photograph was not
F-actin has been visualized in the present paper on useful in this case, because the rhodamine bleached as
onion epidermal cells as more or less coextensive the initial image was captured.
with the ER and with other features of the endomem- On the other hand, even though large F-actin bundles
brane system (such as organelles and small vacuoles). and ER strands were not seen separately in the present
In evaluating these images, it must be appreciated that covisualizations, the possibility of their occurrence
the level of resolution of fluorescence microscopy is cannot be dismissed. Only occasionally and uninten-
inadequate to discriminate how fine a mesh of fila- tionally were moving strands imaged because, given
ments is associated with a membrane. However, current CCD cameras and stepping motors, they are
based on the entire set of images we have produced, not well captured by optical sectioning. Higher reso-
we speculate that in living cells the endomembranes lution may prove a disunion of moving ER tubules
are in general more thoroughly coated with a finer and long distinctive tracks of actin that has not been
fibrillar system of actin than is usually imagined. This hinted by the present relatively casual observations.
is not to deny, of course, that fibrils might be well Nevertheless, if strands of actin and ER occur sepa-
aligned during streaming and might gather into fine rately in onion bulb scale epidermis cells, it is sur-
strands, but we suggest that fibrils in any such strands prising that they were never seen thus during the pre-
would remain closely associated with membrane. sent study.
Possibly, such an embracing association between F- The family of plant actin proteins is evidently large
actin and the endomembranes, particularly during (e.g., An et al. 1996), and its roles are numerous. It is
membrane streaming, has been unseen in much pre- clear that not all functions of either the ER or actins
vious research for six reasons. First, cytoskeletal are interdependent. The time approaches when it will
meshes are not well evidenced in electron micro- be possible to examine actin distributions with more
graphs of embedded specimens (Penman 1995). molecular specificity, with better spatial resolution,
Second, fixation may perhaps sometimes promote with temporal dynamics, with covisualized distribu-
bundling of actin. Third, high levels of phalloidin tions of associate protein and lipid molecules, and
may adversely affect the natural dynamic organiza- with in vivo measurements of energy exchanges
tion of F-actin (e.g., Schmit and Lambert 1990). between such associates. Equally importantly, actin
Fourth, in some algal cells actin does form indepen- can be examined with these discriminating method-
dent tracks along which myosin-coated organelles ologies in the wide variety of species, cell types,
and ER may move (e.g., Williamson 1993, Gunning developmental stages, and intracellular processes in
and Steer 1996), and it has been loosely extrapolated which it is well known or speculated that actin is
that this may be a general phenomenon: thus, the involved.
large "bundles" sometimes seen in cells of higher
plants were expected to be actin alone. Fifth, covisu- Covisualization with intermediate filaments
alizations of an endomembrane stain with an F-actin The cytoplasmic distribution of intermediate fila-
stain were not generally attempted. Too, they might ments observed in the present study does not differ in
be challenging to the extent that DiOC6 staining of any notable way from the distribution of the endo-
membranes can depend on the voltage drop across membrane system. In contrast, intermediate filaments
them (Haugland 1992); however, physiologists study- are believed to contribute importantly to a fibrous
ing animal cells have considered that the ER cannot matrix in animal cytoplasm (e.g., Quinlan et al.
194 C. Reuzeau et al.: Integrin and cytoskeletal proteins in onion cells

1994), where their ready phosphorylation and de- sively with anti-spectrin and DiOC6 in plasmalemmal
phosphorylation suggests important regulatory capac- puncta and on the endomembranes when these anti-
ity. In the present study of living onion epidermal bodies are injected into onion cells.
cells, roles in a fibrous cytoplasmic matrix are not Ankyrin and anion exchange protein are also impor-
addressed and cannot be excluded; even if cyto- tant molecules often associating with spectrin in ani-
plasmic intermediate filaments stain with the anti- mals (e.g., Gomez and Morgans 1993, Ruetz et al.
body of Pruss et al. (1981), a dim matrix need not 1993). However, new separation and visualization
show up when juxtaposed with bright dense aggrega- data on these two proteins (Reuzeau et al. 1995a,
tions at the putative adhesion sites and on all the 1997a) are extensive enough to require separate pre-
endomembranes. On the other hand, as in animal cells sentation. In summary, however, codistribution of
and as anticipated for onion cells on the basis of work ankyrin and anion exchange protein with spectrin has
on pea nuclei by McNulty and Saunders (1992), inter- been observed associated both with membrane puncta
mediate filaments are abundant in the cell nucleus. and with the endomembrane system.
Indeed, preliminary images with the 4 • 4 CCD
camera binning that was standard for the present The endomembrane sheath
paper show enough specificity of distribution within Evidence from covisualizations
the nucleolus and within the nucleoplasm at large to The codistribution of several key cytoskeletal pro-
encourage future examination at lower binning and teins with the endomembrane system, including por-
closer spacing of focal planes of a stack. tions associated with puncta arrayed in the plasma
membrane, indicates the presence of a previously
unappreciated feature of the plant cell: the endomem-
Other cytoskeletal elements brane sheath. In this paper we have provided strongly
suggestive evidence only for spectrin, actin, and
In animal cells, talin and vinculin are proteins that are
intermediate filaments in the sheath, but this evidence
often associated with each other and which are inde-
is matched by data published only in abstract that
pendently found at various kinds of adhesion sites
strongly suggest the presence in the sheath of ankyrin.
(e.g., Jockusch et al. 1995, Gilmore and Burridge
These data are supplemented by less extensive evi-
1996). Notably, at these sites they are associated with dence for vinculin and talin. More importantly these
actin (ibid.). Talin and vinculin represent the inade- data are reinforced by strong evidence that there is a
quately understood but evidently large group of pro- distinctly greater concentration of proteins at the
teins (e.g., Jockusch et al. 1995) that serve to link the endomembrane surface than in the "cytosol" or "cyto-
primary elements of the cytoskeleton and its regula- plasmic matrix".
tory molecules together. Although we have not stud- The codistribution of putative ~ integrin with the
ied talin or vinculin antigenicities following separa- endomembrane system/plasmalemmal puncta sug-
tion of proteins from the onion extracts in which they gests how the sheath attaches to the membrane.
were present, these molecules are representative can- Among the signaling and structural roles of proteins
didates for the kinds of homologs of small cytoskele- of the integrin family is the attachment of cytoskeletal
tal interconnecting proteins that one expects to find in elements such as actin (e.g., Schwartz and Ingber
plants (Reuzeau and Pont-Lezica 1995). Moreover, 1994, Kieffer et al. 1995). In this function, so far as is
Quatrano et al. (1991) have presented Western blots known, it is always joined by a host of associated pro-
from electrophoretically separated Fucus extracts in teins many of which may be specific to cell type and
which anti-chicken vinculin identified a band of mo- to location of integrin in the cell. Proteins of the anion
lecular mass similar to that of authentic chicken vin- exchange family, while serving to control pH and ion
culin. And, even if the antigens that are recognized do fluxes, also play important roles in affixing cytoskele-
not in the end prove homologous to the animal pro- ton (e.g., Gomez and Morgans 1993, Ruetz et al.
teins, their presence indicates the relative abundance 1993).
of proteins on the endomembrane surface and at the Thus, while the described codistribution of proteins
putative adhesion sites. Thus, it seems appropriate to suggested to be integrin, spectrin, actin, and interme-
present images (Figs. 1 F, 5 E, and 7 C, D) showing diate filaments appears likely to be diagnostic for the
that anti-animal vinculin and talin colocalize exten- presence of adhesion sites and an endomembrane
C. Reuzeau et al.: Integrin and cytoskeletal proteins in onion cells 195

sheath, it only hints at the large number of transmem- matching monies from Dow Chemical Co. Encouragement from
brahe linkers and cytoskeletal elements that must Thora W. Halstead, f o ~ e r Chief of the Space Biology Program at
NASA, has been particularly important to us. We thank Mary Jane
surely participate in structuring these architectural
Saunders of the University of Southern Florida, Eugene Marcantonio
elements. of Columbia University, and Howard Hughes Medical Institute
Investigator Richard O. Hynes at Massachusetts Institute of Techno-
Evidence from embedment-free electron microscopy logy for gifts of antibodies. We appreciate the use of computers at the
Biomedical Computer Laboratory at Washington University and
A corroborative view of a cytoskeletal cladding of
assistance from its members; we especially thank Josd-Angel Con-
endomembranes has been presented by Penman chello, Chrysanthe Preza and Frederick U. Rosenbergcr for helpful
(1995), based on embedment-free electron micros- attention. We thank Hugh Chou and Gerald C, Johns for care of com-
copy of animal cells. This technique enables the puter systems, Michael M. Veith for photography of gel blots, Paul
visualization of a distinctive mesh of protein through- A. Ely for electronic work, and James R. Pease for shop work. We
out the cytoplasm, densely aggregated around and are grateful to J. Scott Gens for collegial assistance with computa-
tional optical-sectioning microscopy.
specifically anchored to the various membrane sur-
faces of the organelles. It makes readily accessible the
formerly controversial "microtrabecular lattice" References
described by Porter (1984) using high-voltage elec- Allen NS, Brown DT (1988) Dynamics of the endoplasmic reticulum
tron microscopy with unembedded whole mounts. in living onion epidermal cells in relation to microtubules,
The mutual reinforcement of the two new kinds of microfilaments, and intracellular particle movement. Cell Motil
evidence - covisualization by optical sectioning of Cytoskeleton 10:153-163
An Y-Q, McDowell JM, Huang S, McKinney EC, Chambtiss S,
living cells and electron microscopy of embedment-
Meagher RB (1996) Strong, constitutive expression of the Ara-
free ultrathin sections - is the stronger as neither bidopsis ACT2/ACT8 actin subclass in vegetative tissues. Plant
study was carried out with knowledge of the other. J 10:107-121
Ayala J (1994) Transport and internal organization of membranes:
Functions of the sheath vesicles, membrane networks and GTP-binding proteins. J Cell
Sci 107:753-763
Evidence most economically interpreted to demon-
Beck KA, Buchanan JA, Malhotra V, Nelson WJ (1994) Golgi spec-
strate endomembrane integrin and evidence plainly trin: identification of an erythroid [~-spectrin homolog associated
indicating the existence of an endomembrane sheath with the Golgi complex. J Cell Biol 127:707-723
invites inquiry about the function of this system. Bennett V, Gilligan DM (1993) The spectrin-based membrane skele-
Briefly, we propose that the system orchestrates meta- ton and micron-scale organization of the plasma membrane.
Annu Rev Cell Bid 9:27-66
bolic regulation by providing and modulating loci for
Carrington WA, Fogarty KE, Lifschitz L, Fay FS (1989) Three-
channelling, and accelerates channeling as needed by dimensional imaging on confocal and wide-field microscopes.
dragging the ER and organelles through the cyto- In: Pawley J (ed) The handbook of biological confocal micro-
plasm. We propose that to accomplish this "stream- scopy. IMR Press, Madison, WI, pp 151-161
ing", F-actin levers against the rest of the endomem- Cheresh DA, Mecham RP (1994) Integrins: molecular and biological
responses to the extracellular matrix. Academic Press, San Diego
brane sheath and the ER can also lever against the
Chrombach A, Rodbard D (1971) Polyacrylamide gel electrophore-
plasmalemmal control centers. As an important agent sis~ Science 172:440-450
in the control of cellular activities, according to this Cleary AL (1995) F-actin redistributions at the division site in living
proposal, the endomembrane sheath plays a major Tradescantia stomatal complexes as revealed by microinjection
part in responses to diverse signals and stresses, and of rhodamine-phalloidin. Protoptasma 185: t 52-- 165
under extreme stress cell survival depends on the abil- - Gunning BES, Wasteneys GO, Hepler PK (1992) Microtubute
and F-actin dynamics at the division site in living Tradescantia
ity of the system to maintain enough integrity to stamen hair cells. J Cell Sci 103:977-988
direct critical syntheses and degradations. This pro- Conchello JA, McNally JG (1966) Fast regularization technique for
posal proceeds well beyond strict interpretation of the expectation maximization algorithm for computational optical
data of the present paper, and will be elaborated else- sectioning microscopy. In: Cogswell CJ, Kino GS, Wilson T
where (Reuzeau et al. 1997b). (eds) Three-dimensional microscopy: image acquisition and pro-
cessing. Proc SPIE 2655:199-208
Craig S, Staehelin LA (1988) High pressure freezing of intact plant
Acknowledgements tissues: evaluation and characterization of novel features of the
We are grateful for funds from the NASA/NSF Joint Program in endoplasmic reticulum and associated membrane systems. Eur J
Plant Biology (grant IBN941601), from the NASA Space Biology Cell Bio146:80-93
Program (grant NAGW 3046), NIH Biomedical Research Technolo- de Ruijter N, Emons A (1993) Immunodetection of spectrin antigens
gy Program (grant RR01380), and from Glenn L. Allen, Jr. with in plant cells. Cell Biol Int Rep 17:169-182
196 C. Reuzeau et al.: Integrin and cytoskeletal proteins in onion cells

Ding JP, Pickard BG (1993) Mechanosensory calcium-selective 2 integrin chain in vitro. Biochem Biophys Res Commun 217:
cation channels in epidermal cells. Plant J 3:83-110 466474
Enomoto-Iwamoto M, Menko AS, Philp N, B oettiger D (1993) Eval- Knebel W, Quader H, Schnepf E (1990) Mobile and immobile endo-
uation of integrin molecules involved in substrate adhesion. Cell plasmic reticulum in onion bulb epidermis cells: short- and long-
Adhes Commun 1:191-202 term observations with a confocal laser scanning microscope.
Fairbaim DJ, Goodbody KC, Lloyd CW (1994) Simultaneous Eur J Cell Biol 52:328-340
labelling of microtubules and fibrillar bundles in tobacco BY-2 Laemmli EK (1970) Cleavage of structural proteins during the
cells by the anti-intermediate filament antibody, ME 101. Proto- assembly of the head of bacteriophage T4. Nature 227:680~685
plasma 182:160-169 Li Y-X, Keizer J, Stojilkovic SS, Rinzel J (1995) Ca 2+ excitability of
Faraday CD, Spanswick RM (1993) Evidence for a membrane skele- the ER membrane: an explanation for IP3-induced Ca 2+ oscilla-
ton in higher plants: a spectrin-like polypeptide co-isolates with tions. Am jr Physiol 2 6 9 : C 1 0 7 9 ~ 1 0 9 2
rice root plasma membrane. FEBS Lett 318:313-316 Lichtscheidl IK, Url WG (1990) Organization and dynamics of cor-
Fuhrmann C, Bereiter-Hahn J, Br~indle K (1990) Influence of the tical endoplasmic reticulum in inner epidermal cells of onion
cytoskeleton, energy supply, and protein synthesis on the struc- bulb scales. Protoplasma 157:203-215
ture of the endoplasmic reticulum. Protoplasma 158:53-65 - Lancelle SA, Hepler PK (1990) Actin-endoplasmic reticulum
Geiger B, Ayalon O (1992) Cadherins. Annu Rev Cell Biol 8: 307- complexes in Drosera: their structural relationship with the plas-
332 malemma, nucleus, and organelles in cells prepared by high
Gens JS, Reuzeau C, Doolittle KW, McNally JG, Pickard BG (1996) pressure freezing. Protoplasma 155:116-126
Covisualization by computational optical-sectioning microscopy Liebe S, Menzel D (1995) Actomyosin-based motility of endoplas-
of integrin and associated proteins at the cell membrane of living mic reticulum and chloroplasts in Vallisneria mesophyll cells.
onion protoplasts. Protoplasma 194:215-230 Biol Cell 85:207-222
- Fitriani T, McNally JG, Pickard BG (1997) Wall-associated - Quader H (1994) Myosin in onion (Allium cepa) bulb scale epi-
receptor-like kinase 1 (WAK1) comingles with antigens recog- dermal cells: involvement in dynamics of organelles and endo-
nized by antibodies against [~1 integrin and other adhesion site plasmic reticulum. Physiol Plant 90:114-124
proteins in wall/membrane puncta of BY2 tobacco cells. Plant Marcantonio E, Hynes RO (1988) Antibodies to the conserved cyto-
Physiol 114 Suppl: 14 15 plasmic domain of the integrin [31 subunit react with proteins in
Gilmore AP, Burridge K (1996) Regulation of vinculin binding to vertebrates, invertebrates, and fungi. J Cell Biol 106:1765-1772
talin and actin by phosphatidyl-inositol-4-5-bisphosphate. McNulty AK, Sannders MJ (1992) Purification and immunological
Nature 381:531-535 detection of pea nuclear intermediate filaments: evidence for
Goddard RH, Wick SM, Silflow CD, Snustad DP (1994) Micro- plant nuclear lamins. J Cell Sci 103:407~414
tubule components of the plant cytoskeleton. Plant Physiol 104: Michaud D, Guillet G, Rogers PA, Charest PM (1991) Identification
1-6 of a 220 kDa membrane-associated plant cell protein immuno-
Gomez S, Morgans C (1993) Interaction between band 3 and ankyrin logically related to human b-spectrin. FEBS Lett 294:77-80
begins in early compartments of the secretory pathway and is Miller DD, Scordilis SP, Hepler PK (1995) Identification and local-
essential for band 3 processing. J Biol Chem 268: 19593- ization of three classes of myosins in pollen tubes of Lilium lon-
19597 giflorum and Nicotiana alata. J Cell Sci 108:2549-2563
Grabski S, Feijter AW de, Schindler M (1993) Endoplasmic reticu- Mizuno K (1995) A cytoskeletal 50 kDa protein in higher plants that
lum forms a dynamic continuum for lipid diffusion between con- forms intermediate-sized filaments and stabilizes microtubules.
tiguous soybean root cells. Plant Cell 5:25-38 Protoplasma 186:99-112
Gumbiner BM (1993) Proteins associated with the cytoplasmic sur- Morr6 DJ, Keenan TW (1994) Golgi appm'atus buds - vesicles or
face of adhesion molecules. Neuron 11: 551-564 coated ends of tubules? Protoplasma 1 7 9 : 1 4
Gunning BES, Steer MW (1996) Plant cell biology. Jones and Bart- Oparka KJ, Read ND (1994) The use of fluorescent probes for stud-
lett Publishers, Sudbury, MA ies of living plant cells. In: Harris N, Oparka KJ (eds) Plant cell
Hartwig JH (1994) Actin-binding proteins 1: spectrin superfamily. biology: a practical approach. Oxford University Press, Oxford,
Protein Profile 1: 715-749 pp 27-50
Haugland RP (1992) Handbook of fluorescent probes and research - Murphy R, Derrick PM, Prior DAM, Smith JAC (1991) Modifi-
chemicals, 5th edn. Molecular Probes, Inc., Eugene, OR cation of the pressure-probe technique permits controlled intra-
Hepler P, Palevitz BA, Lancelle SA, McCauley MM, Liehtscheidl I cellular microinjection of fluorescent probes. J Cell Sci 98:
(1990) Cortical endoplasmic reticulum in plants. J Cell Sci 96: 539-544
355-373 - Prior DAM, Crawford JW (1994) Behavior of plasma mem-
Hitt AL, Luna EJ (1994) Membrane interactions with the actin cyto- brane, cortical ER, and plasmodesmata during plasmolysis of
skeleton. Curt Opin Cell Biol 6:120-130 onion epidermal cells. Plant Cell Environ 17:163-171
Ito Y, Abe S, Davies E (1992) Co-localization of cytoskeleton pro- Pardi R, Bossi G, Inverardi L, Rovida E, Bender JR (1995) Con-
teins and polysomes with a membrane fraction from peas. J Exp served regions in the cytoplasmic domains of the leukocyte inte-
Bot 45:253-259 grin a~[32 are involved in endoplasmic reticulum retention,
Jockusch BM, Bubeck P, Giehl K, Kroemker M, Moschner J, Roth- dimerization, and cytoskeletal association. J Immunol 155:
kegel M, Rtidiger M, Schlfiter K, Stanke G, Winkler J (1995) 1252-1263
The molecular architecture of focal adhesions. Annu Rev Cell Penman S (1995) Rethinking cell structure. Proc Natl Acad Sci USA
Biol I 1 : 3 7 9 4 1 6 92:5251-5257
Kieffer JD, Plopper G, Ingber DE, Hartwig JH, Kupper TS (1995) Pickard BG (1994) Contemplating the plasmalemmal control center
Direct binding of F-actin to the cytoplasmic domain of the alpha model. Protoplasma 182:1-9
C. Reuzeau et al.: Integrin and cytoskeletal proteins in onion cells 197

-
Ding JP (1993) The mechanosensory calcium-selective ion anion exchange family in onion epidermal cells. Plant Physiol
channel: key component of a plasmalemmal control centre? Aust 114 Suppl: 88-89
J Plant Physiol 20:439-459 - - McNally JG, Pickard BG (1997b) The endomembrane sheath:
Reuzeau C, Doolittle K, McNally JG (1994) High resolution
- a key structure for understanding the plant cell? Protoplasma (in
visualization in onion of distribution patterns of spectrin, talin press)
and vincnlin antigenicities. ASGSB Bull 8:54 Ruetz S, Lindsey AE, Ward CL, Kopito RR (1993) Functional
- McNally JG, Reuzeau C (1995) The endomembrane sheath - a activation of plasma membrane anion exchangers occurs in a
"new" component of the plant cell? ASGSB Bull 9:29 pre-Golgi compartment. J Cell Biol 1 2 1 : 3 7 4 8
Pollard TD, Almo S (1994) Structure of actin binding proteins: Schindler M, Meiners S, Cheresh DA (1989) RGD-dependent link-
insights about function at atomic resolution. Annu Rev Cell Biol age between plant cell wall and plasma membrane: conse-
10:207-249 quences for growth. J Cell Biol 108:1955-1965
Pont-Lezica RF, McNally JG, Pickard BG (1993) Wall-to-membrane Schmit A-C, Lambert A-M (1990) Microinjected fluorescent phal-
linkers in onion epidermis: some hypotheses. Plant Cell Environ loidin in vivo reveals the F-actin dynamics and assembly in
16:111-123 higher plant mitotic cells. Plant Cell 2:129-138
Porter KR (1984) The cytomatrix: a short history of its study. J Cell Schwartz MA, Ingber DE (1994) Integrating with integrins. Mol Biol
Biol 99 Suppl: 3s-12s Cell 5:3t89-93
Pruss RM, Mirsky R, Raft MC, Thorpe R, Dowding AJ, Anderton Sikorski AP, Swat W, Brzezifiska M, Wr6blewski Z, Bisikirska B
BH (1981) All classes of intermediate filaments share a common (t993) A protein cross-reacting with anti-spectrin antibodies is
antigenic determinant defined by a monoclonal antibody. Cell present in higher plant cells. Z Naturforsch 48c: 580-583
27:419-428 Staiger CJ, Yuan M, Valenta R, Shaw PJ, Warn RM, Lloyd CW
Quader H (1990) Formation and disintegration of cisternae of the (1994) Microinjected profilin affects cytoplasmic streaming in
endoplasmic reticulum visualized in live cells by conventional plant cells by rapidly depolymerizing actin microfilaments. Curr
fluorescence and confocal laser scanning microscopy: evidence Biol 4:215-219
for the involvement of calcium and the cytoskeleton. Protoplas- Stickens D, Verbelen J-P (1996) Spatial structure of mitochondria
ma 155:166-175 and ER denotes changes in cell physiology of cultured tobacco
Fast H (1990) Influence of cytosolic pH changes on the organi-
-
protoplasts. Plant J 9:85-92
zation of the endoplasmic reticulum in epidermal cells of onion Terasaki M, Reese TS (1992) Characterization of endoplasmic retic-
bulb scales: acidification by loading with weak organic acids. ulum by co-localization of BiP and dicarbocyanines. J Cell Sci
Protoplasma 157:216-224 101:315-322
- Liebe S (1995) Actin filament-independent formation of tubular Thimann KV, Reese K, Nachmias VT (1992) Actin and the elonga-
endoplasmic reticulum in onion epidermis cells. J Plant Physiol tion of plant cells. Protoplasma 171:153-166
145:71-77 Towbin H, Staehelin T, Gordon J (1979) Electrophoretic transfer of
- Schnepf E (1986) Endoplasmic reticulum and cytoplasmic protein from polyacrylamide gels to nitrocellulose sheets: proce-
streaming: fluorescence microscopical observations in adaxial dure and some applications. Proc Natl Acad Sci USA 76:
epidermis cells of onion bulb scales. Protoplasma 131:250-253 4350-4354
- Hofmann A, Schnepf E (1987) Shape and movement of the Traas JA, Doonan JH, Rawlins DJ, Shaw PJ, Watts J, Lloyd CW
endoplasmic reticulum in onion bulb epidermis cells: possible (1987) An actin network is present in the cytoplasm throughout
involvement of actin. Eur J Cell Bio144:17-26 the cell cycle of carrot cells and associates with the dividing
- (1989) Reorganization of the endoplasmic reticulum in epi-
- -
nucleus. J Cell Biol 105:387-395
dermal cells of onion bulb scales after cold stress: involvement Wachtler F, Stahl A (1993) The nucleolus: a structural and function-
of cytoskeletal elements. Planta 177:273-280 al interpretation. Micron 24:473-505
Quatrano RS, Brian L, Aldridge J, Schultz T (1991) Polar axis fixa- Wang N, Ingber DE (1995) Probing transmembrane mechanical cou-
tion in Fucus zygotes: components of the cytoskeleton and extra- pling and cytomechanics using magnetic twisting cytometry.
cellular matrix. Development Suppl 1:11-16 Biochem Cell Biol 73:327-335
Quinlan R, Hutchison C, Lane B (1994) Intermediate filament pro- - Butler JP, Ingber DE (1993) Mechanotransduction across the
teins. Protein Profile 1:779-800 cell surface and through the cytoskeleton. Science 260:
Reuzeau C, Pont-Lezica RF (1995) Comparing plant and animal 1124-1127
extracellular matrix-cytoskeleton connections - are they alike? Wang Y, Yan L (1988) The membrane protein of leaf cell membrane
Protoplasma 186:113-121 of Viciafaba. Chin Sci Bull 33:231-235
- Doolittle KW, McNally JG, Pickard BG (1995a) Injected anti- - - (1991) Immunochemical identification of spectrins on the
bodies against animal vinculin, ankyrin, talin and spectrin form plasma membrane of leaf cells of Viciafaba. Chin Sci Bull 36:
punctate arrays connected by a fine "lacework" in living onion 862-866
cells. J Cell Biochem Supp121A: 465 Wayne R, Staves MP, Leopold AC (1992) The contribution of the
. . . . (1995b) Spectrin surrounds the endomembrane system of extracellular matrix to gravisensing in characean cells. J Cell Sci
onion epidermal cells. Mol Biol Cell 6 Suppl: 370A 101:611-623
McNally JG, Pickard BG (1995c) [31 Integrin in the endomem- Williamson RE (1993) Organelle movements. Annu Rev Plant Phy-
brane system of onion epidermal cells is colocalized with spec- siol Plant Mol Biol 44:181-202
trin and actin. ASGSB Bull 9:29 Yang C, Xing L, Zhai Z (1992) Intermediate filaments in higher
Fichter CR, McNally JG, Pickard BG (1997a) Colocalization of plant cells and their assembly in a cell-free system. Protoplasma
injected antibodies against spectrin, ankyrin, and members of the 171:44-54