Integrin and spectrin homologues, and cytoplasm-wall adhesion in tip growth

Tip growth is a major form of cell growth, characteristic of fungal hyphae and plant cells such as root hairs and pollen tubes. It involves localized apical extension, which produces highly polarized, tubular cells typical of species and life cycle stages. It requires coordination of apical cell wall deposition, tip extension, apex-directed cytoplasmic migration and organelle movements, the latter two being the focus of this work.

Because turgor is equivalent between the apical and increasingly vacuolate subapical regions, it cannot act as an internal motive force. Thus, with respect to cytoplasmic migration, fungal hyphae have been described as tube-dwelling amoebae (Reinhart, 1892; Isaac, 1964; Heath, 1990a), based on the similarities between amoeboid cells (Bray, 1992) and that of apical cytoplasm in tip growing cells. These include polarized distributions of cytoskeletal filaments, ion gradients and organelles (reviewed by Heath and Kaminskyj, 1989; Heath, 1990a, b; Jackson and Heath, 1993a), cytoplasmic contractility (McKerracher and Heath, 1986; Jackson and Heath, 1992; Kaminskyj et al., 1992a) and analogous responses to Ca²⁺ gradient perturbations (Kaminskyj et al., 1992a; Garrill et al., 1993).

Amoeboid cells require cytoskeleton/substratum attachments, such as focal contacts, in order to anchor the force necessary for movement (Bray, 1992). These contain plasma membrane-spanning proteins called integrins which mediate the connection between the cytoskeleton and the extracellular matrix (Hynes, 1992; Sastry and Horwitz, 1993). Recent models for some plant cell activities suggest that the cell wall is analogous to the extracellular matrix, that cytoplasm-cell wall interactions are required for normal function, and these interactions are mediated by integrin-like proteins (Kropf et al., 1988; Kropf, 1992; Saunders and Lord, 1992; Goodner and Quatrano, 1993; Wyatt and Carpita, 1993). Evidence for such models has been based on immunoblot crossreactivity with antisera to β₁-integrin (Marcantonio and Hynes, 1988; Quatrano et al., 1991) or extracellular matrix proteins (Saunders et al., 1991; Wagner et al., 1992; Zhu et al., 1993), or by inhibition of normal cell function using exogenous peptides containing integrin-extracellular matrix protein binding sequences (Schindler et al., 1989; Wayne et al., 1992; Bachewich and Heath, 1993). The detailed distribution of integrins or extracellular matrix proteins has not yet been shown in walled cells, especially hyphae, nor has there been evidence of integrins linking the plasma membrane and the hyphal cell wall.

With respect to organelle movement, both microtubules and actin can be bases for motor molecules (McKerracher and Heath, 1987; Heath, 1990a, 1994). However, as with cytoplasmic migration, organelle motility requires anchorage of the force-generating ensembles. It is likely that the anchorage is to the cell wall (Aist and Berns, 1981; McKerracher and Heath, 1987; Kaminskyj et al., 1989; Heath, 1990a; Aist and Bayles, 1991), but there are no reports of the distribution of appropriate candidate molecules.

In our ongoing studies on the functions of the cytoskeleton during tip growth, we have identified and localized integrin and spectrin homologues in the hyphal oomycete, Saprolegnia ferax. Integrin distribution and behaviour imply a role in cytoskeleton-cell wall attachment for cytoplasmic migration and organelle motility. Spectrin distribution suggests a role as a membrane skeleton.

Saprolegnia ferax (Gruith.) Thuret (ATCC no. 36051) was grown on 6 mm×35 mm strips of dialysis tubing (Heath, 1987) at room temperature (∼24°C) on a peptone/yeast extract medium solidified with 1.5% agar (OM; Heath and Greenwood, 1970). Colonies were also grown on OM agar containing 250 μM tetrapentyl ammonium chloride (TPAC; Kaminskyj et al., 1992b). Growth rates were calculated from colony radius at 24 hours (control) or 48 hours (TPAC). Sporangia were induced on coverslips from fibrin-enmeshed dialysis tubing colonies (Kaminskyj et al., 1992a) using the method of Holloway and Heath (1974).

Growing colonies were anchored on coverslips (Kaminskyj and Heath, 1994) and allowed to recover in liquid OM for ≥30 minutes. Some colonies were plasmolysed in 2 M sucrose in distilled H₂O, 1 M NaCl in OM, 0.5 M polyethylene glycol-400 (PEG; Baker, Phillipsburg, NJ) in OM, or 0.5 M sorbitol in OM, for 15 minutes before fixation.

Colonies were fixed for 30 minutes in 6% formaldehyde containing 100 μM m-maleimidobenzoyl N-hydroxy succinimide ester (Molecular Probes, Eugene OR). Fixative and staining solutions were prepared in saline containing 60 mM PIPES (Calbiochem, San Diego, CA), pH 7.0, 2 mM EGTA (Sigma, St Louis, MO), 2 mM MgCl₂, 137 mM NaCl, 268 μM KCl. Fixed colonies were stained for actin with rhodamine-phalloidin (RP, Molecular Probes) and for other proteins by immunofluorescence (Kaminskyj and Heath, 1994).

For immunofluorescence, fixed colonies were rinsed for 2× 5 minutes in saline, the wall was degraded for 5 minutes in 10 mg/ml Driselase (Kyowa Hakko Kogyo, Tokyo, Japan), and membranes were permeabilized in 0.05-0.1% (v/v) Triton X-100 (Baker) for 10 minutes. Nonspecific binding was blocked for 20 minutes before incubating in primary antisera (60-90 minutes at room temperature), and again before incubating in secondary antisera (as above) and RP. Blocking solution contained 5% (v/v) fresh egg albumin (Grey Ridge Farms, Toronto, ON) and 2.5% (w/v) skim milk powder (Carnation, Toronto, ON). Colonies were rinsed for 3× 10 minutes in saline after each staining step.

Polyclonal primary antisera, from rabbits, were raised against: (1) the consensus peptide for the cytoplasmic domain of β₁-integrin (Marcantonio and Hynes, 1988); (2) chicken erythrocyte spectrin (ICN, St Laurent, PQ (no longer available), and Sigma); (3) chicken gizzard tropomyosin (ICN); (4) chicken gizzard α-actinin (ICN); (5) yeast actin binding proteins: ABP1, cofilin and fimbrin; (6) peroxidase-conjugated anti-goat IgG (Sigma). Antiserum dilutions were 1:100 for integrin and spectrin, and 1:10 otherwise. Immunoblots using whole anti-spectrin serum showed a band with similar mobility to actin, and immunofluorescence with whole antiserum showed staining patterns which were congruent with RP. Thus, for spectrin immunofluorescence, the antiserum was adsorbed against rabbit myofibrils, 24 hours at 4°C on a rotating tilt table, which were then removed by centrifugation at 20,000 g for 20 minutes at 4°C. Myofibrils were prepared from glycerinated rabbit psoas muscle in 6 mM Na/K phosphate buffer, pH 6.8, containing 100 mM KCl and 5 mM MgCl₂. Using myofibril-adsorbed spectrin antiserum, there was no subapical congruence between spectrin and actin patterns.

Monoclonal antibodies were raised in mice against chicken gizzard vinculin, clones C19 and VIN-11-5 (ICN). These antibodies were used diluted 1:10.

Secondary antisera (Amersham, Chicago, IL) were used diluted 1:50-1:100. They were: (1) FITC-conjugated anti-rabbit, raised in goat; (2) FITC-conjugated anti-mouse, raised in sheep; and (3) Texas Red-conjugated anti-rabbit, raised in donkey. Anti-rabbit sera were adsorbed against lyophilized Saprolegnia powder to reduce nonspecific wall adherence (5 mg powder/200 μl diluted antibody) for 90 minutes at room temperature. The powder was removed before use by centrifugation at 20,000 g for 20 minutes at 4°C.

Equally bright images would have a ratio of 1; relatively faint FITC images would have a ratio greater than 1. Data are presented for cells prepared and examined in one session, since variations in confocal settings and sample preparation affected the amount and apparent staining intensity of apical remnants. Statistically comparable ratios were obtained in replicate experiments. Statistical analyses used Statview SE+Graphics 1.02 (Abacus Concepts, Inc., Berkeley, CA). Figures are contrast enhanced to facilitate visual comparison.

Saprolegnia colonies were grown for 3 days on 35 mm×75 mm sheets of dialysis tubing on solid OM. To reduce protease activity during collection/freezing and sample preparation, 40 μl of protease inhibitor solution containing 50 mM AEBSF (Calbiochem) and 500 μM leupeptin (Sigma) was added to each mycelial mat (∼200 mg fresh weight) before harvest. Mycelia were collected rapidly and frozen (∼10 seconds, total) in liquid nitrogen, ground in a mortar at −80°C, transferred to cold preweighed Eppendorf tubes, and stored at −80°C until use.

Fertilized chicken eggs (Brampton Chick Hatchery, Brampton, ON) were incubated for 4 days at 37°C and 86% relative humidity in a TX6 incubator (Marsh Farms, Garden Grove, CA). Embryos were isolated (∼20 seconds), frozen (∼20 seconds), then prepared as above. Unlike Saprolegnia colonies, the protease inhibitor solution would not wet the embryos. Protease inhibitor was added to the frozen tissue during grinding, using 60 μl solution per embryo.

For electrophoresis, 2 μl/mg of 100°C sample buffer (double strength buffer:distilled water, 3:1) was added to frozen samples, which were briefly vortexed and placed in a boiling water bath. Sample buffer contained 2% SDS (BDH specially pure, Toronto, ON) in 62 mM Tris-HCl, pH 6.8 (BDH), or 5% SDS and 10 mM EDTA (Fisher) in 100 mM Tris-HCl, pH 6.8 (Marcantonio and Hynes, 1988).

Thiol-reducing buffers contained 2-100 mM dithiothreitol (DTT; Bethesda Research Labs, Gaithersburg, MD) or 5% β-mercaptoethanol (Sigma). Samples were boiled for 5 minutes, chilled in icewater, centrifuged at 20,000 g for 20 minutes at 4°C, and filtered through an Acrodisc (0.2 μm pore, Gelman Sciences, Rexdale, ON). Some samples were Centricon-30 or -100 fractionated (Amicon, Beverly, MA) according to the manufacturer’s instructions.

Protein concentrations were determined using a modification of the enhanced copper (Lowry) method, since sample buffer components interfere with standard protein assays (Stoscheck, 1990). SDS was replaced with distilled water in preparing the copper reagent; bovine serum albumin standards and 5 μl samples of unknowns (with ≤2 mM DTT) were diluted in 2% SDS in 62 mM Tris-HCl, pH 6.8. Under these conditions there was a linear relationship between A₇₅₀ and protein for 10-250 μg bovine serum albumin per 400 μl. Protein content of samples extracted with >2 mM DTT was taken to be equivalent to similar buffer with ≤2 mM DTT. Typically, whole protein extracts from Saprolegnia contained 2 μg protein/μl; chick extracts contained 5 μg protein/μl.

Samples were mixed 4:1 with sample buffer containing 50% glycerol and 0.001% Bromophenol Blue (BDH) before loading. Proteins (25-100 μg/4× 1.5 mm well) were separated by SDS-PAGE using methods of Hames (1981), transferred to nitocellulose (Towbin et al., 1979), and stained with 0.2% Ponceau S (Allied Chemical, Morristown, NJ) in 3 mM trichloroacetic acid (Fisher) to visualize the lanes. Some gels were silver-stained for protein (Merrill et al., 1980) or carbohydrate (Tsai and Frasch, 1982). For the latter, periodic acid was replaced (w/w) with sodium periodate. Nitrocellulose blocking was in TBS-T (10 mM Tris-HCl, pH 7.5, containing 150 mM NaCl and 0.05% Tween-20 (Fisher)) containing 5% milk powder, for 45 minutes at 23°C, during which the Ponceau S was eluted.

The blots were probed with the same polyclonal primary antibodies at the same dilutions as used for immunofluorescence. Monoclonal anti-(chicken gizzard) actin, clone N350 (Amersham), raised in mouse, was used at 1:1000. Monoclonal anti-(rabbit myosin heavy chain) was used at 1:50. Incubations were 1-3 hours in a moist box at room temperature. Peroxidase-conjugated secondary antisera (Sigma) were used at 1:200. Incubations were 1-2 hours as above. Peroxidase was developed with H₂O₂ and 4-chloro-l-naphthol (Sigma).

Saprolegnia proteins (from mycelial extracts, Fig. 1a,b) with the expected molecular masses crossreact with antisera to actin (42 kDa; Fig. 1c), myosin (205 kDa; Fig. 1d), and spectrin (246 kDa; Fig. 1e). Under nonreducing conditions there is a β₁-integrin-reactive band at 178 kDa (Fig. 1f). With thiol-reducing agents (2-100 mM DTT or 5% β-mercaptoethanol) this band appears to be converted into one of 120 kDa, which occurs in a DTT concentration-dependent manner (Fig. 1g-i). Both of these proteins appear to be glycosylated, since bands with similar masses and relative intensities to those on immunoblots are also found on carbohydrate-stained gels (Fig. 1j).

Proteins in walled cells are very susceptible to proteolysis (Quatrano et al., 1991), and this appears to be the case for Saprolegnia also, because samples prepared without the addition of protease inhibitors gave multiple lower mass spectrin and integrin crossreactive bands (not illustrated). Similarily, chick embryo proteins showed smaller than expected integrin and spectrin crossreactive bands, in addition to the predicted bands at 116 kDa and 230/240 kDa, respectively (not illustrated).

No Saprolegnia proteins crossreacted with antisera to α-actinin, tropomyosin, vinculin, yeast ABP1, cofilin or fimbrin, or peroxidase-conjugated anti-goat IgG, when these were used as primary antibodies for immunoblotting. Omitting primary antibodies gave no crossreaction.

In hyphae, the integrin antiserum most prominently stained peripheral patches (integrin patches) (Fig. 2b,d), with diffuse, central staining which was excluded from large organelles. The granularity of the immunofluorescence patterns may be technique-related, or may indeed reflect discontinuous distributions, possibly adhesion complexes in the case of integrin staining.

Integrin patches were abundant in the apical, i.e. 0-10 μm, segment of hyphal tips and declined through the 10-70 μm zone to lower levels in the most distal regions measured, 150-250 μm from the tip (Fig. 4A, Table 1). However, all hyphal regions contained integrin patches. In the 0-10 μm segment of hyphal tips, F-actin and integrin patches were so abundant that colocalization was inevitable. Subapically, where the density of both proteins was lower, the integrin patches did not colocalize with either actin filaments or plaques (Fig. 2a-d), confirmed by colour merge analysis (not shown).

The spectrin antisera also stained prominent peripheral patches (spectrin patches) in hyphae. These were superficially similar to the integrin patches (Fig. 2f), and diffuse central staining was again present. Spectrin patches were enriched at the tip (0-10 μm value: 0.548±0.013, n=46; >0.5 with 1-tailed t-test, P=0.0002), and were seen in all regions examined. F-actin and spectrin patches were both so abundant at the tips that they colocalized, but subapically many spectrin patches did not colocalize with actin (Fig. 2e,f).

We could not dual-label for spectrin and integrin as these antisera were both raised in rabbits. However, different masses of crossreactive proteins on immunoblots, and different staining patterns in plasmolysed hyphae (cf. Fig. 6c,d and e,h) and sporangia (cf. Fig. 3b and d), discussed below, suggest that these antisera do not crossreact.

During asexual reproduction, hyphae destined to become sporangia cease extending, expand and accumulate cytoplasm by apex-directed cytoplasmic migration (Gay and Greenwood, 1966). During this stage, sporangia contained integrin (Fig. 3a,b) and spectrin (Fig. 3c,d) patches, which were so abundant in the tips that they could not be resolved for counting. In sporangia whose cytoplasm separated from the cell wall during immunofluorescence processing, integrin remained associated with the wall (Fig. 3b; 11 of 13 sporangia), whereas spectrin segregated with the cytoplasm (Fig. 3d; 5 of 5 sporangia).

Saprolegnia hyphae did not stain with antisera to α-actinin, tropomyosin, vinculin, yeast ABP1, cofilin or fimbrin, and omitting primary antibodies gave no staining. In contrast, actin and tubulin antibodies gave detailed patterns which were very different from integrin or spectrin immunofluorescence (Kaminskyj and Heath, 1994).

To assess whether integrin patch number and distribution in hyphae correlated with growth, we reduced growth rate by half with TPAC (Table 1; Kaminskyj et al., 1992b) and determined the corresponding effect on integrin abundance. The tip-high integrin gradient of control hyphae was perturbed in the extreme apex of TPAC-treated hyphae (Fig. 4B; Table 1) and the abundance of integrin patches per segment was reduced in culture conditions, although the average rate remains steady (Kaminskyj et al., 1992b). However, the length of the filamentous cap of F-actin in the tips correlates with changes in growth rate (Jackson and Heath, 1990), especially in hyphae growing slower than 12 μm/min (r²=0.742; Kaminskyj et al., 1992b). We used F-actin cap length to look for finer correlations between growth rate and integrin patch abundance. As expected, control hyphae had longer actin caps than TPAC-treated hyphae (Table 1, ANOVA, P=0.0001). Control hyphae with longer actin caps had more integrin patches in the 0–10 μm segment; however, the correlation was low (r²=0.06; P=0.0802).

We used the clear difference between control and TPAC-treated hyphae to assess the role of integrins in cytoplasm-wall adhesion in Saprolegnia. If integrins are involved, then adhesion should be stronger in integrin-rich control hyphal tips, i.e. control hyphae should have more apical wall-attached cytoplasm after plasmolysis than TPAC treated hyphae. the 0-10 μm (ANOVA, P=0.0001), 10-40 μm (P≤0.01), and subapical (ANOVA, P=0.001) segments.

Growth rates in hyphae vary erratically under constant Colonies were plasmolysed with hypertonic solutions, and scored for apical-attached cytoplasm (Table 1; Fig 5). All hyphae were plasmolysed either apically or subapically with each treatment. However, apical cytoplasmic attachment was greater in control than TPAC-treated hyphae plasmolysed with sorbitol, sucrose and NaCl (factorial ANOVA, P=0.0312), although not with PEG.

Plasmolysed tips with apical separation had phase dark, wall-attached remnants (Figs 5c,d, 6a), especially prominent after PEG plasmolysis. These appeared to be regions where cytoplasm-to-wall attachment was stronger than cohesion to the adjacent cytoplasm.

Sporangia were more likely than hyphae to have tip-attached cytoplasm after plasmolysis, when all plasmolysing solutions including PEG were compared (factorial ANOVA, P=0.0279; Table 1).

If integrins mediate actin-wall interactions in Saprolegnia, then: (a) the cytoplasmic remnants after plasmolysis-induced tip separation should contain integrin; (b) control hyphae (with more apical integrin patches) should have more remnant material than TPAC-treated ones; and (c) integrin staining should be brighter than spectrin staining in the remnants, since spectrin does not participate directly in transmembrane linkages (Bennett, 1990). These experiments, summarized in Table 2, used PEG-plasmolysis, since apical separation was consistently high, and similar between control and TPAC-treated hyphae.

Plasmolysed hyphae were strikingly similar in their actin and integrin staining patterns (Fig. 6c-f), with both being concentrated in the remnants. There were significantly more apical remnants in control than TPAC-treated hyphae, but in both treatments the fluorescence intensity ratios were similar (Table 2; cf. Fig. 6c,d and e,f). In control hyphae (TPAC-treated hyphae were not examined), the apical remnants were significantly enriched in integrin over spectrin (Table 2; cf. Fig. 6c,d and g,h).

This is the first cytological study of cytoplasmic anchorage in hyphae. We have identified a novel integrin homologue using immunoblotting, shown with immunofluorescence that it is concentrated at the plasma membrane in a non-uniform pattern, experimentally manipulated the integrin distributions and correlated them with cytoplasm-cell wall attachment. We also identified and localized a spectrin homologue, again with a polarized distribution.

Saprolegnia protein extracts contain an integrin homologue which crossreacts on immunoblots with cytoplasmic domain β₁-integrin antiserum. This domain is conserved in vertebrates, invertebrates and fungi (Marcantonio and Hynes, 1988), as well as algae (Quatrano et al., 1991). Reported sizes of β₁-integrins show considerable variation, from 76 to 130 kDa (Hynes, 1987; Marcantonio and Hynes, 1988; Quatrano et al., 1991), spanning that of the reduced Saprolegnia β₁-integrin (120 kDa). β₁-Integrins have conserved intrachain disulphide bridges, so the reduced protein normally has lower mobility (Marcantonio and Hynes, 1988). The unexpected increase in mobility of the Saprolegnia integrin after reduction suggests the presence of an interchain disulphide bond, as reported for the α_IIbβ₃-integrin (Hynes, 1987), glycoprotein IIb (Calvete and González-Rodriguez, 1986). The Saprolegnia integrin homologue appears to be glycosylated, like other integrins (Hynes, 1992): carbohydrate-staining bands in protein extracts comigrated and varied in relative intensity after sample reduction in the same way as the integrin crossreacting bands on immunoblots.

Saprolegnia also contains a spectrin homologue similar in size, 246 kDa, to those of animals (Bennett, 1990), plants (Michaud et al., 1991; de Ruijter and Emons, 1992; Faraday and Spanswick, 1993), and protists (Choi and Jeon, 1989). Finding a single crossreactive band after probing with an erythrocyte antispectrin, rather than a doublet, is likely because erythrocytes contain α and β spectrins, whereas nonerythroid cells contain α and γ subunits (Bray, 1992). Spectrin is a membrane skeleton protein (Bennett and Gilligan, 1993) whose enrichment in Saprolegnia hyphal apices might contribute to membrane stability during extension.

Integrins appear to mediate attachment between the cytoskeleton and the cell wall, which is analogous to an extracellular matrix (Kropf et al., 1988; Kropf, 1992; Saunders and Lord, 1992; Goodner and Quatrano, 1993; Wyatt and Carpita, 1993). In Saprolegnia this model is supported by: (a) the peripheral location of the integrin patches; (b) the correlation between their abundance and the likelihood of the plasma membrane to separate from the wall during plasmolysis; (c) the greater resistance to plasmolysis in the integrin-rich sporangial tips relative to hyphal tips; (d) the formation of integrinand actin-rich cytoplasmic remnants attached to cell walls following plasmolysis; and (e) the colocalization of actin and integrin in hyphal and sporangial tips.

Integrins may anchor hyphal cytoplasm for apex-directed contractions (McKerracher and Heath, 1986; Jackson and Heath, 1992; Kaminskyj et al., 1992a), which may be an extreme form of the manner in which cytoplasm moves forward with the growing tip (Kaminskyj et al., 1992a). Contraction polarity implies differential anchorage of the contractile system, consistent with the gradients of integrin patches in hyphae and sporangia.

As with cytoplasmic migration, organelle motility requires anchorage, possibly to the cell wall (Aist and Berns, 1981; Aist and Bayles, 1991). The apical and subapical integrin patches are well placed for a role in this process, and their distribution along hyphae accords with similarly distributed organelle movements.

Apically, integrin and spectrin may function in tip morphogenesis, since recent observations indicate a role for the actin cap found in oomycete hyphal tips (Jackson and Heath, 1990, 1993b; Kaminskyj et al., 1992a,b; Money and Harold, 1992, 1993). Peptides which compete for integrin-extracellular matrix interactions in animal cells interfere with growth and morphology of Saprolegnia hyphae (Bachewich and Heath, 1993), implying a linkage between the cytoskeleton and the forming wall.

In apical regions, Saprolegnia integrin appears to associate with both actin arrays and the cell wall. In hyphal tips, actin and integrin (but not spectrin) colocalize in cytoplasmic remnants after plasmolysis; in sporangial tips, integrin (again, not spectrin) stays attached to the apical wall even if the cytoplasm is detatched during immunofluorescence preparation. These results are consistent with integrin being a transmembrane link between wall proteins and actin. In contrast, spectrin seems to be more independent of the predominant peripheral actin arrays, as also seen in nonwalled cells (Bennett, 1990; Bennett and Gilligan, 1993).

The apparent lack of coincidence between subapical actin plaques and integrin patches seems contrary to the apical linkages assumed above. However, the nature of the labile actin plaques is unknown (Heath, 1987, 1988, 1990a; Harold and Harold, 1992; Kaminskyj and Heath, 1994). They may form artefactually and separate from integrin if interfilament interactions are stronger than those between actin and integrin.

Integrins mediate signalling as well as anchorage (Hynes, 1992; Sastry and Horwitz, 1993). In Saprolegnia, the integrin distribution correlates with stretch-activated Ca²⁺ channels (Garrill et al., 1992, 1993; Levina et al., 1994). In fibroblasts, integrin-mediated adhesion regulates ion channel activity (Schwartz et al., 1991). Channel function (Garrill et al., 1993) and integrin-extracellular matrix interaction (Bachewich and Heath, 1993), and both are required for growth in Saprolegnia, suggesting that they may be interdependent. Similarly, stretchactivated Ca²⁺ channels and an integrin-like protein are codistributed in characean internode cells and both are required for gravisensing (Staves et al., 1992; Wayne et al., 1992).

This work was supported by an NSERC operating grant to I.B.H. and an Ontario Graduate Scholarship to S.G.W.K. We thank Dr R. Hynes, Dr K. Burridge, Dr D. Drubin, and Dr T. Hill for antisera, Dr R. Lew for the TPAC and PEG-400, Dr G. Hyde and Dr G. Aspinall for helpful discussions.