SEARCH

SEARCH BY CITATION

Keywords:

  • Xenopus;
  • adhesion;
  • cadherin;
  • juxtamembrane domain;
  • catenin;
  • p120;
  • ARVCF;
  • Rac

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

The cytoplasmic tail of cadherins is thought to regulate the strength and dynamics of cell–cell adhesion. Part of its regulatory activity has been attributed to a membrane-proximal region, the juxtamembrane domain (JMD), and its interaction with members of the p120 catenin subfamily. We show that titration of xARVCF, a member of this family, to the plasma membrane disrupts adhesion in the early embryo. Adhesion can be restored by coexpression of constitutively active Rac, suggesting that intracellular signaling is the primary cause in the loss of adhesion phenotype. Our observations suggest that the recruitment of p120 type catenins to the plasma membrane by the cadherin cytoplasmic tail may create protein complexes, which actively modulate the adhesion “status” of embryonic cells. Developmental Dynamics 237:2328–2341, 2008. © 2008 Wiley-Liss, Inc.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

Classical cadherins are major mediators of cell–cell adhesion and represent important determinants of tissue architecture and function, both during development and in mature organisms. In particular, the extensive rearrangements, migration, and sorting of cells that occur during development require a dynamic regulation of cell adhesion. While the importance of regulated cadherin activity during morphogenetic movements has been clearly demonstrated, the nature of this regulation is still unclear (for a recent review, see Gumbiner,2005).

Classical cadherins join neighboring cells by homophilic binding of their extracellular domains (Nose et al.,1988; Niessen and Gumbiner,2002; Duguay et al.,2003). The binding affinity of individual domains is weak, and efficient adhesion appears to be achieved through lateral clustering of cadherin dimers into larger adhesion spots, anchoring to the actin cytoskeleton and structural changes in the underlying actin cortex (Yap et al.,1997; Chu et al.,2004; Mege et al.,2006). The cytoplasmic tail has been shown to be essential for adhesive strengthening (Nagafuchi and Takeichi,1988; Brieher et al.,1996). Cadherin linkage to the actin cytoskeleton is attributed to the β-catenin-binding domain at the C-terminal end of the tail (Ozawa et al.,1990; Nagafuchi et al.,1994). β-catenin connects cadherin to α-catenin (Hinck et al.,1994), which in turn interacts with several other proteins, including actin. Although the β-catenin-α-catenin-actin linkage appears to be more complex than originally proposed (Rimm et al.,1995; Drees et al.,2005; Yamada et al.,2005), cadherin association with β- and α-catenin along with the recruitment of additional regulators of actin dynamics is certainly crucial for the establishment of strong yet dynamic adhesion (Nagafuchi et al.,1994; Kuroda et al.,1998; Perez-Moreno et al.,2003; Sallee et al.,2006).

In early Xenopus laevis embryos, C-cadherin is the main mediator of cell–cell adhesion and depletion of the C-cadherin pool leads to rounding up and dissociation of embryonic cells (Heasman et al.,1994). In addition to its essential role in maintaining tissue integrity, the dynamic regulation of C-cadherin-mediated adhesion is an important component of morphogenetic movements. Even relatively weak perturbations of adhesion, e.g., by overexpression of wild type cadherins or constructs lacking the cytoplasmic tail, can interfere with the orderly progression of gastrulation (Lee and Gumbiner,1995; Kuhl et al.,1996). Overexpression of a cadherin fragment lacking the extracellular domain also leads to dose-dependent defects, including the complete dissociation of cells (Kintner,1992; Broders and Thiery,1995). The inhibitory activity of this construct has been attributed to its ability to bind β-catenin, which is thought to deprive endogenous cadherin from this essential linker to the cytoskeleton. However, shorter fragments that cannot bind β-catenin also decreased adhesion, suggesting that other domains in the cytoplasmic tail might contribute to the inhibitory effect (Kintner,1992).

A membrane proximal region in the cytoplasmic tail, the juxtamembrane domain (hereafter JMD), has also been demonstrated to influence adhesive activity. However, the use of deletion constructs has not been able to determine with certainty whether this region plays a positive or a negative role and its function might be highly context dependent (Ozawa and Kemler,1998; Yap et al.,1998; Aono et al.,1999; Anastasiadis and Reynolds,2000). The function of the JMD has in many cases been related to its ability to bind cytoplasmic proteins of the p120catenin family (hereafter p120). These armadillo repeat proteins, which include p120 and ARVCF (Armadillo repeat gene deleted in velo-cardio-facial syndrome), show some similarity to β-catenin, but they do not interact with the carboxy-terminal β-catenin binding site. All members of the p120 family seem to share certain functions and are probably at least partially redundant (for recent reviews, see Hatzfeld,2005; McCrea and Park,2007). Several mechanisms have been proposed to explain the activity of p120 in cadherin-mediated adhesion. For example, p120 binds and stabilizes cadherins at the plasma membrane in mammalian cells (Ireton et al.,2002; Davis et al.,2003; Xiao et al.,2003). p120 also influences the activity of the small GTPases cdc42, Rac, and Rho, and, consequently, the architecture of the actin cytoskeleton (Anastasiadis et al.,2000; Noren et al.,2000; Grosheva et al.,2001; Ciesiolka et al.,2004; Fang et al.,2004). In addition, the JMD has been shown to be necessary for cadherins clustering at the plasma membrane and p120 has been suggested as a candidate to regulate this function (Yap et al.,1997,1998).

The reported effects of overexpressed cadherin fragments suggest an inhibitory function of the JMD on cell adhesion in the early Xenopus embryo. To study this possibility and the functional relation to its interaction with members of the p120 family, we compared the effects of several differentially modified JMD constructs on the integrity of embryonic tissue. While a plasma membrane–bound JMD construct interferes with adhesion to a certain extent, we found that this effect is significantly weaker than after the expression of constructs comprising the complete cytoplasmic tail. It also appears to require additional parts of the C-cadherin sequence, outside of the JMD. However, coexpression of membrane bound JMD with Xenopus ARVCF (xARVCF; or Xenopus p120 [Xp120]) leads to a complete loss of cell–cell adhesion, similar to the phenotype obtained after expression of the complete cytoplasmic tail. Inhibition of adhesion required the recruitment of xARVCF to the membrane-bound JMD. Coexpression of constitutively active Rac restored adhesion, indicating that intracellular signaling is affected by the accumulation of xARVCF at the plasma membrane.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

The C-Cadherin Juxtamembrane Domain Disrupts Tissue Morphology When Targeted to the Plasma Membrane

Overexpression of CadΔE, a cadherin construct that lacks the extracellular domain but contains the complete cytoplasmic tail, strongly interferes with cell adhesion in the early embryo (Kintner,1992; Broders and Thiery,1995). A further truncation that removes the distal β-catenin binding domain has been reported to still interfere with adhesion (Kintner,1992). However, it is not clear how the phenotypes relate to each other and how interference is mediated.

To characterize the activity of such truncated constructs in more detail, we designed, similar to a construct used in the previous study (Kintner,1992), a cadherin construct lacking both the extracellular domain and the distal part of the cytoplasmic tail (Fig. 1). This polypeptide, called JMDpm, consists of the JMD and the transmembrane domain and should simulate JMD localization at the plasma membrane outside of the context of cell–cell adhesion. To test for membrane-independent effects, the activity of JMDpm was compared to two other differentially localized JMD constructs: a fully soluble cytosolic JMD fragment (JMDcyt), and a JMD targeted to the cytosolic face of the outer mitochondrial membrane (JMDmi) by fusion with the TOM20 signal sequence (Kanaji et al.,2000). All constructs contained a myc-tag at their C-terminal end. mRNA encoding the constructs was injected in the animal region of Xenopus embryos. This region was selected because it gives rise to the ectoderm, a multilayered epithelium, which is a relatively simple and homogenous tissue well suited for detecting changes in cell–cell adhesion and cell shape. The embryos were fixed at the early gastrula stage and the resulting cryosections were double stained using an anti-C-cadherin and an anti-myc-tag antibody (Fig. 2). C-cadherin is homogenously expressed throughout the embryo, and could thus serve as a marker to highlight the cell outlines. The C-cadherin antibody has been raised against the extracellular domain and does not interact with the cytoplasmic cadherin constructs (Yap et al.,1997). Myc-tag staining was used to identify the cells expressing the constructs and to determine the subcellular localization of each construct.

thumbnail image

Figure 1. Structure of C-cadherin cytoplasmic tail constructs. To target constructs to the plasmamembrane, in CadΔE and JMDpm the signal sequence of C-cadherin is directly fused to the transmembrane domain. JMDcyt consists only of the JMD, while in JMDmi a mitochondria targeting sequence is added N-terminal to target the construct to the outer layer of the mitochondria membrane. JMDmyr contains a N-terminal myristylation sequence to target it to the plasma membrane, JMDmyrG-A harbours three glycin to arginin mutations, which abrogates binding of p120 catenin family members. All constructs contain in addition to the depicted domains a myc-tag at the C-terminal end.

Download figure to PowerPoint

thumbnail image

Figure 2. JMDpm perturbs tissue architecture in the animal cap. A: Cross-section of an uninjected early gastrula (stg.10.5), stained for C-cadherin. Dorsal is oriented to the right, animal to the top. Box depicts region magnified in A'). Bar = 200 μm. A': The animal cap consists of an outer (ol) and an inner layer (il). The inner layer appears apposed to the endoderm (ed) due to the collapse of the blastocoel during fixation. B–E: Animal cap of mRNA-injected embryos (1 ng/inj., except CadΔE: 250 pg/inj.). Overlay of staining for C-cadherin (green) and myc-tagged constructs (red). B'–E': C-cadherin alone. Arrows mark adherens junctions. Bar = 50 μm. B: JMDpm localizes to intracellular structures and the plasma membrane. Cells are rounded and loosely arranged. B': Cadherin staining at the plasma membrane appears to be reduced in both layers. C: CadΔE localizes to intracellular structures and the plasma membrane. Cells of the inner layer detach, while the outer epithelium remains a continuous cell layer. C': Cadherin staining at the plasma membrane is reduced. D: JMDcyt localizes to the nucleus and the cytoplasm. D': Cell arrangement is compact and cadherin staining is similar to uninjected control. E: JMDmi localizes to small cytoplasmic spots. E': Cell arrangement is compact and cadherin staining is similar to control.

Download figure to PowerPoint

In the blastula stage embryo, the ectoderm is organized in a single outer layer of polarized cells and a thicker sheet of inner cells. The thickness of the inner layer decreases from 4–5 to 1–2 cell diameters during epiboly, when its cells move between each other in a process termed radial intercalation (Keller,1980). During this early phase of gastrulation, cells of both outer and inner layers tend to be slightly elongated perpendicularly to the plane of the ectoderm (Fig. 2A'). Expression of JMDpm strongly affected the architecture of the animal cap. In both the inner and outer layer, cells were rounded, loosely arranged, and lacked any obvious pattern of orientation (Fig. 2B, B', Table 1). In addition, cadherin staining at the plasma membrane of these cells was generally reduced and the bright cadherin spots corresponding to the apical adherens junctions of the outer layer were absent (compare Fig. 2A' and B'). This phenotype suggested that intercellular adhesion is impaired. Another observation supported this conclusion: while in uninjected, cultured embryos at this stage, the two layers of the animal cap can be manually separated into two coherent cell sheets; in JMDpm-injected embryos, the inner layer fell apart into smaller pieces (not shown). The outer layer, however, remained intact during these manipulations, probably due to the presence of intercellular tight junctions (Muller and Hausen,1995). Compared to JMDpm, CadΔE appears to be a more potent inhibitor of adhesion. Most cells of the inner layer dissociated completely in CadΔE-injected embryos, even with comparatively lower amounts of injected mRNA and expressed protein (Fig. 2C, C'; Table 1 and data not shown). Similar to JMDpm, CadΔE-expressing cells showed reduced cadherin staining at the plasma membrane.

Table 1. Summary of Injected Constructs and Their Effects on Cell Arrangement in the Animal Capa
ConstructEmb.Exp.Cell arrangement in the animal cap, % of embryos
CompactLooseInner cells detached
  • a

    mRNA of the various constructs was injected at 2–4 cell stage into the animal region and the embryos were fixed, sectioned and stained at early gastrula stage (stage 10.5). mRNA concentrations: JMDpm, JMDcyt, JMDmi 1 ng/inj.; JMDmyr. JMDmyrG-A 100 pg/inj.; CadE, Xp120, xARVCF, xARVCFmyr 250 pg/inj. unless stated otherwise. The number of evaluated embryos (emb.) and experiments (exp.) is indicated for each construct.

uninjected20510000
JMDpm20535650
JMDcyt12310000
JMDmi12310000
JMDpm+xARVCF20501585
JMDpm+Xp12019501684
JMDcyt+xARVCF11310000
JMDmi+xARVCF12310000
Xarvcf16410000
Xarvcf 1 ng/inj.12310000
Xp120 1 ng/inj.10310000
CadE12300100
JMDmyr17310000
JMDmyrG-A16310000
JMDmyr+xARVCF18301090
JMDmyrG-A+xARVCF17360400
xARVCFmyr1530694

In contrast to these two plasma membrane–bound constructs, JMDcyt or JMDmi did not seem to affect cell–cell adhesion. While the inner layer was frequently thicker than in uninjected embryos of the same stage, suggesting slower or delayed cell intercalation, cells appeared tightly apposed and generally oriented towards the surface (Fig. 2D, D' and E, E', Table 1). The intensity of cadherin staining was similar to that in the uninjected controls (compare Fig. 2A' with D' and E') and the integrity of the inner cell sheet remained intact when manually dissected (not shown).

Therefore, to interfere with tissue architecture, the localization of the JMD to the plasma membrane appears essential. However, while both JMDpm and CadΔE lead to a reduction of cadherin staining, JMDpm has only a comparatively weak effect on tissue integrity.

Coexpression of xARVCF With JMDpm Abolishes Adhesion

In Xenopus, two closely related members of the p120 subfamily of catenins, Xp120 and xARVCF, are essential for development. In the early embryo, depletion of either of these proteins reduces cadherin levels (Fang et al.,2004; Tao et al.,2007). The simplest explanation for the JMDpm-induced phenotype therefore would be that JMDpm titrates p120-like proteins away from endogenous cadherins. While the observation that JMDmi and JMDcyt had no effect suggests that this is not the case, binding to the JMD might be context dependent (Paulson et al.,2000; Miranda et al.,2003). To further test the titration-hypothesis, we coexpressed xARVCF or Xp120 with JMDpm in a rescue experiment.

To our surprise, coexpression of an HA-tagged version of xARVCF or of untagged Xp120 with JMDpm not only failed to rescue adhesion, but resulted in a much stronger phenotype than expression of JMDpm alone. In 85% of the embryos, cells of the inner layer of the animal cap detached completely (Fig. 3B, Table 1). Cadherin staining at the plasma membrane again appeared reduced. This phenotype is similar to the one observed after injection of CadΔE (compare Figs. 3B and 2C). Loss of adhesion was JMDpm-dependent, since xARVCF or Xp120 alone had no detectable effect on cap morphology or cadherin staining, even after injecting four times more mRNA than used in the coexpression experiment (Fig. 3A, Table 1). Coinjection of xARVCF with JMDcyt or JMDmi also had no effect on animal cap morphology (Fig. 3C and D), confirming that membrane localization of the JMD is necessary for the inhibitory activity of xARVCF. Interestingly, we observed that xARVCF altered the distribution of JMDmi in the cytoplasm from a relatively even, speckled pattern to a small number of large aggregates (compare Figs. 2E and 3D), indicating that xARVCF retains an activity in this context.

thumbnail image

Figure 3. Inner cells of the animal cap detach after coexpression of xARVCF and JMDpm. A–D: Animal cap of mRNA injected embryos (all JMD constructs 1 ng/inj., xARVCF 250 pg/inj.), stage 10.5. Overlay of C-cadherin (green) and myc-staining (red), except A), which is HA-staining (red). Bar = 50 μm. A'–E': C-cadherin staining alone. A: xARVCF, animal cap remains intact. Construct localizes to nucleus, cytoplasm, and plasma membrane (see also Fig. 4A). B: JMDpm+xARVCF, cells of the inner layer (il) detach, outer layer (ol) remains intact. JMDpm locates to intracellular structures and the plasma membrane. B': Cadherin staining at the plasma membrane is reduced. C: JMDcyt+xARVCF, animal cap remains intact. JMDcyt locates to nucleus and the cytoplasm. D: JMDmi+xARVCF, animal cap remains intact. JMDmi accumulates in large, mostly perinuclear patches. A', B',D': Cadherin staining at the plasma membrane is unaffected.

Download figure to PowerPoint

These results show that coexpression of xARVCF or Xp120 with membrane-bound JMD interferes with adhesion to a similar degree as CadΔE, indicating that increasing amounts of these proteins can compensate for the lack of the β-catenin binding domain in the JMD construct.

JMDpm Titrates xARVCF Efficiently to the Plasma Membrane

p120-like proteins are thought to have different functions depending on their subcellular localization and we hypothesize that coexpressed JMDpm and xARVCF interfere with adhesion by titrating xARVCF to the plasma membrane. However, JMDpm staining showed a large proportion of the construct in intracellular structures, rendering it possible that xARVCF acts by a more indirect mechanism, e.g., in the exocytotic pathway. To characterize the effect of JMD constructs on xARVCF distribution, we stained sections for the HA-tag of overexpressed xARVCF.

When injected alone, xARVCF showed a diffuse distribution, which included nuclei, cytoplasm, and an often patchy plasma membrane staining (Fig. 4A). JMDcyt, which has itself a diffuse localization, did not change the nuclear and cytoplasmic distribution of xARVCF. However, the strongly xARVCF-positive patches at the plasma membrane appeared to be reduced (Fig. 4B), indicating that JMDcyt can alter the distribution of at least a certain proportion of xARVCF. When coexpressed with JMDmi, xARVCF accumulated in large cytoplasmic structures that were also positive for JMDmi (Fig. 4D, data not shown). Coexpressed with JMDpm, xARVCF localized almost exclusively to the plasma membrane (Fig. 4C), except for small xARVCF positive spots in the cytoplasm, indicating that the intracellular fraction of JMDpm does not bind xARVCF and might be trapped in the endoplasmic reticulum. xARVCF redistribution was dose-dependent: Decreasing the JMDpm to xARVCF mRNA ratio from 4:1 to 3:1 and then 2:1 led to a progressive decrease in plasma membrane staining and an increase in intracellular xARVCF staining, first in the nucleus and then in the cytoplasm, suggesting that the intracellular localization of xARVCF depends mainly on the availability of binding sites (not shown). Coimmunoprecipitation experiments confirmed that all JMD constructs bound xARVCF (Fig. 4E). However, while the HA-staining indicated that JMDmi titrates xARVCF quite efficiently, we consistently found a larger proportion of xARVCF coprecipitating with JMDpm than with JMDmi or JMDcyt.

These observations show that JMDpm recruits xARVCF effectively to the plasma membrane and indicate that additional factors stabilize this specific interaction, which might contribute to the efficient recruitment.

Expression of JMDpm Leads to a Redistribution of Endogenous Cadherin From the Plasma Membrane to the Cytoplasm

Expression of constructs interfering with adhesion (JMDpm, JMDpm+xARVCF or CadΔE) resulted in a visible reduction of C-cadherin staining at the plasma membrane. We noted that this reduction coincided with a conspicuous accumulation of cadherin-positive spots in the cytoplasm, often together with distinguishable spots of construct accumulations (Fig. 5A, A' arrows). This observation suggested that the reduction/loss of adhesion might be the direct result of a redistribution of cadherin. The number of cytoplasmic cadherin spots per cell in uninjected embryos was very low (on average 0.2 spots/cell, Fig. 5E). The appearance of multiple cytoplasmic cadherin spots was restricted to JMDpm, JMDpm+xARVCF and CadΔE expressing embryos (Fig. 5A',B',D',E), as constructs that did not interfere with adhesion (JMDcyt, JMDcyt+xARVCF or xARVCF alone) also did not increase the frequency of these spots (Fig. 5C',E). On average, JMDpm- and JMDpm+xARVCF-expressing cells contained 3 spots/cell. In cells expressing CadΔE, the number was significantly higher (5–6 spots/cell, significant at α = 0.01). In all three conditions, there was no significant difference in the number of cadherin spots between cells of the outer and the inner layer (Fig. 5E). Thus, reduction of cell adhesion correlates with a shift of cadherin-signal from the plasma membrane to the cytoplasm. However, the equal number of spots in JMDpm and JMDpm+xARVCF-expressing cells suggests that the amount of redistributed cadherin does not necessarily correlate with the intensity of the adhesion phenotype.

thumbnail image

Figure 4. Coexpression of JMD constructs changes intracellular distribution of xARVCF. Animal cap of injected embryos (all JMD constructs 1 ng/inj., xARVCF 250 pg/inj.), stage 10.5. Bar = 20 μm. A–D: xARVCF-HA staining. A'–D': C-cadherin staining. A: xARVCF localizes to nucleus, cytoplasm, and plasma membrane. Arrows: Patches of stronger plasma membrane staining. B: JMDcyt+xARVCF, xARVCF locates mainly to cytoplasm and nuclei, plasma membrane staining appears partially reduced. C: JMDpm+xARVCF, xARVCF localizes to the plasma membrane and to intracellular spots (arrows). C': Cadherin staining shows cytoplasmic spots that colocalize with xARVCF spots (compare arrows). A', B', D': Cadherin staining is restricted to the plasma membrane, no defined cytoplasmic spots are detectable. D: JMDmi+xARVCF, xARVCF localizes to large cytoplasmic patches (arrowheads). E: Co-immunoprecipitation of xARVCF. Embryos were co-injected with mRNAs encoding HA-tagged xARVCF and myc-tagged JMD constructs or a myc-tagged EGFP construct as control. Embryos were extracted at early gastrula stage and the JMD-constructs were immunoprecipitated with an antibody against the myc-tag. Loaded lysate corresponds to the equivalent of 1/4 embryo, loaded myc-IP to the equivalent of 9 embryos. Co-precipitating xARVCF was detected with a HA-tag antibody. xARVCF co-precipitates with the JMD constructs, but not with EGFPmyc. Note that JMDpm appears to bind xARVCF more efficiently than JMDmi and JMDcyt.

Download figure to PowerPoint

thumbnail image

Figure 5. JMDpm expression leads to the formation of cadherin-positive intracellular spots. Animal cap of injected embryos at early gastrula, stage 10.5 (JMDpm, JMDcyt 1 ng/inj.; CadΔE, xARVCF 250 pg/inj.). Bar = 20 μm. A–D: staining of myc tagged constructs. A'–D': C-cadherin staining. A: JMDpm, construct localizes to the plasma membrane and intracellular structures. In some cells, isolated myc-positive spots can be distinguished (arrows). A': Cadherin-positive spots appear in the cytoplasm (arrows), showing the same distribution as myc-positive spots. B: JMDpm+xARVCF, JMDpm shows similar intracellular distribution as in A. B': Cadherin-positive spots in the cytoplasm. C: JMDcyt+xARVCF, JMDcyt is cytoplasmic and nuclear. C': no cytoplasmic cadherin spots are visible. D: CadΔE localizes to plasma membrane and intracellular structures. D': Cadherin-positive spots appear in the cytoplasm. E: Quantification of intracellular cadherin spots. The number of cadherin-positive spots in the cytoplasm was determined per myc-positive cell. Cells of the outer layer (ol) and cells of the inner layer (il) were evaluated separately. If construct expression resulted in an increase of cadherin spots, spots in adjacent myc-negative cells were also counted. Error-bars indicate standard deviation.

Download figure to PowerPoint

Expression of JMDpm Does Not Affect Cadherin Levels or Cadherin-α/β-Catenin Interaction

The strength of cadherin-mediated adhesion can potentially be modulated by regulating its physical association with the actin cytoskeleton via α- and β-catenin, and the loss of adhesion caused by CadΔE expression has been explained by titration of β-catenin away from endogenous cadherin (Kintner,1992). JMDpm does not contain a β-catenin binding domain and neither JMDpm alone nor together with xARVCF binds α- or β-catenin (data not shown). However, it remains possible that the construct disrupts the cadherin–catenin link by more indirect means.

To test for effects on the cadherin–catenin bonds, we injected either JMDpm+xARVCF, CadΔE, or control EGFPmyc mRNA. In these experiments, embryos were injected at the two-cell stage into the marginal zone instead of animal to interfere with adhesion in as many cells as possible. Based on the dissection of some embryos just before extraction, we estimate the proportion of loose cells in injected embryos between 60 and 80%. The expression of the constructs did not change the total amounts of cadherin, α- and β-catenin in embryonic lysates (Fig. 6) except for a small increase in β-catenin levels in CadΔE-injected embryos, which is probably due to the binding and stabilization of cytoplasmic β-catenin by CadΔE. Cadherin immunoprecipitates from lysates of the three conditions contained equal amounts of cadherin, and there was no detectable difference in coprecipitated amounts of β-catenin and α-catenin between JMDpm+xARVCF- and EGFPmyc-injected control embryos (Fig. 6). Even in CadΔE-injected embryos, only a slight reduction in coprecipitated catenins was detectable and even a fourfold increase in injected CadE mRNA did not yield a stronger effect.

thumbnail image

Figure 6. Cadherin levels in the embryo and amounts of associated catenins remain the same after JMDpm+xARVCF injection. Embryos were injected with EGFP-myc as a control (250 pg/inj.), JMDpm+xARVCF (1 ng resp. 250 pg/inj.) or CadΔE (250 pg or 1 ng/inj.) and 30 embryos/condition were extracted at stage 10.5. Immunoprecipitation was done with a cadherin antibody (5G5). Loaded lysate corresponds to the equivalent of 1 embryo. Loaded Cadherin-IP corresponds to the equivalent of 30 embryos. There is no detectable difference in absolute levels of Cadherin, β-, or α-catenin between the EGFP-myc control and JMD+xARVCF and similar amounts of the two catenins are associated with precipitated cadherin.

Download figure to PowerPoint

These results strongly suggest that the JMDpm-xARVCF complex interferes with adhesion by a mechanism that does not disrupt the cadherin to β-catenin or the α-catenin to β-catenin bond.

Constitutively Active Rac Restores Adhesion of Cells of the Inner Layer

Cadherin-mediated adhesion is influenced by the activity of small GTPases (Braga,2002). In Xenopus laevis, the importance of this activity is underlined by the observation that expression of a dominant negative version of Rac disrupts adhesion of cells in the animal cap (Hens et al.,2002). Since p120 and ARVCF are known regulators of small GTPases, we speculated that this regulatory link might contribute to the loss of the adhesion phenotype.

In a first step, we confirmed the ability of dominant negative Rac (RacN17) to interfere with adhesion. Expression of RacN17 in the animal cap led to a loss of adhesion in the inner cell layer, while the outer layer remained intact. The phenotype appeared similar to the phenotype obtained with JMDpm+xARVCF (Fig. 7B,C,) Table 2). Expression of a constitutively active Rac (RacV12) seemed to have the opposite effect, resulting in a very thick animal cap due to an accumulation of cells in the inner layer (Fig. 7D). Coinjection of RacV12 with JMDpm+xARVCF caused cell accumulations similar to RacV12 alone (Fig. 7E). These accumulations mainly consisted of JMDpm-positive cells (Fig. 7E′). Thus, RacV12 indeed restores adhesion of the manipulated cells. Efforts to titrate the amount of injected RacV12 to a level that would not result in a visible phenotype when expressed alone, but would still rescue adhesion, were unsuccessful (data not shown). It is possible that the critical concentration of RacV12 needed to restore adhesion produces additional effects, perhaps caused by the unregulated distribution of the overexpressed protein.

thumbnail image

Figure 7. Coexpression of racV12 restores adhesion in JMDpm+xARVCF-expressing cells. Animal caps of injected embryos at early gastrula stage (stage 10.5). A–E: C-cadherin staining. B',E': Overlay of C-cadherin (green) and JMDpm (red) staining. During fixation, the animal cap has collapsed on the endodermal tissue below (ed). Opposing arrows indicate thickness of animal cap. ol, outer layer; il, inner layer. Bar = 100 μm. A: Uninjected bilayered animal cap. B: JMDpm+xARVCF (1 ng resp. 250 pg/inj.), cells of the inner layer have detached, outer layer is intact. B': JMDpm-positive cells in the outer layer. C: racN17 (200 pg/inj.), cells of the inner layer have detached, outer layer is intact. D: racV12 (25 pg/inj.), cells of the inner layer form a thick cluster. E: Coexpression of racV12, JMDpm, and xARVCF (25 pg, 1 ng resp. 250 pg/inj.). The inner layer forms a thick cluster of adherent cells. E': Thick part of the inner layer consists of JMDpm-positive cells. F: Quantification of intracellular cadherin spots. The number of cadherin-positive spots in the cytoplasm was determined per myc-positive cell. Cells of the outer layer (ol) and cells of the inner layer (il) were evaluated separately. If construct expression resulted in an increase of spots, spots in adjacent myc-negative cells were also counted as control. Error-bars indicate standard deviation.

Download figure to PowerPoint

Table 2. Summary of rac-Induced Effects in the Animal Capa
ConstructCell arrangement in the animal cap, % of embryos
Wild-typeLooseInner cells detachedAdherent clusters
  • a

    mRNA of the various constructs was injected at 2–4 cell stage and the embryos were fixed, sectioned, and stained at early gastrula stage (stage 10.5). mRNA concentrations: Constitutively active rac (racV12) 25 pg/inj.; dominant negative rac (racN17) 200 pg/inj.; JMDpm 1 ng/inj.; xARVCF 250 pg/inj. For each condition, a total of 12 embryos was evaluated in 3 independent experiments.

Uninjected100000
JMDpm+xARVCF012.587.50
racV12000100
racV12+JMDpm+xARVCF000100
racN17001000

JMDpm expression leads to a redistribution of cadherin staining from the plasma membrane to cytoplasmic spots. While coexpression of RacV12 with JMDpm+xARVCF restored adhesion, cadherin spots were still clearly visible and occurred at similar amounts as in JMDpm+xARVCF-expressing embryos (Fig. 7B′,E′,F). In addition, while RacN17 disrupted adhesion of cells in the inner layer, cadherin staining at the plasma membrane remained strong and there was no increase in the number of cadherin spots detectable (Fig. 7C′ and F). Rac, therefore, does not seem to influence the intracellular distribution of cadherin, suggesting that its activity regulates other components of the adhesion machinery.

These results show that constitutively active Rac can counteract the loss of adhesion in JMDpm+xARVCF-expressing cells. The observation that adhesion is restored, although cadherin staining at the plasma membrane remains reduced and intracellular cadherin spots persist, suggests that the phenotype obtained with JMDpm+xARVCF cannot solely be explained by the cadherin redistribution to the cytoplasm.

Loss of Adhesion Upon Plasma Membrane Recruitment of xARVCF Through Binding to a Myristylated JMD or Through Direct Membrane Targeting of a Myristylatable xARVCF Variant

While the strong synergistic inhibition of adhesion by JMDpm and xARVCF suggests that effect is due to xAVRCF recruitment to the membrane, more indirect mechanisms are conceivable that could involve, for instance, two parallel pathways, one triggered by xARVCF overexpression and another by JMDpm, which may act through other yet unidentified JMD-interacting molecules or even through the transmembrane domain. To address this latter possibility, we examined the properties of JMDmyr, a variant where the transmembrane domain and the most proximal amino acids of JMDpm were deleted and replaced by a short myristylation signal. JMDmyr was efficiently targeted to the plasma membrane (Fig. 8B), and showed a remarkable capacity to re-localize xARVCF to the plasma membrane, as 100 pg mRNA were sufficient (Fig. 8F' and J), while as much as 1 ng of JMDpm mRNA was required to achieve a comparable re-distribution (Fig. 4C and data not shown). This high efficiency is probably due at least partly to the strong surface localization of JMDmyr itself (compare the distribution of JMDmyr and JMDpm, expressed alone in Figs. 2B and 8B, and co-expressed with xARVCF in Figs. 3B and 8F). Note that for both constructs, xARVCF co-expression increased JMD cell surface localization.

thumbnail image

Figure 8. Plasma membrane localization of xARVCF is necessary and sufficient to disrupt adhesion. Animal caps of stage-10.5 embryos. Ol, outer cells; il, inner cells; ed, endoderm. Bar = 50 μm. Sections were stained as follows. Green: C-cadherin; blue: nuclei; red: myc-tagged JMD constructs (B–E, 100 pg mRNA/inj.) or HA-tagged xARVCF (F–I, 250 pg/inj.). Red (D') and green (F'–I') channels were omitted to show cadherin and xARVCF distribution, respectively. A–E: Effect of JMDmyr on animal cap integrity: expression of JMDmyr (B) or JMDmyrG-A (C) alone has no detectable effect. Co-expression of JMDmyr+xARVCF causes inner cells to loose adhesion (D), but cadherin staining in the remaining outer layer is not affected (D', compare to control, A). Cells co-expressing JMDmyrG-A+xARVCF remain adherent (E, H). F–H: Effect of JMDmyr on xARVCF distribution. xARVCF expressed alone is mainly cytoplasmic (F, F'), but is titrated to the plasma membrane by JMDmyr (G, G' arrow). xARVCF co-expressed with JMDmyrG-A (H) shows a distribution similar to xARVCF alone. I, I': Inhibition of adhesion by direct membrane targeting of lipid-anchored xARVCFmyr. J: Coimmunoprecipitation of xARVCF with JMDmyr but not JMDmyrG-A. Immunoprecipitations were performed as in Figure 4. Loaded lysate corresponds to the equivalent of 1/8 embryo, loaded myc-IP to the equivalent of 4 embryos.

Download figure to PowerPoint

Injection of 100 pg JMDmyr mRNA alone showed no effect on adhesion or on tissue organization in the animal cap (Fig. 8B and data not shown, Table 1). Higher amounts could not be used as they caused early defects in cell division (data not shown). However, co-expression of JMDmyr with xARVCF led to complete loss of cell–cell adhesion in the inner layer of the ectoderm (Fig. 8D,F), similar to the effect of JMDpm. This phenotype was not accompanied by cadherin redistribution. We did not observe any significant reduction of cadherin staining at the plasma membrane in JMDmyr or JMDmyr+xARVCF expressing cells (Fig. 8D' and data not shown) nor any increase in intracellular cadherin spots (JMDmyr as well as JMDmyr+xARVCF: 0.2 ± 0.1 cadherin spots/cell).

To confirm that binding of xARVCF to the membrane-bound JMD is required, we compared the effect of JMDmyr with a mutated form (JMDmyrG-A). This construct contains a triple glycine to aspartate mutation in the core binding region, which has been shown to abolish the binding of ARVCF (Paulson et al.,2000). Expression of JMDmyrG-A alone had no detectable phenotype (Fig. 8C). After co-expression of xARVCF with JMDmyrG-A, the inner cells also remained adherent (Fig. 8E). There was no detectable interaction of xARVCF with JMDmyrG-A in immunoprecipitation experiments (Fig. 8J) and xARVCF remained mostly cytoplasmic in the presence of JMDmyrG-A (Fig. 8H), with only a very low membrane signal, comparable to the signal in cells expressing xARVCF alone (Fig. 8G).

We finally tested directly whether the observed disruption of adhesion required the xARVCF–JMD interaction, or was solely the result of physical recruitment of xARVCF to the plasma membrane. We thus expressed a variant full-length xARVCF fused to a myristylation sequence (xARVCFmyr). This construct, which concentrated as expected at the plasma membrane (Fig. 8I'), caused strong disruption of cell adhesion in the inner layer of the animal cap (Fig. 8I). Unlike JMDpm + xARVCF, but similar to JMDmyr + xARVCF, xARVCFmyr did not affect cadherin staining (0.1 ± 0.1 cadherin spots/cell). Thus, the inhibition of adhesion can be recapitulated by recruiting xARVCF at the plasma membrane, independent of interaction with the JMD, and unrelated to changes in cadherin distribution.

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

Proteins of the p120 catenin-family have been implicated in multiple compartmentalized signaling events, including regulation of transcriptional activity in the nucleus, regulation of Rho GTPases in the cytoplasm, and modulation of cadherin-mediated adhesion at the plasma membrane. Several potential functions of these proteins in modulating adhesion have been suggested, but their activity has remained rather cryptic. The most firmly established function to date is in the control of cadherin stability. In mammalian cell lines, p120 appears to act as a “rheostat,” in the sense that the binding-ratio of these proteins to the JMD determines the amount of cadherin residing at the cell surface (Ireton et al.,2002; Davis et al.,2003; Xiao et al.,2003). In Xenopus laevis embryos, a reduction in Xp120 or xARVCF levels leads to a decrease in cadherin amounts, indicating a similar function (Fang et al.,2004; Tao et al.,2007). Since p120catenins appear essential to maintain cadherin levels, other potential functions cannot be studied through knock-down approaches.

We use here xARVCF overexpression and manipulation of its intracellular location to assay for activities of this protein and we show that recruitment of xARVCF (as well as Xp120, Table1 and data not shown) to the plasma membrane efficiently interferes with cell–cell adhesion. We further demonstrate that this effect can be fully accounted for by accumulation of xARVCF at the cytoplasmic surface of the plasma membrane, independently of its binding to the JMD.

The observed inhibition of adhesion does not appear to be due to changes in cadherin surface expression, nor to an increased turnover, which would be expected to be reflected by increased intracellular cadherins. We also found no detectable difference in the amount of catenins interacting with cadherins. Considering the severity of the loss of adhesion, if any of the above-mentioned mechanisms would be involved, one would have presumably detected some differences.

Alternative mechanisms could be: (1) titration of other yet unidentified components of the cadherin-based adhesion complex by mislocalized xARVCF, or (2) indirect inhibition of adhesion through regulation of the cortical actin cytoskeleton. We consider the first possibility less likely, because we have shown that xARVCF must be recruited to the plasma membrane for its activity. One would have then to assume that the potential xARVCF-sequestered adhesive component is exclusively accessible to membrane-bound xARVCF. Our results, on the other hand, fit well with the latter model: Efficient adhesion depends on a still not completely understood interaction between the cadherin adhesion complex and the underlying actin cortex (Mege et al,2006). Localization of p120catenins to the membrane independently of cadherins would ectopically modify the cortical cytoskeleton in a non-productive manner that would be eventually detrimental for adhesion. We thus consider that this artificial experiment reveals an intrinsic property of the p120catenins, which is normally contained within the adhesion complex and difficult to discern. Our data complement the recently reported observation that, in these same animal cap cells, the presence of the JMD in C-cadherin is required for the maintenance of a dense actin cortex, consistent with the JMD and associated xARVCF/ Xp120 modulating structural aspects of the actin cytoskeleton (Tao et al.,2007). The prime candidates to mediate actin regulation by p120catenins are the small RhoGTPases (Noren et al.,2000; Grosheva et al.,2001; Yanagisawa and Anastasiadis,2006), and one of the proposed functions of p120catenins is indeed recruitment of these GTPases to the plasma membrane (Anastasiadis and Reynolds,2001). We tested dominant negative and constitutively active versions of the three prototypes Rho, Rac, and cdc42, for their ability to rescue the loss of adhesion phenotype (data not shown). We found Rac to be the most likely candidate target of xARVCF for this phenotype. Indeed, dominant negative Rac was the only component tested that caused cells of the animal cap to detach, similar to membrane-bound xARVCF (Fig. 7 and data not shown). Furthermore, constitutively active Rac, while showing additional effects on cell behavior in the animal cap, clearly restored completely adhesion of cells expressing JMDpm+xARVCF. Whether Rac is indeed the target, and the only target, of xARVCF responsible for the observed phenotypes cannot be demonstrated by our experiments, and other p120catenin interactors (Kausalya et al.,2004; Sallee et al.,2006; Wildenberg et al.,2006; Boguslavsky et al.,2007) might be involved. Understanding the molecular mechanisms that lead from recruitment of xARVCF at the membrane outside of the cadherin complex to de-adhesion and that allowed constitutively active Rac to restore adhesion are equally beyond the scope of this report.

Our data, however, emphasize the extent to which adhesion appears to be modulatable independently of surface levels of cadherin and of changes in the classical cadherin-β-catenin-α-catenin interaction. In our experiments, complete de-adhesion could be obtained without any significant change in the levels of cadherin-associated catenins (Fig. 6). Although cadherin levels at the surface were decreased under certain conditions with concomitant increase in intracellular cadherin signal, such changes did not necessarily correlate with the strength of adhesion. JMDmyr, for instance, efficiently synergized with xARCVF to block adhesion without a parallel effect on cadherin distribution. Most strikingly, constitutively active Rac restored very compact tissues without rescuing the significant decrease in cadherin induced by JMDpm + xARVCF. These observations, taken together with the ability of xARVCF to disrupt adhesion merely by its localization to the plasma membrane, are all consistent with a major role of the cortical actin organization on the strength of adhesion (Drees et al,2005, Yamada et al,2005), and for a potential role of xARVCF in linking cadherins to this regulation (Yanagisawa and Anastasiadis,2006).

Cadherin deletion constructs have been widely used to disrupt or modulate adhesion. Using different types of constructs allowed us to distinguish the effects of different domains on cell adhesion and endogenous cadherin distribution. Firstly, we noted that the expression of fusion constructs containing the signaling sequence and the transmembrane domain (JMDpm and CadΔE, Fig. 2) resulted in a redistribution of C-cadherin from the plasma membrane to distinct spots in the cytoplasm, while JMDmyr (Fig. 8), which localizes to the plasma membrane via a cytoplasmic pathway, did not. These observations suggest that this phenotype must be mechanistically distinct from the cadherin downregulation reported upon depletion of endogenous p120catenins. It might be provoked by the passage of the constructs through the exocytotic pathway, and/or stimulation of endocytosis by lateral interaction of the transmembrane domain of the constructs with other membrane components (including perhaps endogenous cadherins). In any case, as discussed above, the strong de-adhesion phenotype appears to be independent of this effect on cadherin distribution. Secondly, our data give a different perspective on the role of the β-catenin binding domain of these constructs. Although we confirmed that this domain participates in this inhibitiory activity, since the CadΔE construct interfered with adhesion more efficiently than JMDpm (Fig. 2), our data are not consistent with the model initially proposed, where CadΔE would interfere with adhesion by titrating β-catenin away from endogenous cadherin (Kintner,1992). In our hands, CadΔE-injected embryos showed a substantial amount of α- and β-catenin associated with cadherins and the total amount of β-catenin was even increased compared to controls. It thus seems that CadΔE does not compete for cadherin-bound β-catenin, but rather recruits and stabilizes more β-catenin from the soluble pool. A similar conclusion has been drawn from experiments in a carcinoma A431 cell line (Nieman et al.,1999). CadΔE may, therefore, interfere with adhesion by sequestering other factors interacting directly or indirectly with the β-catenin binding domain. Like p120 and ARVCF, α- and β-catenin have been shown indeed to bind several other cytoskeletal regulators at the plasma membrane (Gumbiner,2000; Woodfield et al.,2001; Kobielak et al.,2004; Sallee et al.,2006). The approach we present here will provide a robust assay to further investigate a synergistic interaction between different domains of the cadherin cytoplasmic tail.

EXPERIMENTAL PROCEDURES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

Embryos and Microinjections

Eggs were obtained from Xenopus laevis females injected with 400 U human chorionic gonadotropin (Sigma) and in vitro fertilized with macerated testis. The jelly coat was removed with a 2% cysteine-solution (pH 8). Embryos were kept in 0.1× MBS-H (88 mM NaCl, 1 mM KCl, 2.4 mM NaHCO3, 0.82 mM MgSO4, 0.41 mM CaCl2, 0.33 mM Ca(NO3)2, 10 mM HEPES [pH 7.4], 10 mg/ml streptomycin sulfate, and 10 mg/ml penicillin). Capped mRNA for injections was synthesized from NotI linearized plasmids using the SP6 messageMachine Kit (Ambion) according to the manufacturer's protocol. Animal injections were performed either at the 2-cell stage into the animal pole of both blastomeres or at the 4-cell stage into the animal pole of the two ventral blastomeres. Dorsal injections were performed at the 4-cell stage into the equatorial region of the two dorsal blastomeres. Injection volume was 5 nl.

Constructs

pCSXp120 (Kim et al.,2002); pCSCadΔE-myc (Kurth et al.,1999); pCSxARVCF1B-HA pCSJMDcyt-myc, pCSJMDmyr-myc, pCSJMDG-Amyr-myc (Paulson et al.,2000); pCSJMDpm-myc (Reintsch et al.,2005), pCSracV12 and pCSracN17 (Tahinci and Symes,2003). pCSJMDmi-myc: To generate a mitochondria-bound JMD, the Xenopus database was searched with the protein-sequence of rat TOM20 (Kanaji et al.,2000). Primers were generated from the closest matching sequence (gi: 27371034) to PCR-amplify a fragment encoding the minimal mitochondria localization signal (amino acid 1 to 35) from a stage-10.5 embryo RT. The fragment was cloned N-terminally into pCSJMDcyt-myc to obtain pCSJMDmi-myc. To generate pCSxARVCFmyr, the JM domain of pCSJMDmyr-myc was eliminated by ClaI digest and relegation. xARVCF1B was PCR-amplified and cloned C-terminal into pCSmyr-myc.

Histology and Immunofluorescense

Histological samples were prepared using a previously described protocol (Fagotto et al.,1999). Briefly, embryos were first fixed in 3.7% formaldehyde/PBS and then in DMSO/methanol. After embedding in fish-gelatin, 10-μm serial cryosections were prepared. Antibodies were diluted in 5% milk/PBS. Secondary antibodies were goat-anti mouse, respectively, goat-anti rabbit antibodies coupled to Alexa488 or Alexa546 (Molecular Probes). Nuclei were counterstained with DAPI (0.5 μg/ml) and yolk with 0.2% Eriochrome.

Primary antibodies: mouse anti-myc 9E10 (Evan et al.,1985), rabbit anti-myc (gift from Thomas Joos), mouse anti-HA 12CA5 (Field et al.,1988), rabbit anti-C-cadherin (Yap et al.,1997). With the exception of the images in Figure 5, pictures were obtained using a Zeiss TV135 inverted microscope with a 25× (0.8 N.A.) water-immersion objective (Carl Zeiss MicroImaging, Inc.) and a CCD-Video camera (DXC-950P, Sony). Images were acquired and processed with AnalySIS-software (Soft Imaging System GmbH, Muenster). Large high-resolution images were obtained by collating pictures of adjacent regions (Schohl and Fagotto,2002). Pictures in Figure 5 were obtained using a Zeiss LSM510 Confocal microscope with a 25× (0.8 N.A.) water-immersion objective. Pictures were acquired with an open pinhole, which resulted in a comparable image quality as with the Zeiss TV135.

Analysis of Cadherin Spots

Cryosections were stained for C-cadherin and the tagged constructs. Starting from a randomly chosen construct-expressing cell in the inner or outer layer of the animal cap, intracellular cadherin spots were counted in neighboring construct-expressing cells. If the number of cadherin spots was increased compared to uninjected controls, cadherin spots in neighboring wild-type cells were also counted as additional control. For each condition (inner or outer layer and construct expressing or wild-type cells), 10 cells/section were evaluated. For each construct, sections from at least 8 embryos, obtained in 3 independent experiments, were used. Results were tested for significant differences with the Student's t-test.

Immunoprecipitation

Injected embryos were extracted at stage 10.5 in 0.5% NP40, 120 mM NaCl, 1 mM MgCl2, 1 mM CaCl2, 50 mM Tris-HCl, pH 7.4, supplemented with Proteinase-Inhibitors. Cadherin-IP: Precipitation was done with monoclonal anti-C-cadherin antibody 5G5 (gift from B.M. Gumbiner), bound to ProteinG-beads (Sigma). Blots were probed with antibodies against β-catenin P14L (Schneider et al.,1993), α-catenin (rabbit polyclonal, Sigma) and C-cadherin (5G5). Myc-tag-IP: Myc-tagged JMD constructs were precipitated with the mouse anti-myc antibody 9E10. Blots were probed with anti-HA antibody to visualize xARVCF-HA and with anti-myc antibody.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

We thank Rudi Winklbauer for constructive comments, Carolyn Brown for thorough and critical reading of the manuscript, and Drs. Horb and Ruthazer labs for technical support. This work was supported by CIHR grant 62898 and NCIC grant 017162 to F.F. and CIHR grant 68970 to C.A.M.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES